Modes of Existence: Papers in Ontology and Philosophical Logic 9783110327533, 9783110327281

Citation preview

Andrea Bottani / Richard Davies (Eds.) Modes of Existence

Philosophische Forschung Philosophical Research Herausgegeben von / Edited by Johannes Brandl • Andreas Kemmerling Wolfgang Künne • Mark Textor Band 5 / Volume 5

Andrea Bottani / Richard Davies (Eds.)

Modes of Existence Papers in Ontology and Philosophical Logic

ontos verlag Frankfurt I Paris I Ebikon I Lancaster I New Brunswick

Bibliographic information published by Die Deutsche Bibliothek Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliographie; detailed bibliographic data is available in the Internet at http://dnb.ddb.de

North and South America by Transaction Books Rutgers University Piscataway, NJ 08854-8042 [email protected]

United Kingdom, Ire Iceland, Turkey, Malta, Portugal by Gazelle Books Services Limited White Cross Mills Hightown LANCASTER, LA1 4XS [email protected]

Livraison pour la France et la Belgique: Librairie Philosophique J.Vrin 6, place de la Sorbonne ; F-75005 PARIS Tel. +33 (0)1 43 54 03 47 ; Fax +33 (0)1 43 54 48 18 www.vrin.fr



2006 ontos verlag P.O. Box 15 41, D-63133 Heusenstamm www.ontosverlag.com ISBN 10: 3-938793-12-1 ISBN 13: 978-3-9387932006 No part of this book may be reproduced, stored in retrieval systems or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use of the purchaser of the work

Printed on acid-free paper ISO-Norm 970-6 FSC-certified (Forest Stewardship Council) This hardcover binding meets the International Library standard Printed in Germany by buch bücher dd ag

Contents Introduction Andrea Bottani and Richard Davies (University of Bergamo, Italy)

7

Facts, Formal Objects and Ontology Kevin Mulligan (University of Geneva, Switzerland)

31

Fictional and Aesthetic Objects: Meinong’s Point of View Venanzio Raspa (University of Urbino, Italy)

47

Russell’s Descriptions and Meinong’s Assumptions Frederick Kroon (University of Auckland, New Zealand)

81

McGinn on Existence Peter van Inwagen (University of Notre Dame, USA)

105

The Talk I Was Supposed To Give Achille C. Varzi (University of Columbia, USA)

131

Two Interpretations of “According to a story” Maria E. Reicher (University of Graz, Austria)

153

Madame Bovary as a Higher-Order Object Carola Barbero (University of Turin, Italy)

173

Identity across Time and Stories Francesco Orilia (University of Macerata, Italy)

191

A Problem about Reference in Fiction Giuseppe Spolaore (University of Verona, Italy)

221

Introduction Richard Davies and Andrea Bottani The aim of this introduction is to comment briefly on the background to the three primary topics touched on from various directions by the contributors to this collection. The background is the very general field of what is meant by the verb ‘to exist’ and, hence, of what we have called in our title ‘modes of existence’. At the risk of some artificiality, the subdivisions of this field considered by our authors might be summarised as: first, the general principles of how ontological primitives are to be established; second, the question of the status of negative existentials; and, third, the treatment of fictional discourse. These questions are closely interrelated in ways that might be made to emerge by recasting them, at an equal risk of facetiousness, as the ‘is’ question; the ‘is not’ question; and the ‘as if’ question. Put that way, shorn of abstract terminology, it should be plain why most of the essays here presented touch on more than one of the primary topics. Any answer to the ‘is’ question will have consequences for the ‘is not’ question. Indeed, one of the most influential texts on these issues, Quine’s ‘On What There Is’, announces on its very first page ‘the ontological problem’, which is understood as the ‘riddle of nonbeing’ (Quine 1948). Likewise, any answer to the ‘is not’ question will have consequences for the ‘as if’ question. Again, a – or we might say, advisedly, ‘the’ – classic in the field, Russell’s groundbreaking article, ‘On Denoting’ (Russell 1905), witnesses a very direct relation between a characteristic handling of sentences whose grammatical subject has no referent and the view that myth-making, story-telling and the like do not produce truths, at least not in the way that, say, zoology produces truths. Let us proceed to consider what is at stake in our three questions, considered separately and jointly. I. ‘Is’ The approaches adopted by all the authors here represented to the ‘is’ question fall within the taxonomy presented in the useful overview

8

Introduction

offered by Peter van Inwagen in introducing his discussion of Colin McGinn’s recent book Logical Properties. Although van Inwagen individuates a total of six distinguishable positions, we may say that they fall into two principal classes. On the one hand, there are theories that regard being and existence as identical; for instance, Kant, Frege and Russell regard assertions of being/existence as ascriptions of a relational property to an abstract object, such as a property, a concept or a propositional function. Thus, Kant famously observes that, ‘“Being” is obviously not a real predicate; that is, it is not a concept of something which could be added to the concept of a thing’1, by which he seems to have meant that the satisfaction of the concept is quite a different matter from its definition. Likewise, for Frege, existence cannot but be a predicate of predicates2; and what Russell calls the ‘feeling for reality’ should ensure that logic contemplates ‘only one world, the “real” world’3. Also among those whom van Inwagen classifies as not seeing any call to make any distinction between ‘there is’ and ‘there exists’ is Quine, for whom the existential quantifier (plus identity) expresses both indifferently. As we shall see in considering the ‘is not’ question, van Inwagen places himself squarely in the Quinean tradition of those who say that ‘there is an x’ or ‘there exists an x’ both mean exactly that, for some x and for some y, x is identical to y. On the other hand, van Inwagen sets out four theories that make some distinction between being and existence. The theories that distinguish between being and existence clearly do not rely, as the Quineans do, on the alleged basicness of unrestricted quantification, but range instead over a variety of domains of quantification that are not co-terminous with concrete actuality. In Meinong’s view, for instance, we ought to distinguish between existence and subsistence; in addition to the things that have existence (concrete, real objects) or subsistence 1 Kant 1781, p. 504 (A598, B626), emphasis original. 2 Frege 1884, §53. 3 Russell 1919, p. 169.

Richard Davies and Andrea Bottani

9

(abstract, ideal objects)4, Meinong’s characteristic claim is that some things have nonbeing or are outside being (Aussersein)5. More recent champions of the Meinongian tendency have inclined, rather, to distinguish the concepts of being and existence, and to attribute the former to everything whatever, while allowing that there are things that do not exist. It may be that the verbal distinction here does not amount to a real philosophical novelty but rather a terminological preference for using ‘exists’ in place of ‘exists or subsists’ and ‘has being’ for ‘there is’. Nevertheless, the thesis that there are things that do not exist is one that the neo-Meinongians share not only with those who envisage an ontology of possibilia, which are things that possibly exist, but do not actually do so; but also with McGinn, whose theory of existence is set out in some detail and is subjected to close criticism regarding its consequences for what we are calling the ‘is not’ question. The dichotomy that van Inwagen helpfully proposes as basic to posing and resolving the ‘is’ question clearly excludes, as van Inwagen candidly acknowledges, a great many of the ways that that question has been approached in the tradition. While it may well be that philosophers of the sort that van Inwagen calls ‘our sort’ are right to think that clarification of how to pose and resolve the ‘is’ question is a primary philosophical objective, a doubt might be raised about just how to identify ‘our sort’ of philosopher. While, again, van Inwagen is surely right to say that the ontological musings of, for example, Heidegger and Sartre are not to be taken to be clarifications – even in intention – of the ‘is’ question, he also names and sets aside Aristotle and Ryle as not ‘of our sort’. In so doing, he makes it clear that, for philosophers of ‘our sort’, the debate about the ‘is’ question and its cognates is not a question of ‘analytic’ against ‘Continental’. For, though only local legend might have it that Aristotle was an early Student of Christ Church, Oxford, there is no doubt that 4 For this account of Meinong’s considered position, see Grossmann 2005. In

some moments, it seems that being is the genus of which existence and subsistence are mutually exclusive species; in others, that subsistence is possessed also by the things that exist and is the timeless aspect of their being. We are grateful to Venanzio Raspa for exegetic clarification on a point about which, as Lambert observes, Meinong expresses himself ‘confusingly’ (Lambert 1983, p.4). 5 Cf. Meinong 1904, p. 82.

10

Introduction

Gilbert Ryle counts as an ‘analytic’. Rather, ‘our sort’ embraces all those who recognise that a crucial stand must be taken over how quantification is to be construed, and thus it cannot but include Frege, Russell and Quine. On the other hand, though Meinong, the neoMeinongians, the possibilists and McGinn do not form a homogeneous team, they do count as of van Inwagen’s ‘sort’ because their claims can be translated without obvious residue into the canonical notation of first-order predicate logic with identity. And, once translated, those claims can be tested for their consequences. While Heidegger would not have admitted the tellingness of the test (cf. Carnap 1932), most modern Meinongians see that the machinery of quantification theory permits their characteristic claims about the ‘is’ question to be expressed perspicuously. What, then, of Aristotle and Ryle? Relative to the ‘is’ question, what sets them, and with them many others, apart from van Inwagen’s sort is, as he notes, that, in their different ways, they deny the univocity of being. They deny, that is, that the ‘is’ question with regard to the category of the subjects of predication (substances) exhausts what there is to be said about being6. If they were right about this, it would pretty clearly imply the inapplicability of first-order predicate logic with identity to large tracts of the ‘is’ question. For that canonical notation presupposes that the only candidates for being are substitutioninstances of the variables envisaged by the notation. The substitutioninstances of the variables of first-order logic are individuals; and the category-theoretic approaches of Aristotle, Ryle and many others will allow not merely that being is said in many ways7, but that it is said of instances of categories other than that of individual substance. Of course, many philosophers have been persuaded that instances of other categories can be treated as if they were substitution-instances of the variables of first-order predicate logic, and have quantified gaily over events, times, places and whatnot (though rarely, for example, over postures8) as if they were individuals in just the way that substances are individuals. That is, because these other kinds of particulars 6 See, e.g., Aristotle, Categories, v. 7 See, e.g. Aristotle, Metaphysics, Γ, ii, 1003a33. 8 Cf. Aristotle, Categories, v, 1b27.

Richard Davies and Andrea Bottani

11

can be subjects of predication, the canonical notation treats them on a par with what the category-theorists will think of as the focal bearer of ‘is’. About this, it is not our present intention to judge or even to opine, beyond the observation that there may be more to the ‘is’ question than can be captured by translation into the canonical notation of the Quineans. Nevertheless, even those who doubt or deny the univocity of being may learn a great deal about the ‘is’ question by attending to translatability into first-order predicate logic with identity; but they may continue to doubt or deny that such translatability fixes a limit to what is to be done in ontology. Relative to the dichotomy we have borrowed from van Inwagen, the first two contributions to this volume stand in fairly stark opposition. The distinction between theories that do not and theories that do distinguish between being and existence may be applied to the two thinkers who stand more or less openly behind Kevin Mulligan and Venanzio Raspa: respectively, the early Wittgenstein and Alexius Meinong. In Quine’s image, the ontology of the Tractatus is undoubtedly a ‘desert landscape’9, insofar as Wittgenstein announces that ‘the world is the totality of facts’10; being a fact is thus the only way for something to be part of the world or to exist. The question, then, that Mulligan raises formulates our ‘is’ question as the question, ‘what is a fact?’ One of the most prominent theories of what it is to be a fact, elaborated under the pretty direct influence of the Wittgenstein of the Tractatus, is that associated with F.P. Ramsey, according to which a fact is a true proposition (Ramsey 1920). If this is so, then, the world is the totality of true propositions. Against a suggestion of this sort many objections might be raised, among which we may mention two, which appear separately damaging and jointly fatal to it. A first objection is that, if the world is made of true propositions, then it becomes entirely mysterious what it is in virtue of which they are true. If, that is, ‘what is the case’ is itself propositional, the distinction between truth and falsehood seems either to be downplayed or to run counter to the intuitive thought that truth is a product of or supervenes upon reality. 9 Quine 1948, p. 4. 10 Wittgenstein 1921, 1.1

12

Introduction

Furthermore, truth is treated as identical with being, and what vanishes is the idea that truth and falsity are linguistic notions. A second objection is that, while there is nothing particularly mysterious about a proposition’s being false of something else (of the that-which-it-isabout), the notion of a fact’s being ‘negative’ puts the cart before the horse, because it seeks to build the notion of truth, rather than that of, say, existence, into the basic ontological notions without explaining what is true of what. If a (positive) fact is a true proposition, then a negative fact would have to be a true proposition that is false, which certainly looks incoherent. Hence, it is very implausible to propose a theory of facts on which they are both true propositions and ontologically fundamental. An alternative approach has been to say that a fact should be construed as a sui generis entity that has some internal structure, but that is nevertheless primitive. For instance, one might say that a fact is a combination of n particulars or properties with an n-placed predicate. The notion of combination here is to be taken as in some way predicative or exemplificative: the particulars exemplify the predicate. Were the question answered of how this combination is supposed to be wrought, Mulligan proposes that, even on this conception of what facts are, they are not ontologically fundamental because exemplification is not the most basic sort of ontological tie. It is not the most basic because it can itself be construed as a case of the more general notion of ‘obtaining’ or ‘being thus-and-so’. In this respect, if there are facts, then they are to be understood as obtaining states of affairs rather than as any sort of entity that has a predicative structure. More positively, Mulligan argues that things (substances), states, processes, perhaps space-time and, for the non-nominalist, kinds of things, kinds of states and kinds of processes, should be counted as fundamental for the very simple reason that it is because of these that any facts that one might wish to cite are true. The fundamentalness of things (etc.) is an ontological fundamentalness relative to facts, properties and relations in two leading respects. One is that none of the former is identical with any of the latter. For instance, a person endures in time and has boundaries in space whereas a fact about, a property of, or a relation to a person has neither of these features and, hence, cannot be identical with the person. The other is that things (etc.) cannot

Richard Davies and Andrea Bottani

13

be constructed out of facts (etc.). What this means is that, while there might be something that makes a certain person sad, and that person’s being sad makes true the fact that the person is sad, that fact is not among the things that makes the person sad. Rather, the converse holds: the sad person is the truth-maker of the derivative fact. While there may remain some uncertainty about how to understand what it is for a fact to be made true by a sad person, it is pretty certain that it will have to call on the notion of an object. As already indicated, this notion is itself by no means neutral in those theories that propose some distinction between a wide sense of ‘being’ and narrower senses of ‘existence’ and ‘subsistence’. While the initial moral drawn from Russell’s theory of descriptions was that the beings over which ‘object’ should range are restricted to those recognised in the “real” world, anti-Russellians of various stripes have continued to subdivide objects according to the species of being that they instantiate. In discussing Meinong’s account of the status of fictional and æsthetic objects, Venanzio Raspa assumes the fundamental theses of his theory of objects. It is worth noting that the theory is in large measure motivated by Meinong’s desire to make sense of Brentano’s claim that every mental state must have an object as a component. Thus, when one thinks about, imagines, or deplores something, there has to be a something about which one is thinking, that one is imagining or deploring; this something is the ‘intentional object’ of the state in question. Consider, then, the apparently inconsistent triad: (i) all intentional states are relations to their objects; (ii) an n-placed relation requires that n relata exist; and (iii) some intentional states have as their objects things that do not exist11. Where Husserl moved in the direction of denying (i) by saying that intentional states are other than ‘quasi-relations’ and Russell would have refused (ii) in the case of non-referring or ‘merely intentional’ states, Meinong takes (iii) head on. The Meinongian answer to the ‘is’ question is thus formulated in terms of a taxonomy of the ways that various categories of objects are related or not related to being. The distinction between existent and non-existent objects is coordinated in the first instance with that between real (concrete) and ideal (abstract) objects, and motivated by 11 This formulation is owing to Tim Crane in an unpublished paper on intention-

ality in Husserl.

14

Introduction

what Meinong calls ‘the principle of the indifference of the pure object to being’12. As several commentators have pointed out, this principle is closely related to the Kantian argument, alluded to above, that the being of an object is ‘alien or extrinsic’ to what it is for an object to be the object it is13. This, in turn, derives from the notion that an object’s being what it is is determined by the properties attributable to it: being determined or being thus-and-so (Sosein) is therefore independent of and prior to being (Sein). There appears, however, to be an asymmetry in the treatments of this ‘indifference’. On the one hand, Kant’s objection to what he dubbed the Ontological Argument for the existence of a God is that the concept of a God cannot include, as one of its ‘real’ properties, the property of existence; on the other, Meinong allows that there are objects, such as the round square, the list of whose properties, which makes up its being thus-and-so, includes an absurdity, so as to ‘carry in itself the guarantee of its own non-being in every sense’14. Why, one might wonder, should not a certain listing be such that it does carry in itself the guarantee of its own being in some sense or other? If, that is, ‘indifference’ is a relation between being thus-and-so and existence, then the non-existence of the round square should have a parallel in the existence of something-or-other in virtue of the properties (other than existence) that constitute its being thus-and-so. If the something-orother in question is not God, then perhaps we might think of the empty set or whatever else makes for the necessary truths of arithmetic. This is not to suggest that properties, such as self-identity, which are necessary for something to exist, are also sufficient; but rather, it is a hanging query about whether or not some being thus-and-so is such as to guarantee – to provide sufficient conditions – that something should instantiate it. Nevertheless, Meinong does envisage an ‘infinite totality of objects which are outside being’15, from among which we select when talking about fictions (a point to which we shall return in considering the ‘as if’ question). What makes the totality a totality is that it contains 12 Meinong 1904 p. 86. 13 Cf. Findlay 1933 p. 49; Lambert 1983, p. 20. 14 Meinong 1904, p. 86. 15 Meinong 1910, p. 197.

Richard Davies and Andrea Bottani

15

all the listings of properties that are attributable to objects, including absurd ones. Given that the listing of the properties is identical with what it is to be a determinate object (its being thus-and-so), it becomes a merely verbal matter whether we say it is the listing, such as one including both ‘round’ and ‘square’, or the object itself, such as a round square, that is absurd. Just as one might easily think that Ramsey, with his facts, was insufficiently alert to the merely possible, one might also worry that Meinong’s objects are overly hospitable to the impossible. The point can be made by reference to a recurring fiction whose incoherence has perhaps become a commonplace among philosophers without thereby affecting Hollywood box-office. Consider the many stories of backwards time-travel, in which the superposition of what actually happened and what is successively made to have happened is so successfully brushed over as not to impede the unfolding of the plot: the Terminator both was and was not in the police station at a time t. Of course, the story might be told without apparent absurdity, but it is not a matter of indifference that there is an absurdity lurking. For this reason, it might be well to say that no action taken by the Terminator after t can be such as to bring it about that he wasn’t if he was or that he was if he wasn’t, just as no town is such that every man in it is shaved either by himself or by the barber (but not both). The fact that such absurdities may be more or less covert or may take time and reflection to be unveiled does not make them any the less absurdities: whether they carry in themselves a guarantee of their own non-being implicitly or explicitly, the guarantee is in force. It might be noted that, these sorts of logical or conceptual absurdity are just one species of how the non-existence of something fitting a certain description might be guaranteed. There may be physical reasons for rejecting the existence of certain categories of things. For instance, we might wonder whether, conjointly, equine physiology (especially regarding muscle efficiency and the structure and tensile strength of the rib cage) along with the principles of aerodynamics are sufficient to carry a guarantee of the non-being of flying horses, such as Pegasus. In short, the ‘is’ question as it is faced in this volume is a question primarily about the nature of the objects of which it can be said that they are. Whereas Mulligan proposes that the concrete particular is

16

Introduction

prior and offers a fiercely parsimonious account of the ontological fundamentals, Raspa’s Meinong begins with listings of properties and thus enlarges the class of objects that are independent of what exists or even subsists.

2. ‘Is not’ Proceeding, then, from the ‘is’ question to the ‘is not’ question, it is helpful to bear in mind the distinction that Frederick Kroon draws between the Russellian ‘feeling for reality’ that allows us to use terms like ‘pseudo-object’ for the likes of round squares and, perhaps, flying horses, and the question of the correct logical analysis of propositions that are at least apparently about them. In particular, we need to be attentive to the question of how propositions denying that such things exist can be true. On a simplified version of Russell’s theory of descriptions, ‘the present King of France is bald’, is false because it is equivalent to the conjunction, ‘there is at least and at most one present King of France, and he is bald’, whose first conjunct (‘there is at least one present King of France’) is false. By analogy, then, we would expect to have to say that, ‘the present King of France does not exist’, is false because it is equivalent to the conjunction ‘there is at least and at most one present King of France, and he does not exist’, whose first conjunct (‘there is at least one present King of France’) is false. Yet, it is apparent that ‘the present King of France does not exist’ is true. The trouble here presumably arises from the way that ‘the present King of France does not exist’ appears to say of the present King of France that he does not exist; that is, for some x, x is a present King of France and x does not exist. What we have already heard Quine calling the ‘riddle of nonbeing’ is just that, if a sentence beginning with the quantifier, ‘for some x’, is to be true, there has to be an x that satisfies the predicates attributed to it; as Kroon notes, this line of thought risks landing Russell and others with the thesis that ‘the present King of France does not exist’ must be treated not merely as false, but as a contradiction because it appears both to assert and to deny of something that it is a present King of France. But there is no contradiction here; so something has gone wrong.

Richard Davies and Andrea Bottani

17

On the Quinean reading advanced by van Inwagen, the nonbeing predicate denies of everything that it is a present King of France: for all x, x is not a present King of France. So it may be worth pausing for a moment on the issue of how the quantifiers of the canonical notation of first-order predicate logic are to be read. For the purposes of criticising all those (‘of our sort’) who distinguish between being and existence, Van Inwagen himself makes do with the symbol for universal quantification, ‘∀’, read as ‘for all’, and merely alludes to another, which he describes as beginning with ‘e’ and rhyming with ‘residential’. One motivation for this austerity might be expressed by considering one way that introduction of the rhyming quantifier can give rise to puzzles. Consider what Whitehead and Russell say when introducing ‘∃’; they say, ‘it stands for “there exists”’16. So far forth, the symbols adopted by logicians are not responsible to anything but the other things that logicians use them for or say about them. In this respect, a reading of a bit of symbolism differs from the reading one might give of a word in a foreign language; for instance, a direction, of the sort that Giuseppe Spolaore cites, to read the French word ‘canard’ to stand for the same as the English word ‘water’ would pretty quickly lead to downright error because the French word is no longer up for stipulation of the sort that logicians are free to give to their symbols when they introduce them. Here, as Humpty Dumpty might say, the logician is master17. It is instructive to note how it is hard to find an English (or French or Humpty-Dumptian) formula that stands to ‘∀’ as ‘there is’ and ‘there exists’ do to ‘∃’. Of course, the formula ‘∀x Fx’ could be read ‘there is not (or does not exist) an x such that not-Fx’. That would do the trick. But it has not caught on and few would feel at home with it as a reading of the universal quantifier. Rather, van Inwagen’s ‘for all’ 16 Whitehead-Russell 1910, ‘Summary of *9’, p. 127. 17 Wittgenstein affirms that ‘in logic a new device should not be introduced in

brackets or in a footnote’, and then complains (in brackets) that, in the Principia, ‘there occur definitions and primitive propositions expressed in words. Why this sudden appearance of words? It would require justification, but none is given, since the procedure is in fact illicit’; Wittgenstein 1921, 5.452. What is alleged to be illicit is that words should be master in such a context.

18

Introduction

sounds just right. That is, the direction in which readings of the quantifiers should be made uniform is by taking the reading of ‘∀’ as the paradigm: it might just be advisable to allow the symbolism to be master for a moment, and to read ‘∃x’ as ‘for at least one x’. The instruction to be had here is to the effect that we can use the notion of quantification without introducing distracting talk of being or existence; and the advantage of doing so is that we can postpone the philosophical questions until after we have got the logical means for expressing them perspicuously. Quite apart from the reasons that have motivated substitutional readings of the quantifiers18, we may recall a particularly intriguing moment in the Principia at which Whitehead and Russell feel the need to introduce some extra symbolism, in addition to the quantifiers, for saying that something exists. This moment occurs in the introduction of the symbol ‘(ιx)’, which the authors say is to be read ‘the term x which satisfies’. The manoeuvre they then perform appears to constitute a denial that the use of this symbol involves any ontological commitment. For, Whitehead and Russell proceed to introduce the formula ‘E! (ιx) (Fx)’, which they read as ‘the x satisfying Fx exists’19. What makes this moment intriguing is that it appears to pose a choice. On the hand, the symbol ‘(ιx)’ might be read, as has occasionally been heard, as a ‘function’ (when it is called ‘the iota-function’) or as a ‘prefix’20, that is, as a symbol for a mapping from some domain onto some range, in which case, we might understand why it needs to be prefaced with ‘E!’, so as to ensure in some way that the two ends of the mapping are not, so to speak, vacant. But this is a curious doctrine on at least two counts. First, given that the only variable in play is ‘x’, it is not clear that we have a function in any of the standard senses: how 18 Such as that, if ‘∃’ is read as ‘there is a’ (or ‘there exists a’), then the logical

truth ‘(∀x) (Fx v ¬Fx)’, insofar as it implies ‘(∃x) (Fx v ¬Fx)’ and implication is transitive, will imply that there is or there exists at least one thing. But the existence of even one thing is not obviously implied by any logical truth, because logical truths imply only logical truths and it at least appears a contingent matter that anything exists. 19 Whitehead-Russell 1910, *14.02, p. 174. 20 W.v.O. Quine 1952, pp. 230-4.

Richard Davies and Andrea Bottani

19

is ‘the’ a mapping? And, second, it is hard to see what difference there is between ‘the x satisfying Fx exists’ and ‘the x satisfies Fx’. Though what is not clear or hard to see in these cases might be made clear or easy to see, the other salient option, which is the line that Russell unequivocally takes in the Introduction to Mathematical Philosophy21, is to read ‘(ιx)’ as a quantifier, meaning something like ‘for at least and at most one x’. In that case, the ‘E!’ is redundant. If it were not redundant, then it could be placed in front of any other quantified expression; thus a stretch of symbols like, ‘E! (∃x) (Fx)’ might encourage us to read ‘for at least one x that exists, Fx’. Since there is a glaring redundancy in such a case, we may infer that, if ‘(ιx)’ is a quantifier, ‘(ιx) (Fx)’ does not stand in need of ‘E!’ to say that the x satisfies Fx. Although Kroon and van Inwagen have different dialectical and expository purposes, they converge interestingly on one important point about the ‘is not’ question, namely that it is unwarranted to expect there to be a single logical form or semantical profile displayed by all negative existentials. Reconstructing the passage in Meinong’s thought as between the first (1902) and the second (1910) editions of On Assumptions, in which ‘The Theory of Objects’ (1904) is a crucial moment, Kroon suggests that Russell’s first proposal of the theory of descriptions (in 1905) is vitiated as a criticism of Meinong by his having as a target a composite (and incomprehensible) position that Meinong never occupied all at once. Specifically, on the later view, Meinong is free to explain away negative existentials such as ‘the golden mountain does not exist’ by allowing that the object ‘the golden mountain’ is assumed as a pretence, and its non-existence is asserted from outside the pretence. A solution of this sort seems also adapted to dealing with denials of the existence of objects that are picked out only by means of a description that is then said to be wrong. In the case of the assertion, sometimes attributed to Hobbes22, that the Holy Roman Empire is not holy, not Roman and not an empire, we might say that the first occurrences 21 Russell 1919, p. 177. 22 The assertion does not appear in Leviathan, IV, 47, where one might expect it.

20

Introduction

of ‘Holy’, ‘Roman’ and ‘Empire’ figure as it were in scare-quotes: they are used to evoke an entity that does not answer to the description23. In such cases, we have a phenomenon related to what Donnellan (1966) called the ‘referential’ use of descriptions, in which a description may be false of the object to which it is applied but may nevertheless succeed in picking out the object referred to in virtue of the beliefs or readinesses to collude in the pretence of those to whom the description is proffered. Where Donnellan began to probe, it has become apparent that the complexities of linguistic uses in this area cannot all be caught with a simple dichotomic distinction. As van Inwagen illustrates in his conclusion, we can count at least three different ways in which ‘doesn’t exist’ can operate, according to the kind of context in which it is deployed. Even if, then, the ‘is not’ question calls for attention to precisely what is being said in the issuing of a negative existential, there remains a question about what we are to say in general about the metaphysical status of non-existent objects. If we are to say that there are objects that are not identical to anything, this might be formalised as ‘(∃x) ¬(∃y) (y = x)’. If the quantifiers are read as unrestrictedly objectual, that is, as ranging over objects as the values of the variables within a domain identified as ‘the universe’24, then it would appear that the formalisation expresses a contradiction, if for no other reason than that it says that, for at least one x, x is not even self-identical. We might envisage two ways to defend, or at least explicate, the thesis common to Meinong, the neo-Meinongians, the possibilists and McGinn25. The first may be called ‘primitive existence’, which is a refusal of the equivalence between ‘there are non-existent objects’ and ‘there are objects not identical to anything’. The primitive existence approach appears to be an obstinate reiteration of the thesis to be defended rather than a contribution to understanding of what is meant by distinguishing between being and existence. The other line is what we may call ‘quantifier weakening’ where we might distinguish two 23 The case is not unique, of course. If we apply the same treatment to ‘Lord Privy

Seal’, we find that nouns and adjectives get up and walk about. 24 Cf. Haack 1978, p. 42. 25 These options are canvassed in an unpublished paper by Philip Percival on the metaphysical status of non-existent objects.

Richard Davies and Andrea Bottani

21

sub-versions of the weakening involved. On the one hand, there is the option of saying that the ‘there are’ in ‘there are objects not identical to anything’ should not be read objectually but substitutionally, as ranging over terms; on the other, someone might say that the ‘anything’ should not be read as unrestricted but as limited to some subdomain of objects (such as those in time and space). But, not only do these responses seem ad hoc, they also seem to miss the point. Rather, a solution of the problem might lie in finding some reading of the apparently contradictory formalisation ‘(∃x) ¬(∃y) (y = x)’ that is both clear and consistent. One approach might be to say that the key is not so much a matter of the construal of the quantifiers, as one of how to interpret the notion of identity. What is called for is clarification of the modalised notion that is in play when we speak of what might have been had a certain entity – one that does exist – not existed, and more in particular, what might have been true of that very entity itself in the case of its non-existence. In a certain sense, this approach to the ‘is not’ question is the converse of the issues raised by the ‘as if’ question: rather than consider what makes it true to say of a certain fictional character, say, Sherlock Holmes, that, had he existed, he would have lived in Baker Street, we take a certain actually existing person, say, Tony Blair, and examine what could have been said of him had he not existed. We shall return in a moment to some of the proposed ways of construing the fictional and mythological names that allow us to speak truly of things that do not exist as actual, concrete or real objects. But, first, we may consider the status that it is very tempting to attribute to a wide range of what Achille Varzi calls ‘negative events’, such as failures to happen, omissions or non-occurrences. As with ‘negative facts’, where we have a true proposition that is not true, here we seem to be up against a happening that doesn’t happen. Nevertheless, as Varzi notes, what does not happen seems to play a role in our thinking about explanations, giving rise to a tendency to think of negative events as implicated in causation and, hence, as forming part of the history of the world. For instance, it sounds perfectly natural to say that John’s failure to water the flowers was responsible for their withering. In general, there are fair grounds for resisting the idea that talk of events, whether positive or negative, can carry us to the deepest or most general level of ontological commitment. For one thing, the very

22

Introduction

phrase ‘ontology of events’ has an air of oxymoron about it: an event is what happens, rather than what is. For another, an event always presupposes some objects to which it happens: we can have things to which nothing happens, but we cannot have happenings that do not happen to something. Even if there are some turns of speech, such as ‘it is raining’, that ostensibly refer to events without an obvious substance to which they happen (it is not the sky, the cloud, the weather or even the rain that is doing the raining), it remains the case that you cannot have rain without water, but you can have water without rain. It will not quite do to say, as Donald Davidson says26, that there is a ‘symmetry’ between ‘substances and their changes’; for, if there really were symmetry, then one would be equally ready to assert that there is symmetry between changes and their substances. And this symmetry does not sound as symmetrical as the one that Davidson asserts. What stimulates Varzi’s enquiry is the widely-held view, promoted by Davidson and his acolytes, that causation is a relation among events. On that view, therefore, our referring to and quantifying over events, including negative ones, such as John’s failure to water the flowers, is a commitment made in order to report causal relations. In order to restrict the range of the events that are to be admitted as legitimate terms of causation, Varzi’s own ‘feeling for reality’ induces him to separate two elements that might mislead us into the admission of negative events. The first concerns the ways that descriptions of causal factors can be cast as either positive or negative. Thus, if John stays where he is, that event could also be described as a non-leaving. Of course, Mary’s impatience may be due to John’s not leaving, because his leaving was what she expected of him; but his non-leaving just is the ordinary (positive) event of his staying. The mere appearance of a negation-sign in the linguistic expression does not make what happens negative. The second point that Varzi makes involves a distinction between reporting a causal relation and offering an explanation, where the former would, at least in the long run, call on a causal law and the latter demands no more than some merely psychological satisfaction with a reference to a factor that helps us understand what was untoward in the case in hand. More specifically, whereas the context ‘... because –––’ 26 Davidson 1969, p. 175.

Richard Davies and Andrea Bottani

23

of causal explanation is referentially opaque, oblique or intensional, and hence dependent on the particular descriptions that flank it, the Davidsonian thesis is that the context ‘––– caused ...’ is referentially transparent, direct and extensional, and hence satisfied by any terms co-referential with the actual cause and effect. Employing a distinction of this sort, we might concede that John’s failure to water the flowers explains the death of the flowers, inasmuch as Mary had given him a task that, if he had carried it out, would have helped the flowers to survive, but we might still think that no real cause of their death has been cited. Mobilising these two devices, Varzi is satisfied that the temptation to allow negative events can be kept at bay, in much the way that the invocation of fictional contexts can be used to avoid serious ontological commitment to there having been someone who was Sherlock Holmes, when we say, truly, that Sherlock Holmes lived in Baker Street. 3. ‘As if’ While it is true to say of Sherlock Holmes that he lived in Baker Street, it is not true to say of Baker Street that Sherlock Holmes lived in it. This asymmetry is due to the fact that Conan Doyle is in some way authoritative as to some of the things that it is true to say with respect to the non-referring term ‘Sherlock Holmes’ while, even within the fiction of Conan Doyle, the reference of ‘Baker Street’ is Baker Street. The papers by Maria Reicher, Carola Barbero, Francesco Orilia and Giuseppe Spolaore address, from slightly different angles, features of how fictional discourse can generate true predications and identity statements without thereby committing us to an ontology on which fictional objects such as Sherlock Holmes increase the population of the world we live in. This is the ‘as if’ question. The most widespread strategy for dealing with fictional discourse is to say that sentences that ostensibly refer to characters such as Sherlock Holmes are to be understood within the context of some story-telling, of a pretence or of a game of make-believe. Maria Reicher investigates the scope that should be ascribed to such a context of story-telling by considering the logical behaviour of the ‘story operator’. The story operator is a device for making sentences

24

Introduction

like ‘Sherlock Holmes lived in Baker Street’ true by the addition of the rider ‘according to a story’. As it is mostly generally expounded, the story operator is read as a sentence-forming operator on sentences, which therefore embraces the whole of the initial sentence within its scope. Thus, ‘Sherlock Holmes lived in Baker Street’, becomes, ‘according to a story (Conan Doyle’s), Sherlock Holmes lived in Baker Street’, and so comes out true because, from A Study in Scarlet onwards, there are repeated indications in Conan Doyle’s Sherlock Holmes stories that our hero’s address was as reported. On this understanding of the story operator, its logical form could be rendered as, ‘Story (Fa)’. The reading of the story operator as a sentence-forming operator on sentences is offered as having the beneficial effect of ensuring that everything that falls within its scope is taken to be fictional and, hence, of not landing us with ontological commitment to the entities, such as a man by the name of Sherlock Holmes with whom we could have shaken hands and to whom ostensible reference is made. Reicher raises two objections to the standard reading of the story operator. The first is that, appearances to the contrary, there are sentences to which it cannot be applied as a way of avoiding commitment to fictional entities. For, it is pretty clear that ‘Sherlock Holmes is a detective invented by Conan Doyle’ is true, while ‘according to a story (Conan Doyle’s), Sherlock Holmes is a detective invented by Conan Doyle’ is false, because, whatever other slips there might be in the Sherlock Holmes stories, Conan Doyle never mentions himself as an inventor of anything in the stories he wrote. Hence, we would do well to admit ‘Sherlock Holmes’ as a possible grammatical subject of true sentences that do not carry the story-operator rider. The other objection to the standard reading is that, rather than treat the story operator as a sentence-forming operator, we would do well to regard it as a predicate modifier. That is to say, in place of saying, ‘according to a story (Conan Doyle’s), Sherlock Holmes lived in Baker Street’, we say, ‘Sherlock Holmes, according to a story (Conan Doyle’s), lived in Baker Street’. This has the logical structure of a simple predication of the form, ‘(Fa)’. The advantage here is that it permits us to build a more consistent and plausible theory of fictional entities. Consider, for instance, the claim that Sherlock Holmes was swifter but less subtle than Hercule Poirot. On the sentence-forming reading of the story operator, there would have to be a story such that

Richard Davies and Andrea Bottani

25

Conan Doyle’s detective is swifter but less subtle than Agatha Christie’s. But there is no such story of a suitably authoritative sort27. Hence, it is better to read the operator’s scope in such a way as to produce a sentence like: ‘Sherlock Holmes, according to a story (Conan Doyle’s), was swift to degree n and subtle to degree m, while Hercule Poirot, according to a story (Agatha Christie’s), was swift to degree nx and subtle to degree m+y’. On this predicate-modifier reading of the story operator, the comparison is across different stories. If such an interpretation of the story operator leaves us with some commitment to there being such things as Sherlock Holmes and Hercule Poirot about which stories can be told, the question arises of the status of the things in question. Whatever else we end up saying, we had better not allow ‘Sherlock Holmes’ as a term referring to a man with whom we could have shaken hands. Rather, a creature of fiction, such as Sherlock Holmes, might be conceived as a contingent abstract object that is neither physical (because we cannot shake his hand) nor mental (because no longer dependent on particular acts of Conan Doyle’s consciousness). As Venanzio Raspa explains, the Meinongian elaboration of this is to say that, while the characters of stories do not exist, they do nevertheless possess certain properties, which are not themselves created by their authors. Here, the distinction is between an act of creation, which is performed on words and sentences to express an imaginative experience, and an act of choice or selection, which involves something predetermined about its proper objects. For Meinong, the objects in question are complexes of determinations and can be regarded as higher-order objects, that is, as abstract particulars. One important feature of such objects is that they are incomplete or intensional: not all their aspects are determined. Thus, there is no truth about whether Sherlock Holmes had an odd or an even number of hairs on his head at the moment of his fight with Moriarty at the Reichenbach Falls quite simply because Conan Doyle did not specify the matter. All the same, Conan Doyle could have picked out a Sherlock Holmes with an even number of hairs (perhaps specifying which even number, or perhaps not) if it had served his purpose. 27 Fruttero-Lucentini (1989) imagine the two detectives at work on closing the

case of Edwin Drood.

26

Introduction

Pursuing this indeterminacy, Carola Barbero investigates the ways in which a fictional character can be understood as a collection, listing or cluster of descriptions that is not strictly identical with their sum. The analogy she deploys is that of the relation between a melody and the tones of which any given playing of it is made up: the melody can be transposed up or down the scale and yet remain the same melody. What remains unaltered through such a transposition is the form-quality (Gestaltqualität) of the original set of tones. On this analogy, the set of properties attributed by Conan Doyle to his character Sherlock Holmes is not one and the same with its object-correlate. In this way, the characteristically Holmesian descriptions can be transposed into a different story and even undergo significant mutations in its constitutive (or ‘nuclear’) properties, while still remaining the same fictional entity. Thus, when Michael Dibdin adopted the name and the personality for his The Last Sherlock Holmes Story28, he is free to tell a story about Sherlock Holmes in which the doings at the Reichenbach Falls are quite different from those recounted in Conan Doyle’s ‘The Final Problem’ and ‘The Adventure of the Empty House’. Where Barbero’s musical analogy stresses the formal nature of the characters of fiction, Francesco Orilia faces and responds to the question of intertextual identity on the model of identity across time. Especially when we are considering artefacts, there is reason for being shy of commitment to strong metaphysical theses about criteria of persistence in time. Thus, in the case of the ship of Theseus29, a criterion of reidentification that selects the ship at whose helm Theseus finds himself at the end of the journey, rather than the ship reconstituted of the planks and other parts jettisoned along the way, will be a conventional or pragmatic criterion. Likewise, to count Conan Doyle’s Sherlock Holmes as the same character as Dibdin’s, appeal may be made to some salient features that constitute a shared conventional essence. Orilia usefully distinguishes between, on the one hand, what optimists might think of as ‘our’ conceptual scheme and, on the other, par28 Dibdin 1978. Near the close of the book, Dibdin’s Watson says that, after the

‘apocryphal’ adventures made up by Conan Doyle, ‘Holmes had ceased to be remembered as a real figure [...]. He had become a fictional character’ (p. 189). 29 Here, Hobbes may be cited without reservation: De Corpore, XI, 7.

Richard Davies and Andrea Bottani

27

ticular theories of identity across time or of the reidentification of fictional objects. ‘Our’ conceptual scheme is, at least until corrupted by a theory, neutral as between the various considerations that do (or do not) make Conan Doyle’s Sherlock Holmes identical with Dibdin’s. For it is within the scope of our conceptual scheme that competing theories are formulated and assessed30. This leads Orilia to grant a variety of ways in which partial reidentification criteria can be adopted according to our purposes. Rather than think of a fictional character as some sort of abstract or ideal particular, Orilia’s tendency is towards Sherlock Holmes – both Conan Doyle’s and Dibdin’s – as being constituted by a listing of properties, some of which are to count as conventional fictional essences that must coincide if an identity-statement across texts is to count as true. One condition for the success of this enterprise that Orilia does not spell out, but that his account adequately meets, is that fictional characters may count as the same fiction even though they do not bear the same name. Indeed, sameness of name seems neither necessary nor sufficient for identity across texts. It does not appear necessary if we allow ourselves to say that Nero Wolfe is really Mycroft Holmes, rather than his energetic brother Sherlock. And it is not sufficient if we find ourselves faced with the 1965 film A Study in Terror and are moved simply to deny that the John Neville character is Sherlock Holmes, despite what it says in the cast-list. Returning once more to Baker Street, Giuseppe Spolaore takes up the question of the logic of reference to real individuals within the scope of the story operator. In particular, he applies two plausible-looking theses to a difficult case. The first thesis is one that we have already enunciated, that a genuine and successful referring term, such as ‘Baker Street’, that appears in a work of fiction retains its reference to its non-fictional bearer, namely a certain street in London. The second 30 The reason why we say that there is optimism in assigning this role to a con-

ceptual scheme is that it is by no means obvious that any given individual is capable of formulating more than one theory at a time and, hence, of making comparative assessments of a plurality. Indeed, Orilia’s distinction is useful, though rarely made, precisely because there is a widespread tendency to identify one’s own favoured theory with what is dictated by ‘our’ conceptual scheme and hence to regard rivals as merely incoherent.

28

Introduction

is that, in being referred to in a work of fiction, a real individual is also a character of the work in question. The difficult case is one in which two terms that are, in reality, co-referential are used in a work of fiction as names for distinct (fictional) objects. Spolaore explains how, in such a case, it seems that the two terms will refer both to distinct characters of the story and also to the same character of the story. Which means that something must give, though it is far from clear either how the rejection or qualification of the original theses or strategies of disambiguation or paraphrase will reduce the tension without generating further perplexities about the insertion of what ‘is’ into what is ‘as if’. The various approaches adopted to the interconnections among our three basic questions bring to light the ways in which the evaluation of how it is possible to speak truly and even falsely of things that are not part of Russell’s ‘real’ world still remains a stimulating and revealing testing-ground for some very general metaphysical intuitions and principles regarding the modes of existence. Earlier versions of the papers published in this volume were presented orally and discussed at a conference held under the title ‘On What (Perhaps) There Is’ in May 2005, sponsored by the University of Bergamo (Italy) under the patronage of the Italian Society for Analytic Philosophy (SIFA), with the participation of the Philosophy Department and Centre for Theoretical and Applied Ontology of Turin University, and of the Philosophy Department of the University of Eastern Piedmont at Vercelli.

REFERENCES Carnap, R., (1932), ‘Überwindung der Metaphysik durch Logische Analyse der Sprache’, English translation A. Pap, ‘The Elimination of Metaphysics through the Logical Analysis of Language’ in (ed.) A.J. Ayer, Logical Positivism, Glencoe, Free Press, 1959, pp. 60-81. Davidson, D., (1969), ‘The Individuation of Events’, reprinted in his Essays on Actions and Events, Oxford, Oxford University Press, 1980, pp. 163-80. Dibdin, M., (1978), The Last Sherlock Holmes Story, New York, Random House, 1978.

Richard Davies and Andrea Bottani

29

Donnellan, K.S., (1966), ‘Reference and Definite Descriptions’, reprinted in (ed.) S.P. Schwartz, Naming, Necessity and Natural Kinds, Ithaca, NY, Cornell University Press, 1977, pp. 42-65. Findlay, J.N., (1933), Meinong’s Theory of Objects and Values, 2nd ed., Oxford, Clarendon, 1963. Frege, G., (1884), Die Grundlagen der Aritmetik, English translation J.L. Austin, The Foundations of Arithmetic Oxford, Basil Blackwell, 2nd ed., 1953. Fruttero, C., Lucentini, F., (1989), La verità sul caso D, Turin, Einaudi. Grossmann, R., (2005), ‘Meinong’ in (ed.) T. Honderich, The Oxford Companion to Philosophy, 2nd ed., Oxford, Oxford University Press, pp. 581-4. Haack, S., (1978), Philosophy of Logics, Cambridge, Cambridge University Press. Kant, I., (1781), Kritik der Reinen Vernunft, English translation N.K. Smith, Critique of Pure Reason, London, Macmillan, 1929. Lambert, K., (1983), Meinong and the Principle of Independence, Cambridge, Cambridge University Press. Meinong, A., (1904), ‘Über Gegenstandtheorie’, English translation I. Levi (et al.) ‘The Theory of Objects’, in (ed.) R. Chisholm, Realism and the Background of Phenomenology, New York, The Free Press, pp. 76-117. –– (1910) Über Annahmen, English translation J. Heanue, On Assumptions, University of California Press, Berkeley, 1983. Quine, W.v.O., (1948), ‘On What There Is’, reprinted in his From a Logical Point of View, Cambridge (Mass.), Harvard University Press, 1953. –– (1952), Methods of Logic, 3rd ed., London, Routledge & Kegan Paul, 1974. Ramsey, F.P., (1920), ‘Facts and Propositions’ in (ed.) D.H. Mellor, Foundations, London, Routledge and Kegan Paul, 1978, pp. 40-57. Russell, B., (1905), ‘On Denoting’ Mind, 14, pp. 479-93. –– (1919), Introduction to Mathematical Philosophy, London, George Allen & Unwin. Whitehead, A.N., Russell, B., (1910) Principia Mathematica, 2nd edition (1927), abridged, Cambridge, Cambridge University Press, 1962. Wittgenstein, L., (1921), Logisch-Philosophische Abhandlung, English translation D.F. Pears, B.F. McGuinness, Tractatus Logico-Philosophicus London, Routledge, 1961.

Facts, Formal Objects and Ontology Kevin Mulligan What is a fact? Are there such things? Three answers to the first question are facts are true propositions; facts are not true propositions but rather obtaining states of affairs, atomic and non-atomic; facts are neither propositions, nor obtaining states of affairs, and are always atomic. In order to understand the relation between the three answers it is useful to suppose that there are propositions, obtaining states of affairs and facts and then ask what relations there might be between such entities. One plausible answer to this question suggests that, if there are facts, then facts are obtaining states of affairs (§1). But are there such things? One argument in favour of facts, the argument from knowledge, is presented (§2). Although no ontology should be incompatible with epistemology, a realist metaphysician will not attach much importance to an argument for facts from knowledge or from any other minddependent phenomenon. This is because such a metaphysician assumes that, if there are facts, then facts are ontologically basic. Perhaps he thinks that facts are ontologically basic and that all facts are atomic. I give some reasons for thinking that no fact is ontologically fundamental (§3). If this is correct, the argument that knowledge requires facts tells us nothing about what is ontologically basic. A further consequence of the view that no fact is ontologically basic is that the motivation for the view that facts are always atomic disappears. And that there are obtaining states of affairs of all types of logico-ontological complexity, if there are any obtaining states of affairs, is indeed suggested by the account of the relations between propositions, states of affairs and facts in §1. §1 The Ontological Zoo Propositions, states of affairs, facts, concepts, classes and properties clearly all belong together. They are creatures of a kind. Call them

32

Facts, Formal Objects and Ontology

formal objects. Certain properties and relations borne by formal objects clearly also belong together: being true, obtaining or subsisting, falling under concepts, exemplifying properties. Call these formal properties and relations. Romanesque metaphysics and ontologies dispense as far as possible with formal objects, properties and relations. Baroque metaphysics and ontologies embrace them happily. Much discussion of formal objects concentrates on the (in)eliminability of (apparent) reference to such objects or on their (in)dispensability. The interrelations between formal objects, properties and relations, on the other hand, are rarely attended to. In what follows, two types of interrelations will be important: equivalence and explanation. We suppose, to begin with, that there are propositions, facts, concepts and properties and ask what the relations between the species in our ontological zoo might be. Consider (1) Sam is sad (2) The proposition that Sam is sad is true (3) The state of affairs that Sam is sad obtains (4) Sam falls under the concept of sadness (5) Sam belongs to the class of the sad (6) Sam exemplifies the property of sadness If there are propositions etc., then the following seems very plausible: (7) (1) iff (2) iff (3) iff (4) iff (5) iff (6) Consider now the relation of explanatory priority. Should we accept (8) If (1) & (2), (2) because (1) or (9) If (1) & (2), (1) because (2)? There is one reason for thinking that we should accept neither. Instances of (10) p because p, the “because” of the exasperated parent, are all false. So a philosopher who thinks that (1) and (2) are completely synonymous or that they

Kevin Mulligan

33

express the same proposition, should reject both (8) and (9). But the fact that there are heated disagreements about the existence of propositions together with the fact that there is no such disagreement about the existence of creatures like Sam is reason enough to think that (1) and (2) are not completely synonymous. Of the two, (8) and (9), (8) is clearly more plausible than (9). There are certainly true instances of (11) If (1), (1) because p, for example, (12) If (1), (1) because Sam has had some bad news, the causal “because”. But (9) is not a true instance of (11). By the same token, we should accept (13) If (1), (3) because (1) (14) If (1), (4) because (1) (15) If (1), (5) because (1) (16) If (1), (6) because (1) But one might accept (8) and (13) and think that propositions just are states of affairs and that being true is just obtaining. Similarly, one might accept (14) and (16) and think that concepts just are properties and that falling under a concept just is exemplifying a property. There are good reasons for thinking that concepts and properties are quite different creatures. We understand concepts but not properties; concepts, unlike properties, are clear or unclear1. We perceive properties but not concepts. Concepts are expressed by predicates and properties are the semantic values of predicates. We understand a predicate if we understand what concept it expresses. Philosophers sometimes talk of our grasp of properties and of concepts. But to grasp a concept is to understand it or to understand which concept a predicate express. To grasp a property, on the other hand, is to perceive it or to know which property is the semantic value of a predicate or concept. If concepts and properties are quite different creatures, then there is a good reason for thinking that propositions and states of affairs are in

1 Cf. Strawson 1987 404, Schnieder 2004 55-6.

34

Facts, Formal Objects and Ontology

their turn very different creatures. Propositions contain concepts, only concepts, simple and complex. States of affairs, on the other hand, contain properties, simple and complex. State of affairs, not propositions, are the bearers of modality; propositions, not states of affairs, are true are false. If we continue to suppose that there are propositions, states of affairs, concepts and properties and suppose that propositions are not states of affairs and that concepts are not properties, then we may ask whether we should accept (17) If (2) & (3), (2) because (3) If the proposition that Sam is sad is true and the state of affairs that Sam is sad obtains, the proposition that Sam is sad is true because the state of affairs that Sam is sad obtains or (18) If (2) & (3), (3) because (2)? And similarly we may ask whether we should accept (19) If (4) & (6), (4) because (6) If Sam falls under the concept of sadness and exemplifies the property of sadness, Sam falls under the concept of sadness because Sam exemplifies the property of sadness or (20) If (4) & (6), (6) because (4)? One good reason for preferring (17) to (18) is the widespread intuition that truth is truth in virtue of something. It is true that not all philosophers share the intuition. But that is because these philosophers deny the existence of states of affairs, facts or properties or deny that facts differ from true propositions or concepts from properties. We are supposing that there are facts and properties and we have seen that are reasons for thinking that these are distinct from true propositions and concepts. One reason for preferring (19) to (20) is that in order to find out whether Sam falls under the concept of sadness we try to see whether he exemplifies the property of sadness. We make estimates about the

Kevin Mulligan

35

exemplification of the property of sadness not about falling under concepts. We formulate the results of our estimates in sentences containing the predicate “is sad” which expresses the concept of sadness. (1) is an atomic sentence. The state of affairs that Sam is sad is an atomic state of affairs. (17) is an instance of the completely general claim: (21) If the proposition that p is true and the state of affairs that p obtains, the proposition that p is true because the state of affairs that p obtains. In order to defend (21) an argument must be given to show that there are obtaining states of affairs. Further, some reply must be given to the objection that there are indeed facts but they are one and all atomic. In §2 I give an argument for states of affairs. In §3 I give one reply to the objection. §2 An Argument for Facts from Knowledge Does intentionality require us to accept facts? An account of the intentionality of belief and of judgement, as we shall see, does not seem to require us to accept facts. They seem to be merely an option. An account of the intentionality of a type of mental act, state or attitude should indicate whether the type has conditions of correctness and conditions of satisfaction and, if so, state these. The conditions of satisfaction and the conditions of correctness for beliefs and judgements may mention facts and propositions but need not do so. Consider (1) x judges that p (2) x judges correctly that p (3) The proposition that p is true (4) The state of affairs that p obtains (5) p If we already have reason to believe that there are propositions and facts, we may assert (6) If (2), then (3) and

36

Facts, Formal Objects and Ontology

(7) If (2), (2) because (3) (8) If (2), (2) because (4). But whether or not we endorse propositions and facts, we should accept (9) If (2), then (5) (10) If (2), then (2) because (5). But then, for the purposes of an account of the correctness conditions of judgements, facts and propositions are superfluous to requirements. The same is true of accounts of the satisfaction conditions of judgements. Consider (11) x’s judgement that p is satisfied (12) If (11), then (3) (13) If (11), (11) because (5). And what holds for judgement holds, too, for belief. Indeed, it may seem that in the case of judgement and belief there is no difference between conditions of correctness and conditions of satisfaction. Whether or not this is true, there is a clear difference between conditions of correctness and satisfaction in the case of attitudes or states other than judgement and belief. Consider (14) x desires to F (15) x correctly desires to F (16) x’s desire to F is satisfied (17) x ought to F (18) Fx becausecausal (14) (14) has both a correctness condition, (19) If (15), then (17) and a satisfaction condition: (20) If (16), then (18) Oughtness, like being true and obtaining, is formal property; it is a property which (here) takes a property to make a property. No formal property is attributed in the satisfaction condition for desires.

Kevin Mulligan

37

Emotions and sentiments, on the other hand, have correctness conditions: (21) x prefers y to z (22) x correctly prefers y to z (23) y is better than z (24) If (22), then (23) but no satisfaction conditions. And betterness, like exemplification, is a formal relation. An account of the intentionality of belief and judgement, then, apparently does not require us to accept facts. Does any type of intentionality require us to accept facts? Here is an argument to show that a plausible account of one type of knowledge requires facts. Knowledge or epistemic contact with the world comes in at least four distinct kinds. There is knowledge that p, which is no episode, and there is coming to know that p (“erkennen, dass p”), which is an episode. There is acquaintance with objects, which is no episode, and there is coming to be acquainted with (seeing, hearing) objects, making their acquaintance, which is. Philosophers sometimes mention, alongside the distinction between cognitive episodes and cognitive states or dispositions a distinction between propositional and non-propositional knowledge. This distinction may be understood as a grammatical distinction between uses of “know” where the verb takes sentential complements and uses where it does not. But it may also be used to mark a distinction between propositional and non-propositional acts or attitudes, where a propositional act involves thinking a thought and so conceptual representation. I shall assume that there are cases of coming to know that p where the subject does not think any thought – Erna sees that Sam is smiling but she thinks no thoughts. And I shall call only cases where one comes to know that p by having the thought that p “propositional coming to know that p”. The different European terms for what is supposed to be one and the same philosophical discipline highlight different members of the knowledge family – “Erkenntnislehre”, “Wissenschaftslehre”, “théorie de la connaissance”, “theory of knowledge”. One trait common to all

38

Facts, Formal Objects and Ontology

four kinds of epistemic contact with the world is identification. In visual, non-propositional perception identification and reidentification of one and the same object or property goes on continuously. Similarly, in coming to know that p and in propositional coming to know that p there is a process of identification. The visual acquaintance which is the result of coming to be acquainted with an object inherits the result of the process of identification. Similarly, the state or disposition which is knowledge that p, the result of coming to know that p, inherits the result of a process of identification. How, then, does identification work in the case where one propositionally comes to know that p? Consider (35) x perceptually and propositionally comes to know (erkennt) that Sam is smiling According to the theory of identification, (36) x comes to know → x identifies y and z. What, then, is it to identify and what is identified? The psychological process of identifying is the psychological counterpart of the binary predicate of identity although it is not any judgement or thought of identity. Just as the binary predicate cannot be flanked by sentences, so too, identification identifies objects. It is thus unlike judgings and beliefs, which connect animate beings and non-objects. What objects are identified in identification? One bad answer is (37) *x identifies that Sam is smiling, what x sees, and that Sam is smiling, what x represents There are, of course, true instances of (38) x identifies Sam’s smile, which x sees, and Sam’s smile, which x represents, perhaps by judging that this is Sam’s smile But these identifications are not examples of coming to know that Sam is smiling but are peculiar to one type of coming to be visually acquainted with something, one which involves “simple seeing” and demonstrative judgements of identity. An alternative to (37) is:

Kevin Mulligan

39

(39) x identifies the obtaining state of affairs that Sam is smiling, which x sees, with the obtaining state of affairs that Sam is smiling, which x represents (40) (35) → (39) Since I cannot think of any alternative to (39) as an account of propositionally coming to know that p, I conclude that propositional coming to know that p requires obtaining states of affairs, that is, facts. Two comments on the identification theory of knowledge are in order. First, the locution, the obtaining state of affairs that Sam is smiling, which x represents is ambiguous. It might be read as referring to a judgement or belief of x. Then the identification theory is committed to the view that propositionally coming to know that p contains, as a proper part or component, a belief or judgement, which is formed on the basis of perception. But the locution may also be read as referring simply to a propositional representation of an obtaining state of affairs. Then the identification theory is not committed to the view that propositionally coming to know that p contains belief or judgement. On each reading of the locution, the identification theory is committed to the view that propositionally coming to know that p is complex. But on the second reading, propositionally coming to know that p involves no doxastic complexity. In favour of the second reading is the fact that one may wonder whether Sam is smiling and then see that he is and so come propositionally to know that this is the case, a process which involves neither belief nor judgement. Secondly, as we have seen, judgements and beliefs have correctness conditions and, so it seemed, we may but need not say that these refer to facts. Knowledge, on the other hand, has neither correctness conditions nor satisfaction conditions. Knowledge is, so to speak, already correct and self-satisfied2. One very good reason for thinking that 2 I believe that the argument for facts from knowledge presented here is, at the very least,

in the spirit of Husserl’s account of knowledge. Certainly, Husserl clearly argues that identifying is central to knowledge and that identifying is not to to be confused with judgements of identity (LI VI §47, LI p. 790). Perusal of what Husserl says at LU VI

40

Facts, Formal Objects and Ontology

knowledge does not contain belief is the fact that our beliefs and convictions are reactions to our coming to know that p or to our apparently coming to know that p and have degrees. Reactions are positive or negative and belief, too, is positive or negative (belief and disbelief). Knowledge is never a reaction and is never either positive or negative, nor does it come in degrees. If our beliefs are reactions to propositional knowledge (“Erkenntnisüberzeugungen”) or to apparent knowledge, rather than components of knowledge, and if such knowledge requires facts, then it is apparent why we should indeed say that a belief that p is correct if the state of affairs that p obtains and because of this. Belief is a reaction to knowledge that or apparent knowledge that and knowledge that involves identifications of facts. §3 Facts are not Ontologically Fundamental In §1 we considered If the proposition that Sam is sad is true, the proposition that Sam is sad is true because Sam is sad If the state of affairs that Sam is sad obtains, the state of affairs that Sam is sad obtains because Sam is sad If the proposition that Sam is sad is true, the proposition that Sam is sad is true because the state of affairs that Sam is sad obtains These are instances of purely general claims, of, respectively, (41) If the proposition that p is true, the proposition that p is true because p (42) If the state of affairs that p obtains, the state of affairs that p obtains because p (43) If the proposition that p is true, the proposition that p is true because the state of affairs that p obtains.

§8, LI p. 696 (cf. VI §7 LI p. 691), however, will make clear why I hesitate to attribute the view presented here to Husserl. It is also true that many, although not all of Husserl’s formulations, suggest that he thought that propositional knowledge contains belief.

Kevin Mulligan

41

(41) is uncontroversial, if there are propositions. (42) is highly controversial and so is (43), which is one version of truth-maker maximalism. One reason for rejecting (42) is the view that assertions, or propositions to the effect that, for example, Sam is sad, represent or depict or are in some other way “about” the state of affairs that Sam is sad. But the assertion or proposition that Sam is sad is about Sam, it contains no representation, nominal, predicative or sentential, of any state of affairs. The fact that the state of affairs that Sam is sad is mentioned in the correctness condition for the judgement that Sam is sad does not entail that the judgement or the proposition are “about” any state of affairs. Perhaps the main reason for rejecting (43) is adherence to (44) Every obtaining state of affairs is ontologically fundamental Some friends of facts have occasionally accepted negative and disjunctive obtaining states of affairs or facts and taken these to be as ontologically fundamental as atomic facts. But most philosophers who have been prepared to accept facts have denied that non-atomic facts could be fundamental and so have denied that there are such things. Thus friends of facts who endorse (44) typically accept only those instances of (43) which countenance atomic facts. In order to decrease the implausibility of (43) I propose to argue that no fact is ontologically fundamental. Answers must therefore be given to the following three questions. What does it mean to say that something is ontologically fundamental? What are the most plausible candidates for the role of what is ontologically fundamental, if facts cannot play this role? How are the ontologically fundamental and the ontologically non-fundamental related to one another? We already possess the beginnings of an answer to the first question. Consider again (1) Sam is sad (2) The proposition that Sam is sad is true (3) The state of affairs that Sam is sad obtains (8) If Sam is sad and the proposition that Sam is sad is true, then the proposition that Sam is sad is true because Sam is sad

42

Facts, Formal Objects and Ontology

(9) If Sam is sad and the state of affairs that Sam is sad obtains, then the state of affairs that Sam is sad obtains because Sam is sad (17) If the proposition that Sam is sad is true and the state of affairs that Sam is sad obtains, then the proposition that Sam is sad is true because the state of affairs that Sam is sad obtains The “because”s in (8) and (9) tell us that (1) is more fundamental than (2) and more fundamental than (3). The “because” in (8), (9) and (17) is the essential “because”, not any causal “because”3. As far as I can see, there is no true instance of (45) (1) becauseessential p If that is right, then not only is (1) more fundamental than (2) or (3), it is fundamental. But what does it mean to say that something is ontologically more fundamental than something else and that something is ontologically fundamental tout court? The answer to our second question will help to provide an answer to this question. The second question was: What are the most plausible candidates for the role of what is ontologically fundamental if facts cannot play this role? Sam is ontologically more fundamental than any proposition and ontologically more fundamental than any fact. Now there are two rival views about just what ontological category Sam belongs to. For a very long time, it has been thought that Sam and his ilk are enduring things or substances. A more recent view has it that Sam is in fact a rather long process. The categories of substance and process are good candidates for the role of what is ontologically fundamental. To these two categories we may add that of states, such as states of sadness. It is true that if there are both substances and processes, then we may want to argue that processes are less fundamental than substances. And if we add to the list of what is ontologically basic space-time itself, then we 3 The essential “because” is essential to a correct formulation of truth-maker max-

imalism. But it is not enough. The truth-maker principle itself holds because of the nature of truth and of propositions. This “because” is the “because” of essence. On this view and its history, cf. Mulligan 2006, 2006a, 2006b.

Kevin Mulligan

43

may want to say that space-time is more fundamental than any thing, if space-time is no thing. Our candidates, so far, for the role of what is ontologically fundamental or at least relatively fundamental are all acceptable to the nominalist ontologist or metaphysician. An anti-nominalist list of candidates would add the category of kinds – kinds of things, kinds of processes, kinds of states. If substances, processes, states and kinds are ontologically fundamental or ontologically more fundamental than facts, properties and relations, then (a) nothing on the list of what is fundamental is identical with anything on the list of what is not fundamental and (b) nothing on the first list is such that it can be constructed out of what is on the second list. Let me briefly sketch how one might begin to argue for such claims. Ad (a) (a) implies all of the following: States and processes are not properties, neither multiply-exemplifiable properties nor unit-properties. Relational states, processes and events are not relations, neither multiply-exemplifiable relations nor unit-relations. Admiration, collisions and fights are relational or relation-like but are not relations. It is perhaps preferable to talk of one-legged states, processes and events, twolegged, three-legged… states and processes. Kinds are not properties. Sam’s property of barking is either a unit-property (a bearer-specific property) or a multiply-exemplifiable (bearer non-specific) property. In neither case is it identical with his barking, which is a process. The process of barking has temporal parts. Properties have no temporal parts. Properties, it may be thought, are always determinable or determinate properties. Processes and states are neither determinables nor determinates. Sam’s property of being sad is not identical with his state of sadness. His state of sadness lasts for a period of time. Sam has or exemplifies the property of being sad for a period of time. His state of sadness has boundaries. Properties have no boundaries. Things, states and processes and events all exist or do not exist in the sense of the existential quantifier. They also have modes of being. Things and states endure. Processes go on. Properties perhaps enjoy a mode of being, too; they obtain, like states of affairs and numbers. But properties neither endure nor do they occur.

44

Facts, Formal Objects and Ontology

Binary relations have order properties and converses; two-legged processes and states have no order-properties and no converses. Kinds are instantiated by instances. Sam, if he is a man, instantiates homo sapiens. His state of sadness instantiates Sadness. His jump instantiates the kind Jump. Properties are exemplified, if they are multiply-exemplifiable; they inhere if they are bearer-specific. Instantiation is not exemplification. Instantiation is always an internal relation. Exemplification may be an external relation. Kinds, like properties, may be understood in Aristotelian or in Platonistic fashion. An Aristotelian kind is always instantiated, unlike a Platonic kind. There are kinds of properties but no kind is a property. Properties are predicable. Kinds are not predicable. Man cannot be predicated because “man” is a singular term. We can of course predicate the property of being a man. On the standard view, “is a man” has no internal structure. It has been claimed that things or particulars are identical with facts4. But this view is somewhat unclear and is most plausible as a claim to the effect that we can consider things by themselves and as parts of states of affairs. Ad (b) It has been claimed that processes and states can be constructed out of properties, objects and intervals. But the many reasons for thinking that there are neither properties nor relations, for endorsing the nominalist view that there are processes and perhaps states and things but no propositions, facts, properties or relations, are presumably available also to the metaphysician who wants to distinguish between what is ontologically basic and what is ontologically secondary. According to such a metaphysician, there are propositions, facts, properties and relations but these entities are not ontologically basic. After all, there are social entities but social entities are not ontologically fundamental. The picture we have arrived at is this: Logical level: propositions, concepts Logico-ontological level: objects, properties, relations, facts Ontological level: space-time, things, states, processes, – and kinds thereof 4 Cf. Johansson 2004.

Kevin Mulligan

45

Our third question was: How are the ontologically fundamental and the ontologically non-fundamental related to one another? How is the logico-ontological level related to the ontological level? Suppose that Sam is an enduring substance. Then although it is wrong to say (46) *Sam makes Sam exist, the following are all plausible: (47) Sam makes the state of affairs that Sam exists obtain (48) Sam’s sadness makes the state of affairs that Sam is sad obtain (49) Sam’s jump makes the state of affairs that Sam jumps obtain (50) Sam’s jump over the fence makes the state of affairs that Sam jumps over the fence obtain Suppose, now, that Sam is a process. Then (46) is still unacceptable or false and (47)-(50) still hold. The least ambitious version of truth-maker maximalism, the one which involves no toil at all (because it is a simple, a priori truth), (43) If the proposition that p is true, the proposition that p is true because the state of affairs that p obtains, looks much more plausible, I suggest, when we drop the assumption that obtaining states of affairs must be ontologically fundamental. But fact-maker maximalism (51) If the state of affairs that p obtains, something makes the state of affairs that p obtains is obviously false. Nothing makes the state of affairs that there are neither witches nor genders obtain although this state of affairs certainly obtains. Friends of truth-making who reject truth-maker maximalism should really endorse (43) and reject (51). Many friends of truth-making have repeatedly asserted that truthmaking is really a relation (an internal relation). Is it? Nominalisation of part of (43) yields (52) If the proposition that p is true, the obtaining state of affairs that p makes the proposition that p true In (52) “makes” is elliptic for “because” the binary functor. A binary functor, in particular a functor that takes two sentences, is not a

46

Facts, Formal Objects and Ontology

relational expression. Nor is its semantic value, if it has one, a relation5. However, “makes” in (47)-(50) is not elliptic for the binary connective “because”. Fact-making really is a relation when the maker is ontologically basic. It is this relation and its relata that friends of truth-making should be toiling over. I have given an argument for facts from knowledge and tried to explicate and make plausible the claim that no facts are ontologically fundamental. The two projects are complementary in a way which should be attractive to all metaphysical realists. The facts which are the objects of propositional coming to know that p are, like all facts, ontologically secondary. They are the tips of icebergs. Knowledge is always given as knowledge of the tip of an iceberg. We know that there is more to what we know than what we know. References Husserl, E. 1975 Logische Untersuchungen (1900-01, 1913), (ed.) E. Holstein, The Hague, Martinus Nijhoff. Logical Investigations, English translation. by J. Findlay, London, Routledge and Kegan Paul, 1973 (abbr. as LU, LI). Johansson, I. 2004 Ontological Investigations, Frankfurt, Ontos Verlag. Mulligan, K. 2006 ‘Ascent, Propositions and other Formal Objects’, Proceedings of the May 2004 Padua conference on Propositions. –– 2006a ‘Wahrheit und Wahrmachen in 1921’, forthcoming in a volume edited by G. Imaguire & C. Schneider. –– 2006b ‘Two Dogmas about Truth-Making’, forthcoming. Schnieder, B. 2004 Substanzen und (ihre) Eigenschaften, Berlin, de Gruyter. Strawson, P. 1987 ‘Concepts and properties – or: Predication and Copulation’, Philosophical Quarterly, 37, pp. 402-406.

5 The causal “because” which takes two sentences to make a sentence may be

thought to entail that a causal relation between, say, events, holds. But the causal relation is not the semantic value of the “because”.

Fictional and Aesthetic Objects: Meinong’s Point of View Venanzio Raspa In contemporary philosophy there are a great many texts on aesthetics and the ontology of narrative discourse, that discuss Meinong’s theses, agreeing or disagreeing. But Meinong never wrote a single book or article on aesthetics or the ontology of art works, nor did he provide a structured system to deal with fictional aesthetic objects. What he wrote on these matters is to be found in texts dealing with the wide horizon represented by object theory. To assess whether object theory provides the right tools for the treatment of fictional aesthetic objects and hence deserves to be looked at in greater depth and developed further, I have thought fit to consider certain critical points from a general point of view, passing over specific analyses. Meinong does not use the term ‘fictional,’ indeed he only speaks of ‘aesthetic objects’; having brought together the two terms to delimit my discourse, I will not deal with fictitious objects in their totality, but only with the subclass of fictional aesthetic objects. By ‘fictional’ I mean ‘figuring in fiction,’ i.e. in a story or narrative context – of course not all that is fictitious is necessarily given in a narrative context; whereas by ‘aesthetic object’ I mean an object with aesthetic properties – and again, not all that is fictitious (including the fictional) is also necessarily an aesthetic object. Meinong himself refers above all to literature, sometimes to music, seldom to the visual arts, when he speaks of aesthetic objects. 1. Meinong’s discourse on fictional objects follows from certain fundamental theses of object theory. Such theses, put forward for the first time in ‘The Theory of Objects [Über Gegenstandstheorie]’ (1904), mainly come out of a process of theoretical elaboration which began years before, especially with ‘On Objects of Higher Order and their Relationship to Internal Perception [Über Gegenstände höherer Ordnung und deren Verhältnis zur inneren Wahrnehmung]’ (1899) and

48

Fictional and Aesthetic Objects: Meinong

On Assumptions [Ueber Annahmen] (1902)1. Later they underwent further development, but were never abandoned. Here I take them as premises, and do no more than set them out, without going into detail2. First of all, Meinong embraces Brentano’s thesis of intentionality, according to which every mental phenomenon (representation, judgment, feeling or desire) is directed towards an object3: For no one doubts that we cannot have a representation, without having a representation of something, and likewise, that we cannot judge without judging something4.

This intentional character secures an autonomous and specific domain for the mental and allows it to be distinguished from the physical. To these two classes Meinong adds a third, that of ideal objects, which are neither physical nor mental. He comes to this, following Twardowski, by connecting the intentionality thesis with the distinction between the act, content and object of a representation5. 1 Meinong’s works, except the first edition of Ueber Annahmen (1902), are quot-

ed from the Alexius Meinong Gesamtausgabe (1968-1978), abbreviated as GA. Translations are mine, unless indicated otherwise; references to English translations appear in brackets. 2 But cf. the already classic texts by Findlay 19632 and Grossmann 1974; for more specific logical aspects, see Lambert 1983 and Jacquette 1996. A brief exposition of the main theses of object theory can be found in Raspa 2006. 3 Cf. Brentano 18741/1924-19282: I, 124 ff., 136 ff.; II, 32. 4 Meinong 1899: GA II, 381 (1978: 141). Cf. also Meinong 1904: GA II, 381 (1960: 76): ‘That knowing is impossible without something being known, and more generally, that judgments and ideas or representations (Vorstellungen) are impossible without being judgments about and representations of something, is revealed to be self-evident;’; 1905: GA I, 582: ‘one cannot be cheered up without being glad about something. Hence one cannot feel any joy, without apprehending such a ‘something’, an object’ (see infra, fn. 8); 1921: GA VII, 15. 1974: 224. I translate ‘Vorstellung’ with ‘representation’, and not with ‘idea’ (like SchubertKalsi) or ‘presentation’ (like Levi, Terrell and Chisholm), because we find in Meinong both the terms ‘Idee’ and ‘Präsentation’. On this question cf. Lindenfeld 1980: 17, fn. 6. 5 Cf. e.g. Twardowski 1894: 3-4, 12, 18 (1977: 1-2, 10, 16); Meinong 1899: GA II, 381 ff. (1978: 141 ff.).

Venanzio Raspa

49

An act is a psychical experience (Erlebnis) directed toward an object; content is that part of such an experience that varies or remains constant with the object, and in dependence upon it6; the object is, still following Twardowski7, the summum genus, in other words that something to which no other concept is superordinate8. This is fundamental because it means that anything that can be apprehended by a mental experience or expressed by language is an object; but furthermore that ‘object’ for Meinong is not binding as regards the being or essence of something. Meinong can then distinguish between what exists and what – while either subsisting or being outside of being (außerseiend) – does not: existence is characterized by persistence in time, and is a predicate not possessed by all objects; so there are objects that, rather than existing, only subsist, having no spatio-temporal determination (such as the objects of mathematics)9, as well as objects that neither exist nor subsist, and hence are outside of being (impossible objects). Being can hence be either existence or subsistence. Such distinctions imply a criticism of the prejudice in favour of existence or being10, according to which ‘we may speak of a so-being 6 Cf. Meinong 1899: GA II, 384 (1978: 142-143); 1917: GA III, 339 ff., 347 f.

(1972: 49 ff., 55 f.). 7 Cf. Twardowski 1894: 37, 38, 40 (1977: 34, 35, 37). 8 Cf. Meinong 1921: GA VII, 14 (1974: 224). According to Meinong (1904: GA

II, 483-484 (1960: 77)), there is an equivalence between ‘object’ and ‘something’. This meaning of object is similar to that of object in general (Gegenstand überhaupt) found in the Critique of Pure Reason: the concept of an object in general – affirms Kant (17811-17872: A 290 = B 346) at the end of the Transcendental Analytic – is the highest concept (beyond the dichotomy of possible and impossible), with which it remains undecided whether the object is something or nothing. 9 Cf. Meinong 1902: 189 = GA IV, 467; 1904: GA II, 519 (1960: 108); 1906: GA V, 377 fn. 2, 387-388; 1910: GA IV, 64, 74-75 (1983: 52, 58-59); 1915: GA VI, 56-57, 61 ff.; 1918: GA V, 544; 1921: GA VII, 20-21 (1974: 228). 10 Actually Meinong speaks of a ‘prejudice in favour of the real’ (1904: GA II, 485, 505 (1960: 78, 96); 1906-1907: GA V, 235, 255), or ‘of the knowledge of reality’ (1904: GA II, 488 (1960: 81)), in addition to a ‘prejudice in favour of existence’ (1904: GA II, 489, 494 (1960: 82, 86); 1915: GA VI, 181, 201), or ‘of being’ (1904: GA II, 494 (1960: 86)). According to him ‘existence’, ‘reality’ and

50

Fictional and Aesthetic Objects: Meinong

(Sosein) only if a being (Sein) is always presupposed’11, and only reality is worthy of theoretical treatment12. They conversely presuppose the postulation of an existence-free science (daseinsfreie Wissenschaft), that is ‘an aprioristic science, independent from experience’13. The concept of Daseinsfreiheit finds expression in two fundamental principles: the principle of independence of so-being from being and the principle of the Außersein. According to the former the so-being (Sosein) of an object is not affected by its non-being (Nichtsein). […] this principle applies, not only to objects which do not exist in fact, but also to objects which could not exist because they are impossible14.

The other principle, of Außersein, often translated as ‘principle of the indifference of pure object to being’, affirms that the object as such, [...] the pure object, stands ‘beyond being and non-being.’ [...] The object is by nature outside of being [or indifferent to being (außerseiend)], although at least one of its two objectives of being, the Object’s being or non-being, in every case subsists15.

Later on I will explain what Meinong means by Außersein. Regarding the classification of objects, in addition to the above ontological one, Meinong also gives a gnoseological one: namely the classification of objects according to the psychical experiences that apprehend them, so that ‘objecta’ (Objekte) are the objects of representations, ‘objectives’ (Objektive) those of thoughts (judgments and assumptions), ‘dignitatives’ (Dignitative) those of feelings, and ‘desideratives’ (Desiderative) those of desires. We can concentrate on the first two classes of objects. The former, the objecta, can be objects of lower or of higher order, these last being characterized by an ‘intrinsic non-independence’ ‘being’ are not synonymous, but they are, in order, concepts with an increasing extension. This use of different expressions is not a symptom of linguistic inaccuracy or lack of conceptual clarity; rather, they are indicative of the various levels in which the prejudice operates (cf. Barbero & Raspa (eds.) 2005: 7-8). 11 Meinong 1904: GA II, 489 (1960: 82). 12 Cf. Meinong 1904: GA II, 486 (1960: 79). 13 Meinong 1906-1907: GA V, 239, 256-257. 14 Meinong 1904: GA II, 489, 490 (1960: 82). 15 Meinong 1904: GA II, 493-494 (1960: 86); the translation has been slightly modified (V.R).

Venanzio Raspa

51

(innere Unselbständigkeit)16, because they can only be thought in reference to other objects, on which they are built. They can be both ideal, like the relation of similarity between two things, and real, like the combination between a colour and an extension. What can exist is real17, what can subsist ideal18. Objectives are also objects of higher order: as judgment needs representations as its essential basis, so the objective needs objecta; in fact an objective can also be the inferius of another objective, but at the end of the downward series we always find objecta, by the principle of obligatory infima19. Unlike objecta, which can exist or not, objectives can never exist as a piece of reality next to the objecta that occur in them, but they can only subsist, when they are true. I have stated known Meinongian principles and definitions that I will use later on. I have done this for two reasons: because the discourse that I will develop is of a hypothetical kind, in the sense that, having assumed certain ontological theses, it investigates their consequences of an aesthetic nature; and because, since the non-acceptance of the presuppositions means that we can (not necessarily that we must) disagree on the results, proceeding in this way helps to make clear the points of divergence. My paper is structured in three parts, each of which moves from a polemic that involved Meinong; therefore, the theory is immediately subjected to a first verification. Each part intends to answer a question: (i) what kind of existence or of being is possessed by fictional aesthetic objects? (ii) what kind of objects are fictional aesthetic objects? (iii) what makes a fictional object an aesthetic one? 2. Let us begin with the first question: what kind of existence or of being is possessed by fictional aesthetic objects? To enter directly in medias res, we should consider the following passage:

16 Meinong 1899: GA II, 386 (1978: 144). 17 Cf. Meinong 1899: GA II, 394 (1978: 150). 18 Cf. Meinong 1899: GA II, 395 (1978: 150). 19 Cf. Meinong 1910: GA IV, 62-63 (1983: 50-51); 1917: GA III, 389-390 (1972:

94); 1921: GA VII, 17 (1974: 226).

52

Fictional and Aesthetic Objects: Meinong

I know that Goethe’s Mephistus is a mere poetic form, that there has never been a Mephistus, that therefore none of the words that Goethe makes him say have ever been uttered by him. Nevertheless, I can discuss how Mephistus replies to Faust or to the Lord in a certain passage. I can say that he ‘effectively’ responds in a certain way and not in another. And it should be noted that with this I do not want to formulate a judgment either on my or on Goethe’s imaginative activity, but I formulate it on the person of Mephistus. […] for me or for my consciousness Mephistus’ answer does not belong to the past, but to the immediate present. Contrarily, Goethe’s imaginative activity is without doubt a past fact. On the other hand, I do not speak at all of the historical Mephistus, but rather of Goethe’s, or more precisely of the Mephistus of the story. But he has a peculiar form of existence. Without any doubt he was once called into existence by Goethe. But having been called into existence and having reached artistic representation in the words of literature, he has a type of reality; what he does and says is in a certain sense beyond any doubt a verifiable ‘fact’20.

What type of reality does Mephistus have? An ‘aesthetic reality’ (ästhetische Wirklichkeit), is the answer of the author of this passage, where the adjective ‘aesthetic’ means that what is concerned is an invented reality, that does not have anything to do with the real world, but only with its being thought, therefore it can also be defined as an ‘existence for me’ (Dasein für mich)21. It is not Meinong who is talking, but one of his contemporaries, the philosopher Theodor Lipps, in an essay of 1905 ‘Weiteres zur ‘Einfühlung’ [Further considerations on ‘Empathy’]’, in which Meinong is called to account on questions regarding judgement-feelings, aesthetic feelings and aesthetic pleasure. Meinong responds to Lipps the same year with an article published in the same review, the Archiv für die gesamte Psychologie, and entitled ‘Judgement-feelings: what they are and what they are not [Über Urteilsgefühle: was sie sind und was sie nicht sind].’ This was written shortly after the publication of the essay ‘The Theory of Objects’, and it not only presents an aesthetic application of object20 Lipps 1905: 487-488. 21 Cf. Lipps 1905: 489; 1906: 27.

Venanzio Raspa

53

theoretical concepts which were presented in that essay, but it also anticipates many elements that will be developed more completely in Meinong’s mature works. I will not follow the whole development of the polemic, which continued with another article published by Lipps the following year under a similar title, ‘On ‘Judgement-feelings’ [Über ‘Urteilsgefühle’]’; I will instead confine myself to examining those aspects that help us to understand better Meinong’s position on fictional aesthetic objects. From the passage of Lipps the following theses emerge: (L1) the Mephistus of Goethe’s Faust never existed and, if he did not exist, he neither uttered the words that Goethe makes him say nor performed the actions that he makes him perform; (L2) but it is a fact that we can discuss what Mephistus said and did, and when we do this, we refer to Mephistus, the one of the literary story, and not to our or Goethe’s imaginative activity. How is this possible? And this brings us to a third thesis: (L3) once he has been called into existence by Goethe, Mephistus possesses a type of reality that does not have anything to do with proper reality, a type of reality that Lipps names ‘aesthetic’, and that corresponds to being thought. It is of little consequence that subsequently Lipps replaces the term ‘aesthetic reality’ with that of ‘aesthetic objectivity’ (ästhetische Objektivität)22; and we can also disregard his interchangeable use of the terms ‘existence’ and ‘reality.’ What is however significant is that Mephistus is according to Lipps an object provided with a specific kind of reality corresponding to being thought. Now, Meinong shares theses (L1) and (L2) in accordance with his critique of the prejudice in favour of the real, but he disagrees with (L3). In his opinion, we can talk about an ‘aesthetic reality’ only in a metaphoric sense. Here is what he says: Strictly speaking there is only one reality, that of the empirical world, and an ‘aesthetic reality’ that is outside of it is not reality at all23. 22 Cf. Lipps 1906: 28. 23 Meinong 1905: GA I, 599.

54

Fictional and Aesthetic Objects: Meinong

The same can be said for the notion of a ‘simple and indisputable existence of an object for me in general’, because an existence for me, in my mind, exists only in the sense of an ‘existing in a representation’, that is in the sense of a pseudo-existence (as defined by Meinong in the essay ‘On Objects of Higher Order’24). In such a case, what exists is the representational act including its content, but Mephistus cannot be understood as a psychical content, for the simple fact that he possesses some properties and accomplishes some actions that can as little be attributed to a representational content, as being cold and sweet can be attributed to the psychical representation of a ice-cream25. Despite the dissent, what remains from Lipps’s lesson is that with reference to Mephistus, on one hand we are dealing with real psychical experiences, on the other hand with an unreal object. It is a fact that literary stories arouse in us not only thoughts, but also feelings and emotions, and that from this we derive aesthetic pleasure. But how can something unreal, that does not exist, generate real effects, that do nevertheless exist? To answer this question, it is necessary to examine the notion of a ‘given’ (Gegebenes). Meinong observes that generally nobody has any difficulty accepting that reality represents a ‘given’ to which our intellectual activity is directed, but we are less prepared to consider as a ‘given’ the objects of our aesthetic attitude, because the latter addresses without distinction both the real and the unreal, and aesthetic objects are the result of an arbitrary stipulation by the author; in short, we are not prepared to see a bond (Gebundenheit) beyond existence and nonexistence26. Now, if we consider theses (L1) and (L2), it follows that even if Mephistus neither exists nor has ever existed, he has some properties. Meinong develops such a thought from his own point of view: of objects such as Mephistus it is possible to affirm the so-being independently from the being – this is the principle of independence. Moreover, again according to Lipps, what Mephistus says, i.e. his answer to Faust, ‘does not belong to the past, but to the immediate present’. For his part, Meinong assimilates this present to that which is 24 Cf. Meinong 1899: GA II, 382-383 (1978: 142). 25 Cf. Meinong 1899: GA II, 383-384 (1978: 142). 26 Cf. Meinong 1905: GA I, 600.

Venanzio Raspa

55

expressed in the proposition ‘the equilateral triangle has equal angles’, which means that the present concerned is rather a timelessness – a characteristic peculiar to non-existing objects. The action of a drama, even of a historical drama, does not have any strict time placement, at least not in absolute time, while it [the action] no more lacks relative time determination than its characters lack the past and also the future27.

Actually, not only is it difficult to identify the temporal (and spatial) determinations of a literary text (or of a symphony); it is also difficult to say what constitutes a literary work (or a symphony). Does it coincide with the original manuscript of the work (or of the score), or with every copy or authentic reproduction of the same? And what shall we say of ancient texts like Aristotle’s Metaphysics or Machiavelli’s Prince, whose originals are missing and whose copies differ on several points? The answer that Meinong suggests is that (M1) the being of a (literary or also musical) work is not existence at all, but it is a being which is disconnected from space and time, so that in certain circumstances the work can also be lost to humanity, but it can never be deprived of its own being. This audacious thesis is not free from difficulties. In fact, if, as regards timelessness, abstract objects (such as geometric ones) and fictional aesthetic objects show some affinities, from an epistemological point of view it seems that there are none: while the proposition concerning the equilaterality of the triangle holds necessarily, the one concerning the words with which Mephistus addresses Faust does not possess the same character of necessity, because the words could have been different. This means not only that in the work of art arbitrariness holds sway, but that our knowledge of aesthetic objects, as it is mediated by the text, is of the same type as our knowledge of reality, that is empirical – despite the fact that such objects are unreal. But how is it possible to consider outside of space and time what can be known only empirically? Let us see how Meinong articulates his thought in response to the objection. Here the creative activity of the artist comes into play, but we have to be quite clear about the concept of creation. If aesthetic objects do

27 Meinong 1905: GA I, 601.

56

Fictional and Aesthetic Objects: Meinong

not exist, but are given outside of space and time, then they cannot even be created. What the artist ‘creates’ is a more or less composite reality, that has the property, for those who apprehend it, to ‘mean’ something more or less composite, specifically the aesthetic object, which in this way, for those who apprehend that reality, is picked out from among the infinite totality of the objects outside of being and from whose viewpoint it can appropriately be designated as a predetermined object28.

The passage presupposes Meinong’s semiotic theory presented in the second chapter of On Assumptions, according to which among linguistic signs, psychical experiences and objects subsist the three relationships of expressing, presenting and meaning. In short, linguistic signs (words and propositions) express experiences, that is, respectively, representations and judgments or assumptions; these mean objects, or more precisely representations mean objecta, judgments and assumptions mean objectives. Therefore, meanings are always objects; these are independent of the fact of being apprehended by experiences or expressed by signs, but they become meanings only if they are presented to the thought by the corresponding psychical experiences29. Presenting means offering an object to thought by a psychical experience30. In this sense, the aesthetic object is the meaning, while the text is the sign of it. Creation has to do with the material the artist uses, signs, not objects. The writer uses real material, words and propositions, in order to signify an object that is not real, the aesthetic object, which ‘is picked out from among the infinite totality of the objects outside of being.’ A thesis confirmed in On Assumptions: objects cannot be made by us, but can only be chosen out of the infinite abundance of the objects outside of being (aus der unendlichen Fülle des Außerseienden)31 28 Meinong 1905: GA I, 603. Cf. also 1915: GA VI, 49 ff., esp. 52 fn. 1. 29 I have dealt with this subject in Raspa 2001. Cf. also Morscher 1973, Dölling

1998 and 2005, Simons & Morscher 2001. 30 Cf. Meinong 1917: GA III, 291 (1972: 6) and passim; cf. also 1910: GA IV, 244 (1983: 177). 31 Meinong 1910: GA IV, 274 (1983: 197); my italics, the translation has been modified (V.R.).

Venanzio Raspa

57

This is a consequence of a fundamental view of Meinong’s, that is, that the object is always the logical prius32. That the aesthetic object is outside of being, explains those cases in which we do not have the original of a work, but a series of copies that diverge on individual points. Moreover, as Karl Schuhmann observed, such a view ‘accounts for the fact that aesthetic properties need not be parts of real objects’ and also explains the enjoyment of unreal objects, which we can only imagine, such as those of which literary texts speak; thus it explains aesthetic pleasure. But since the real object ‘becomes an aesthetic object only indirectly, namely to the degree it is capable of directing us toward the aesthetic object itself’33, it would have the disadvantage of doubling the object of aesthetic experience, since it distinguishes between a real object and one outside of being. On the other hand, it seems that this is precisely the situation we are facing: we have a text, that speaks about something else, a story; the text and the story do not coincide; the point is to explain how they hold together. This is what Meinong does with the notions of predetermining reality and predetermined object. Meinong’s thesis is the following: representations and the real thoughts of the artist, which through real words and propositions arouse in the reader equally real representations and thoughts, mediate between the reader and the aesthetic object outside of being, whose knowledge and enjoyment, without such a mediation, would be denied to us. The objects of which a novel talks are genuine non-existing objects, and not representations in the author’s mind that correspond to those of the reader; they enter nevertheless into relation with reality, and this relation consists in the ‘being predetermined’ of objects through something real34. The artist’s thoughts, which are expressed in propositions, bring out the object from the Außersein, that is, they ‘predetermine’ it. This predetermination should be understood not as a ‘determining previously’, in a chronological sense, because Meinong maintains that the object is not created, but chosen; on the contrary, it is a logical predetermination or, if we prefer, a ‘determining for’ – for 32 Cf. Meinong 1917: GA III, 300-301, 354 (1972: 15, 62); 1921: GA VII, 22,

45, 47. 33 Schuhmann 2001: 534. 34 Cf. Meinong 1906-1907: GA V, 254.

58

Fictional and Aesthetic Objects: Meinong

itself and for others35. The predetermining reality works as such not only as long as it exists, but from the moment in which it begins to exist, and also in the case in which the knowledge of such a function and therefore of the predeterminedness itself have been lost. That is, the predetermined object remains outside of being. The knowledge about the predetermined object has the character of an empirical knowledge, inasmuch as one reaches it only through the predetermining reality, which is of course empirical. But if the nature of an aesthetic object is knowable only empirically (a posteriori), on the basis of such knowledge – Meinong affirms – its nature can be then attributed to it analytically. In this way Meinong answers the objection raised above. What is deducible from the essence, from the so-being of an object, aside from the fact that the object exists or not, and even if the way through which we get to know it is empirical, assumes then an analytical character; it assumes this, not while we are getting to know the object, but after the cognitive process is concluded. Meinong’s reasoning is similar to that introduced by Kant in the Prolegomena to any Future Metaphysics as regards the judgment ‘Gold is a yellow metal’. According to Kant all analytic judgments are a priori even when the concepts are empirical, as, for example, ‘Gold is a yellow metal’; for to know this I require no experience beyond my concept of gold as a yellow metal. It is, in fact, the very concept, and I need only analyze it, without looking beyond it36.

Likewise – Meinong affirms – we know a diamond empirically, but once it is known, the property of being nonflammable can be attributed to it analytically; equally, through Faust we know Mephistus, and what he says results analytically from the empirically-known nature of Faust. The timeless present does not pertain to the predetermining reality – Goethe’s thoughts – but to their object37. Therefore, an ‘astonishing sovereignty of poetic or, more generally, artistic imagination’ holds sway over fictional objects, nevertheless, once assumed – but we could also say stipulated – that a certain object possesses a certain property, it then has that property. 35 Cf. Meinong 1905: GA I, 604. 36 Kant 1783: Ak. IV, 267 (1997: 14-15). 37 Cf. Meinong 1905: GA I, 603.

Venanzio Raspa

59

So, for example, in modern drama, exact information is frequently given as to the age or other properties of the characters in the play. At first this information can only be of the order of assumptions. But once such assumptions are made, the characters in question are indeed of the indicated age, as if the playwright were free to do with them as he wished. He has, in fact, nothing beyond the right to make analytic judgments in Kant’s sense, according to which the gold mountain is in fact (tatsächlich) made of gold. In respect to these rights of poets, being is at disadvantage in relation to so-being38.

Predeterminedness not only pertains to aesthetic objects; actually, ‘every written or spoken discourse predetermines an object, certainly every thought, whether or not it is expressed in words’39. Our judgments about the world correspond to choices. When we formulate a judgment and express it in a proposition, we do nothing more than make a figure emerge from a background of innumerable figures. If I look before me and I say ‘the bookcase is full, I need another one’, I bring out a state of affairs, but what I see is much more. I could say what the bookcase is like, what wood it is made of, what colour, that one book is standing, another lying flat, and so on. The states of affairs, or the objectives (in Meinongian terms), are infinite if we consider all the properties, both positive and negative, that the individual objects in the bookcase possess40. In this sense, the predeterminedness can also be lost; this does not change anything as regards the given of the object, that is of the objective. On the basis of the fact that predeterminedness proceeds with absolute predominance and is conditioning both for the subject who predetermines a certain object and for those people whose intellectual activity is connected to this subject, Meinong speaks about an analogy between fictional aesthetic objects and definitions: in both cases there is a freedom in the predeterminedness that, once made, turns into a bond; in both cases the predetermined object remains outside of being; if the definition expresses the analiticity, then it is a judgment; if it 38 Meinong 1917: GA III, 374 (1972: 80); the translation has been modified

(V.R.). 39 Meinong 1905: GA I, 604. 40 Meinong shares the Kantian principle of complete determination; cf. Kant 17811-17872: A 571-572 = B 599-600; Meinong 1915: GA VI, 168 ff.; cf. also

Haller 1986: 76.

60

Fictional and Aesthetic Objects: Meinong

takes part in ‘forming’ a fictional aesthetic object, then it is an assumption41. Leaving aside the cases in which determining object and determined object seem to coincide, as in the case of an architectural construction or a natural landscape, in substance ‘the true object of an aesthetic attitude is not at all touched, at least theoretically, by the existence of reality, but nevertheless it reveals that peculiar rigid immutability [Unabänderlichkeit] of the predetermined object’42, an immutability which is connected to timelessness. Before concluding this first part of my paper, something more on Außersein should be said. What is it? It is the sphere of the pure object, in which its existence or non-existence, possibility or impossibility has been put in brackets. Certainly, the properties of an object can be indicative of its being: ‘an absurd object such as a round square carries in itself the guarantee of its own non-being in every sense’43; nevertheless, if we leave aside, as in the case of the objects of stories, their determinations of being and consider only those of so-being, the Außersein belongs to all objects, not only to fictitious, absurd or eccentric ones, but also to those that exist and that subsist44. Außersein embraces therefore all that is ‘given’, that is all possible combinations among properties and objects45. For example, Meinong says: besides the factual objective ‘Pitch is black,’ there are (in the sense of Außersein) also objectives such as ‘Milk is black’46.

In both cases we are combining two objecta into an object of higher order, which is the corresponding objective of so-being. It is not then such an eccentric idea: in Death Fugue [Todesfuge] by Paul Celan we

41 Cf. Meinong 1905: GA I, 605; 1910: GA IV, 274 (1983: 197-198). 42 Meinong 1905: GA I, 605. 43 Meinong 1904: GA II, 493 (1960: 86). 44 Cf. Meinong 1917: GA III, 306 (1972: 19); 1921: GA VII, 21 (1974: 228). An

exception is constituted by defective objects (defekte Gegenstände); cf. Meinong 1917: GA III, 304-309 (1972: 18-22); Jacquette 1996: 37 ff. 45 Cf. Meinong 1904: GA II, 492-493, 500 (1960: 85, 92); 1910: GA IV, 80 (1983: 62); 1915: GA VI, 181. 46 Meinong 1910: GA IV: 277 (1983: 199-200).

Venanzio Raspa

61

read: ‘Black milk of daybreak we drink it at evening / […] Black milk of daybreak we drink you at night’47. To have a more plastic idea of Außersein, we can think about Borges’ Library of Babel, which records all possible combinations of the twenty-two letters of the alphabet plus the space, the comma and the full stop; note: not everything possible, but all possible combinations, therefore also those that are eccentric, bizarre and contradictory. These are the axioms on which the Library is founded: ‘First: The library exists ab eterno’, that is, it has not come into existence in a certain moment, nor it will end – just like what is outside of being. ‘Second: The orthographical symbols are twenty-five in number’48 (the space, the full stop, the comma and the twenty-two letters of the alphabet), and it also contains eccentric books, such as the one which repeats the letters M C V from the first line to the last; – the analogue is given by the theory of objects of higher order and by the principle of the obligatory infima, with the difference that these are not in a defined number like that of the Library, even though all that is apprehended can also be expressed in linguistic signs, that is with the twenty-five orthographic symbols. ‘This much is already known: – affirms Borges – for every sensible line of straightforward statement, there are leagues of senseless cacophonies, verbal jumbles and incoherences’49. – As regards Meinong, we have eccentric and contradictory objects. From the axioms on which the Library is founded one can deduce that: ‘in the vast Library there are no two identical books’, which means that ‘the Library is total and that its shelves register all the possible combinations of the twenty-odd orthographical symbols’50 – the whole infinite abundance (die unendliche Fülle) of the objects outside of being51.

47 Celan 1952/1983: 41-42 (2001). 48 Borges 1956: 87 (1962). 49 Borges 1956: 88 (1962). 50 Borges 1956: 89 (1962). 51 Borges adds: ‘(a number which, though extremely vast, is not infinite)’, the

same does not hold for the Außersein.

62

Fictional and Aesthetic Objects: Meinong

‘I repeat: – it suffices that a book be possible for it to exist. Only the impossible is excluded – explains Borges –. For example: no book can be a ladder, although no doubt there are books which discuss and negate and demonstrate this possibility and others whose structure corresponds to that of a ladder’52. – Also for Meinong impossible objects do not exist, but are outside of being, and yet people talk about them, which means that they too belong to the ‘given’. If a representation or a thought could have an object only if this object were something which is, i.e. which exists or subsists, then those representations and thoughts that are not directed to beings (as in the case of literary fiction), would have no object at all; but this contradicts intentionality thesis, so strongly assumed by Meinong that it excludes the possibility of a pseudo-relation in the case of fictitious entities. The solution proposed by Meinong is to cut the bond between objectuality and existence or being; this can be done if the object is considered as outside of being53. 3. Having ascertained that the aesthetic object is not the real object – the text – but rather the object outside of being – that is, the meaning of the real object –, we return to the question: how can something unreal like a fictional object arouse real emotions and feelings? Inasmuch as it enters into connection with real objects, that is with the material that, manipulated by the artist, constitutes the work of art. Explaining how this is possible implies the answer to the second question: what kind of objects are fictional aesthetic objects? As concerns literary texts, Meinong’s thesis is that (M2) the true fictional aesthetic objects are objectives54, which are apprehended by assumptions. In a certain sense we have anticipated this answer in our brief outline of Meinong’s semiotic theory: texts are made up of words and propositions, which mean objecta and objectives, which take on the role of meanings, in that they are presented either by representations or by judgments and assumptions. Below I will speak about psychical 52 Borges 1956: 93 fn. 1 (1962). 53 Cf. Meinong 1910: GA IV, 233-234 (1983: 170). 54 Cf. Meinong 1910: GA IV, 318, 319 (1983: 227, 228); 1917: GA III, 372 (1972:

79).

Venanzio Raspa

63

experiences and aesthetic feelings not because these are the direct subject of my discourse, which focuses instead on ontological questions, but because the psychological treatment constitutes, according to Meinong, the access route to the ontological treatment of aesthetic objects: we have said that the way we approach aesthetic objects is empirical, only later can we pass over to the aprioristic treatment; the thesis should be proved through the analysis of the empirical material that puts us in contact with aesthetic objects, that is then the only material which is in first place accessible to us. Let us first of all define assumptions, relating them to judgements and representations. Judgment possesses the two moments of conviction and position, which means that it claims truth and is either affirmative or negative; besides this, it also needs representations as its essential fundaments, because every event of the psychical life, which is not itself a representation, presupposes the representing55. Assumptions occupy an intermediate position between representations and judgments56: they are affirmative or negative like judgments, but without claiming truth like representations57. Assumptions are expressed by interrogative, optative, and imperative propositions, by subordinate clauses (daß-Sätze) that occur in propositions such as ‘I fear, I suppose, I contend that p’58, in hypothetical reasoning, in lies, in games and in narrative works. In all such cases the conviction that the object exists is not required. According to Meinong, also our attitude toward the aesthetic object does not demand at all the belief that it exists. Previously, he had maintained that our aesthetic attitude is based above all on representing, because the judgment has as its peculiarity the moment of conviction; but he decidedly changed his mind when he discovered assumptions which, insofar as they lack the moment of conviction, can take the place of judgments and express objectives that – as we have said – are 55 Cf. Meinong 1902: 1-3, 256; 1910: GA IV, 1-4, 46, 339 (1983: 9-11, 39, 242);

1917: GA III, 290, 294 (1972: 6, 9). 56 Cf. Meinong 1902: 277; 1910: GA IV, 367 (1983: 262). 57 Cf. Meinong 1902: 257; 1910: GA IV, 3, 340, 368 (1983: 10, 242, 262-263);

1921: GA VII, 33. 58 Cf. Meinong 1902: 26 ff.; 1910: GA IV, 33 ff. (1983: 30 ff.)

64

Fictional and Aesthetic Objects: Meinong

also objects of aesthetic feelings, for whose expression representations are therefore inadequate. Hence the value attributed to assumptions. In On Assumptions there are different arguments to support the aforesaid thesis, i.e. that (M2) the true aesthetic objects are objectives and that these are apprehended by assumptions. I will introduce three of them; the first dates back to the first edition (1902), the others appear in the second (1910). The first argument (M2a) concerns the assumptions themselves. These occur – as we said – in the cases of fiction, within the realm of ‘as if’, and conspicuously in lies, in games and, as far as we are interested, in art59. There is a close analogy between the game of the child, who ‘pretends’ (fingiert) or ‘imagines’ (sich einbildet) to be somebody else, and art, particularly the art of acting. But, if we consider the point of view of the writer, he too, while creating the work, puts himself in the place of all the people he represents; sometimes he tells true stories, but what he mostly deals with is fiction and ‘fiction is just assumption’60. Let us now put ourselves in the reader’s place: what happens to the reader reading a story he does not believe, because he knows that the story never happened? Although not believing it, while reading he makes the same assumptions as the narrator; obviously, this does not exclude an elaboration of the material assumed by the reader. Certainly, as I said above, the narrator does not express only assumptions, but he can also express true judgments; the effect on the reader is often however identical and even if the narrator enunciates truths, the reader does not have to accept them as they are. Anyway, Meinong concludes, we stay on the level of assumptions61. Subsequently, in On Emotional Presentation [Über emotionale Präsentation] (1917), Meinong develops a distinction that accounts for this difference, distinguishing between ‘shadowy’ assumptions and assumptions that are similar to judgments. I will not discuss this matter here, as I will not deal with the indeterminedness of fictional objects, even if discussing them would help to give a more complete picture of the whole matter62. 59 Cf. Meinong 1902: 36-37; 1910: GA IV, 107 (1983: 81). 60 Meinong 1902: 45; 1910: GA IV, 115 (1983: 86). 61 Cf. Meinong 1902: 57-60; 1910: GA IV, 127-130 (1983: 94-97). 62 But cf. Raspa 2001 and 2005b.

Venanzio Raspa

65

For our purposes, two points are fundamental: first, wherever there are assumptions, there are objectives too63, because objectives are the objects of assumptions, as of judgments; and since in literature we are dealing primarily with assumptions, though judgments are not excluded, objectives are the true aesthetic objects of narrative works; the second point is that assumptions play a prominent role in art64, since our attitude toward aesthetic objects does not demand at all the conviction that these exist, and indeed the objectives that occur in art works are not generally believed, but assumed. The second argument (M2b) examines the representations. Those expressed in a literary text are produced or imaginary representations, at most they can be reproductions65; in every case we are dealing with composite representations, to which correspond, from an ontological point of view, objects of higher order, that is complexes66. But an objective is implicit in every complex; therefore, to grasp a fictional complex we need to grasp the objective implicit in it. The reasoning which supports this thesis was suggested to Meinong by Russell67: in the object of the form ‘A and B in the complex C’ there is already present an objective of the form ‘the complex that is constituted by A and B’, or ‘A, which with B constitutes a certain complex’, even if the form tends to hide it. So ‘where there is a complex, there is also an objective as an integrating factor in it, and one who wants to apprehend the complex cannot do it otherwise than by apprehending the objective’68. That is why representing is not sufficient to apprehend objects of higher order, whether they are objects or objectives, but assuming is the most suitable experience to apprehend aesthetic objects, and again it follows that these are mainly objectives. This thesis (M2), which Meinong maintains in the first edition of On Assumptions and reasserts in the second, is firmly opposed by Stephan Witasek, one of Meinong’s pupils, who in his Outlines of General 63 Cf. Meinong 1910: GA IV, 131 (1983: 98). 64 Cf. Meinong 1902: 210-211; 1910: GA IV, 168-169 (1983: 124). 65 Cf. Meinong 1910: GA IV, 11, 16, 377-378 (1983: 15, 18-19, 269). 66 On this cf. Raspa 2005a. 67 Cf. Russell 1904/1973: 50 ff. 68 Meinong 1910: GA IV: 280 (1983: 202).

66

Fictional and Aesthetic Objects: Meinong

Aesthetics [Grundzüge der allgemeinen Ästhetik] (1904) developed a structured theory which, though it agrees on the whole with Meinong’s view, diverges from it concerning the role of objectives and assumptions69. But before discussing Witasek, let us briefly examine the third argument (M2c) by which Meinong answers the question about the effects of aesthetic objects. (M2c) According to Meinong, besides intermediate mental facts between representations and judgments, that is assumptions, there are also intermediate mental facts between representations and feelings, such as the feelings related to works of art, which Meinong calls ‘phantasy-feelings (Phantasiegefühle),’ and which arise for example when a tragedy evokes fear or compassion70. As assumptions resemble judgments in virtue of the opposition of affirmation and negation, but they are not judgments, so phantasy-feelings are feelinglike in virtue of the opposition of pleasure and displeasure, but they are not real feelings. At the theatre the spectator does indeed experience something in himself, for example something similar to compassion, which however is not literally compassion. The compassion for the tragic heroine Margaret is not only of a lower degree, but even qualitatively different from the compassion that I feel for an unhappy girl whose sad tale I know71. It is necessary to distinguish between phantasy-feelings and aesthetic feelings, which are genuine feelings, for instance aesthetic enjoyment. Indeed, phantasy-feelings, together with assumptions, play a fundamental role in the arousal of aesthetic feelings. From reading works of art arise also phantasy-desires (Phantasiebegehrungen), as when one wishes a novel to end in a certain way, and also these are no more genuine desires than phantasy-feelings are genuine feelings. The presuppositions for such mental phantasy-experiences are assumptions, which are expressed by the propositions of the work, but they

69 A reading of Witasek’s aesthetics is given by Smith 1996 and Schuhmann 2001.

According to Schuhmann (2001: 518), ‘Witasek’s theory is based on Meinong’s early general theory of value as worked out mainly in Meinong’s PsychologicalEthical Investigation on Value Theory from 1894.’ Cf. also Schuhmann 2001: 533. 70 Cf. Meinong 1910: GA IV, 309 (1983: 221). 71 Cf. Meinong 1910: GA IV, 312, 316-317 (1983: 223, 226).

Venanzio Raspa

67

refer to objectives72. Therefore, assumptions go beyond the intellectual sphere and extend to the emotional one; it follows that objectives belong also to phantasy-feelings and phantasy-desires. Hence, since objectives considered as aesthetic objects arouse in us feelings and emotions, they are also often objects of our emotional activity. Witasek recognizes the value of objectives, but he does not believe that they are aesthetic objects73. Like Meinong, Witasek believes that objectives play a fundamental role in the narrative arts, since novels, plays and poems are composed of propositions and, to understand and enjoy a literary text, it is necessary to understand the objectives meant by propositions74. Certainly, objects of representation also play their part, because no objective can subsist without objecta, but they play a more indirect role, since every reader imagines what is described differently, nor is it necessary that he should imagine it in a distinct way. The main role is played by situations, events, narrated facts and, along with them, by the relations subsisting between the characters and objects of the story. ‘But these are objectives, not objects of representation’75. What has aesthetic effectiveness in a literary work is in the first place its content, and this is constituted by the meanings of the single propositions which – as we know – correspond to objectives. Nevertheless, Witasek maintains that (W1) ‘objectives are not, as it appears at first sight, the genuine aesthetic objects, on the contrary they are mediators of aesthetic objects’76. If we mean by ‘aesthetic object’ what our feeling of pleasure and displeasure is directed to, then, since this feeling is never directed to an objective, which is neither beautiful nor ugly, the objective cannot be an aesthetic object. The 72 Cf. Meinong 1910: GA IV, 313-315 (1983: 224-225). 73 At first, Witasek (1904: 53 ff.) puts objectives in the class of the so-called ‘aes-

thetic elementary objects (ästhetische Elementargegenstände),’ that is objects ‘whose aesthetic qualities do not come down to the sum of the aesthetic qualities of the components of the object, but belong only to the object as a whole and are lost when one analyses it’ (p. 35); however, after a careful critical examination – briefly presented above –, he excludes them (p. 179). 74 Cf. Witasek 1904: 56. 75 Witasek 1904: 57. 76 Witasek 1904: 167-168; my italics (V.R.).

68

Fictional and Aesthetic Objects: Meinong

aesthetic quality, the beauty of a literary work depends not only on its language and style, but undoubtedly on its content, and finally on the objectives in which it consists, since those who do not understand the content of a work cannot say they have enjoyed it. Yet, an objective is not beautiful in itself, but it is mediator of beauty. The aesthetic properties of a work depend on objectives, ‘as the perfume of flowers on the wind that blows’77, says Witasek. Witasek distinguishes an epistemological level and an aesthetic one. An objective contains the objects of representations and it serves to describe facts and situations, in which – to put it colloquially – we always speak about people and things. The objective is therefore fundamental for the understanding of the work, since understanding a work – as aforesaid – means understanding the objectives of the propositions that compose it. Nevertheless the aesthetic effectiveness of an objective depends on the words and the representations through which it is communicated to the subject or thought by him; and only if words and representations have aesthetic properties, do objectives also possess them, but only indirectly. What counts is the choice of the expressions, their position, etc.78. To demonstrate this, Witasek advances two arguments. (W1a) First of all, for the aesthetic effect it is important how an objective is expressed. Witasek therefore invites us to transform a line of poetry into a non poetic form but with the same meaning, i.e. the same objective – in short, to translate it into prose: naturally the line loses all aesthetic value. The aesthetic properties of the saying ‘youth is a garland of roses, old age a crown of thorns’, depend on the words and the representations expressed by them. If we paraphrase it in this way: ‘In youth life is easy, happy and cheerful, in old age difficult and lacking in pleasures, but in exchange it is venerable’, the objective loses aesthetic effectiveness. (Leaving aside the additions, Witasek believes that the two propositions express the same objective.) The thought of the objective enriches the material of representation and hence the material which arouses aesthetic feelings in us, but the objective is not the direct object of our aesthetic attitude79. 77 Witasek 1904: 169. 78 Cf. Witasek 1904: 170-171. 79 Cf. Witasek 1904: 171-173.

Venanzio Raspa

69

(W1b) The second argument put forward by Witasek to show that objectives are only mediators pertains to feelings, which are in the case of the narrative arts above all empathy-feelings (Einfühlungsgefühle) and sympathy-feelings (Anteilsgefühle). The former are feelings through which a person who reads, for instance, the scene of ‘Margaret in jail’ feels with the heroine what she feels: torment, confidence, pious resignation, desperation; while the latter are the liking and the compassion that the reader feels for Margaret. In the first case, the subject reproduces in himself the psychical life of the object (Margaret), in the second he reacts emotionally to the object80. These feelings occur where there are objectives, which again act as mediators: through them feelings are aroused in fact in the reader, and only inasmuch as the objectives do this, do they have aesthetic effectiveness81 – which, as we know, depends on the words and the objects of representations. Moreover, if the feeling of aesthetic pleasure were immediately directed at the objectives, then judgments and assumptions should be the presupposition of such a feeling, but judgments and assumptions – Witasek maintains – play such a role in the case of value-feelings (like loving, appreciating, honouring) and knowledge-feelings (doubting, understanding); if they were the presupposition of aesthetic feelings too, the difference between the former and the latter would disappear. This difference consists not in the fact that value-feelings are based on the real world, and phantasy-feelings on the world of imagination; on the contrary, phantasy-feelings are also real, they are characterized by the specificity of their presupposition, namely assumptions, so they are always assumptive feelings82. Not only, but often assumptions too can act as premises for value-feelings, as when I imagine that I can lose my brother and I regret it; in this case, we have a phantasy-feeling of value. Equally it is not excluded that a judgment can arouse aesthetic feelings. The difference between value-feelings and aesthetic feelings lies in the fact that the former have objectives as object, while the latter have objecta, that is the objects of representations. 80 Cf. Witasek 1904: 148-149. On empathy-feelings and sympathy-feelings cf.

Smith 1996: 220 ff. 81 Cf. Witasek 1904: 174-175. 82 Cf. Witasek 1904: 120-121 fn..

70

Fictional and Aesthetic Objects: Meinong

The aesthetic attitude is not directed to objectives at all, but primarily to the objects of representation that they communicate (vermitteln) to the subject83.

Witasek’s thesis (W1) is diametrically opposed to Meinong’s thesis (M2); we see now how the latter answers the criticisms. A part of Meinong’s reply (MR) to Witasek’s first criticism (W1a) is already implicit in the second argument we have seen above (M2b), that is, that objectives are implied in the complex objects of representations. (MRW1a) As regards language, Meinong argues that, indeed, after having transformed a poem into prose, there is very little poetry left in it; the question is, however, whether this transformation has not also changed the objective. One who confers aesthetic dignity on a certain objective, neither sees it deprived of its objectum-material, nor expressed in a passing form, where within certain limits there is no modification of the material. Rather, the aesthetic significance is dependent on the determinations of the real material, and precisely on those referred to by Witasek. The fact remains however that, for the reasons rehearsed above, the true aesthetic object is the objective84. Meinong maintains therefore a close relationship between objective and the linguistic form that expresses it – paraphrase is not admitted in poetry –, but he does not say anything to explain this thesis, leaving the reader disappointed. For instance, the problem of translation remains open, a problem that the theory of objectives on the other hand seems to face; besides, Meinong himself maintains that the same objective can be expressed in different forms: ‘A exists’ is equivalent to ‘the existence of A’85. As regards Witasek’s second argument (W1b), Meinong’s reply is structured in two parts. (MR-W1b') Firstly, he contests the equality between phantasy-feelings and assumptive feelings, as well as the notion that phantasy-feelings are real feelings. Now, that they are not real feelings is shown by the fact that – as we said – the compassion towards Margaret is both of lower degree and qualitatively different from the compassion for an unfortunate girl who is really living. That they are not simply assumptive feelings can be deduced from the 83 Witasek 1904: 179. 84 Cf. Meinong 1910: GA IV, 318-319 (1983: 227-228). 85 Cf. Meinong 1910: GA IV, 56 (1983: 46); 1915: GA VI, 27-28 and fn. 3.

Venanzio Raspa

71

consideration that many phantasy-feelings are not founded on assumptions, as when we remember a strong toothache or a person’s death; therefore, the sphere of assumptive feelings is narrower than that of phantasy-feelings86. The second part of the argument (MR-W1b'') is far more important. Witasek believes that feelings which have objectives as their object can only be value-feelings, not aesthetic feelings. Now, according to Meinong, value-feelings are always directed to being, in the first place to existence, obviously also to the existence of something that is thus and so, to its properties, in every case to existence. This does not hold for aesthetic feelings: they always bear on sobeing, accidentally also on the so-being of something existing, and this so clearly that the kind of treatment from which they arise can be considered existence-free (daseinsfrei). This point of the existence-freedom (Daseinsfreiheit) is fundamental: whether something exists or not is never a concern of the aesthetic attitude, rather the converse happens, that is, that something which can also exist is thus and so; instead, value-feeling is always turned toward an existing thing, even if it is determined in a certain way. Hence, against Witasek, the essence of value-feelings does not depend on judgment and assumption, but always on the nature of the objective. If feeling is directed to the being, it is a value-feeling, if it is directed to the so-being, it is an aesthetic feeling. Certainly, there are value-feelings that are directed to objecta such as a color or a form, and not to objectives, but an objective of sobeing is implicit in qualitative determinations87. It follows that every aesthetic feeling is directed to objectives, if not explicitly, at least implicitly88. It is evident that the contrast between the two authors on the role of the objective, which was underlined with the opposition of (W1b) and (MR-W1b), depends on different conceptions (here very briefly sketched) as regards feelings in their various forms. I do not intend to develop this matter further, which – as aforesaid – is not specifically of an ontological character; I wanted however to mention it, because, according to both authors, in this way one comes to speak of aesthetic objects. 86 Cf. Meinong 1910: GA IV, 316-317 (1983: 226-227). 87 Indeed, as we saw above (M2b), an objective is always implicit in a complex. 88 Cf. Meinong 1910: GA IV, 319-320 (1983: 228-229).

72

Fictional and Aesthetic Objects: Meinong

4. Hitherto we have spoken of aesthetic objects as a proper class of fictional objects, now the question arises of their specificity relative to fictional objects in general, insofar as it is obvious that not all fictional objects are also aesthetic objects. So we come to the third question: what makes a fictional object an aesthetic one? And here another point of divergence between Meinong and Witasek emerges, perhaps even more fundamental than the previous one: (W2) aesthetic objects are not objects of higher order. This thesis, if proven, would significantly reduce the claims to universality of the object theory as regards aesthetic objects: objects of higher order are both objectives and objects of phantasy-representations, that is complexes. In the Outlines of General Aesthetics, Witasek states that ‘an object becomes an aesthetic object, if it is bearer of aesthetic properties’89. Of these the most characteristic is ‘beauty’, which, like the others, ‘proves to be not a real property of its bearer, but an ideal one’90. The beauty of a picture, indeed, is not perceivable as the picture and the masses of colour are; if we listen to a melody or a poem, we hear sounds or words, but beauty consists neither in the individual sounds or words nor it is added as something existing alongside them. Moreover, beauty is not an objectual property of the object, that is, a property which, like real properties, is represented together (mitvorgestellt) with the representation of the object and can then describe it: for example, the colour is an objectual property of the picture but the similarity of this with a copy is not. So beauty is also an extra-objectual determination of its bearer, though rightly it is designated as a property of the latter; even a representation, however complete, of an object does not need to contain the thought of beauty, at least insofar as it intends to be only the representation of this object91.

Again, in the Outlines of General Aesthetics Witasek had said that an object is aesthetic when it is the object to which our feeling of pleasure and displeasure is directed, a definition that obviously depends on that given above.

89 Witasek 1904: 27. 90 Witasek 1904: 14. 91 Witasek 1904: 15-16.

Venanzio Raspa

73

In his last writing, ‘On Aesthetic Objectivity [Über ästhetische Objektivität]’ (1915), as regards aesthetic judgment, ‘This (object) is beautiful (not beautiful, ugly, etc.)’, Witasek maintains that In making the judgment ‘A is beautiful’, one is not normally thinking about a relationship of A with the subject, with their emotional reaction, nor about this same emotional reaction; one means a property of A, that is entirely in the same object A.92.

If we call the object A ‘substratum’, we can say that the aesthetic object consists in this substratum taken together with the aesthetic properties. The bearer of these qualities is not however the transcendent real (existing) object, but the immanent (pseudo-existing) object, and likewise aesthetic properties are only an objectual part of this immanent object. Witasek expressly denies that the bearer can be an existing or subsisting object, insofar as he believes that the evidence of certainty regarding aesthetic judgment holds only for the immanent object93. The essential characteristics of an aesthetic object are the non-independence (Unselbständigkeit) from the substratum, which means that the being of an aesthetic object is founded on the being of another or other objects, and the dependence (Abhängigkeit) on variations of the substratum. An aesthetic property such as ‘beautiful’ requires indeed not only, like the property ‘red’, a substratum to belong to, but also a property or a complex of properties as its basis, without which it could not be and on whose nature it is dependent. Indeed, if one or more elements of the substratum change, the object can change from beautiful to ugly. Changing some notes of a melody, for instance, it can become less beautiful or even ugly94. This dependence is unidirectional in the sense that the aesthetic property is dependent on the substratum and on its properties, not vice versa. This specific dependence relation makes the difference between aesthetic properties and sensory properties: the colour red needs an extension, but it does not vary with the variations of the latter; furthermore, while the aesthetic property can vary if sensory properties like colours vary, the converse does not hold. However, a short consideration will bring out evident 92 Witasek 1915: 91-92. 93 Cf. Witasek 1915: 94, 96-98. 94 Cf. Witasek 1915: 105, 108, 110-112.

74

Fictional and Aesthetic Objects: Meinong

analogies between aesthetic properties and objects of higher order such as, for example, similarity. Indeed, similarity does not occur without similar objects; moreover, whether, and to what extent, two objects are similar, depends on the nature of the objects. Therefore aesthetic properties seem to be ideal objects of higher order95. But four factors lead us to exclude this possibility. Witasek is speaking only of beauty, but his discourse can also be extended to other aesthetic properties. First of all, (W2a) thinking back to the example of similarity, Witasek observes that this is between the two members, for example red and orange, with which it builds a complex; beauty is instead on the red-orange complex: if the difference is removed, the two colours are each for itself, while if beauty is removed, the complex remains. Besides, (W2b) beauty does not need a plurality of inferiora, as objects of higher order do, but a unity that can be, in turn, both complex and simple; for objects of higher order, though, being based on a unity is a limiting case (that of identity)96. In third place, (W2c) objects of higher order are tied to their inferiora by a bond of aprioristic necessity: given red and orange, they are necessarily different or similar (according to their gradation); while there is no necessity relationship between the two colours and beauty. Finally, (W2d) difference is not thinkable intuitively (anschaulich), while beauty is intuitively apprehended through its bearer (melody, picture, poem); in other words, the difference between two empirical objects is actively produced by the subject through an operation of comparison, the beauty of a perceptive object is instead passively accomplished. For all these reasons, beauty cannot be considered as an ideal object of higher order97. But if beauty is not something real close to the object, and it is not an ideal object of higher order, then it neither exists nor subsists. Hence, there must be either a new class of objects which have till now not been contemplated by object theory, but which subsist objectively like ideal objects of higher order, or a third type of being, which Witasek does not identify with the Außersein, but with the immanent being98. 95 Cf. Witasek 1915: 112-114. 96 Cf. Meinong 1899: GA II, 394 (1978: 149). 97 Cf. Witasek 1915: 180-183. 98 Cf. Witasek 1915: 191-192, 198.

Venanzio Raspa

75

If the later Witasek is right, much of Meinong’s theory concerning aesthetic objects comes apart. It is therefore absolutely necessary for him to reply to his friend and pupil. He does so by accepting the first part of Witasek’s discourse, according to which an aesthetic object is non-independent of being and dependent on the so-being, and rebutting the criticisms made by him99. However, Meinong does not answer each of the four criticisms separately, but he develops a single argument, in which he examines the first two, of a strongly ontological character, while he replies very quickly to the last two (of a gnoseological character), concerning the lack of aprioristic necessity and of productive activity. Meinong faces the main theoretical difficulty raised by Witasek against the thesis according to which (M3) aesthetic objects are objects of higher order, more precisely they are complex bearers of beauty, that is, the oneness (Einsheit) of the substratum required by aesthetic properties (as W2a and W2b maintain). I note in advance that Meinong identifies Witasek’s error as his having considered objects of higher order only from the point of view of objecta, and not also from that of objectives. His refutation (MR-W2a-b) can be divided into three steps: (1) objectives are also objects of higher order; (2) there are objectives that are not based on a plurality of inferiora; therefore, (3) if not all objects of higher order need a plurality of inferiora,then there is no difficulty in considering aesthetic objects as objects of higher order. Let us examine the argument in detail. (1) The theory of objects of higher order was initially formulated from the point of view of objecta, but also objectives are objects of higher order. For instance, compared to the objective ‘A is B,’ ‘it is a fact that A is B’ is an objective of higher order. Hence, ‘each objective is an objective of higher order with respect to another objective if the latter occupies the place of the objectum in the former’100. The objective is 99 Cf. Meinong 1917: GA III, 387-388 (1972: 92-93). 100 Meinong 1917: GA III, 389 (1972: 94).

76

Fictional and Aesthetic Objects: Meinong

also an object of higher order with respect to the objecta of which it is constituted, like the objective ‘A is B’ with respect to the objecta A and B; therefore, ‘all objectives as such are objects of higher order’101. As regards the height of order, it holds that objects of higher order are built on objects of lower order in such a way that the former could not be there, if the latter were not there at first. The ‘at first’ indicates a logical priority, not a chronological one. This holds both for objecta and for objectives. An objective is necessarily based on some material, which may, in its turn, contain other objectives, but these are ultimately based on objecta. That means that objectives can never occupy the position of infima. (2) Relations and complexes certainly require at least two elements; the same holds for objectives of so-being of the type ‘A is B’; but objectives of being (‘A is’), of existence (‘A exists’), of subsistence (even of Außersein) are monadic by nature, and it does not make any difference that they can be based both on a simple and on a complex substratum of objecta. This being so, it is wrong to extend to all objects of higher order a characteristic that is common to relations and complexes, that is, that they need a plurality of inferiora102. (3) If not all objects of higher order need a plurality of inferiora, the main obstacle to considering aesthetic objects as objects of higher order, that is the oneness of substratum, no longer subsists. Their dependence on the substrata is a sign – just as Witasek showed – of the superius character of aesthetic objects, which – Meinong concludes – ‘are indeed subsumable under the notion of objects of higher order’103. As regards Witasek’s criticism of a gnoseological character (W2c and W2d), Meinong’s reply is very short and also vague. Regarding the first, (RM-W2c) Meinong denies that the bond of aprioristic necessity between inferiora and superius is a proper character of all objects of higher order: leaving aside real complexes and relations, it would be enough to consider objectives concerning non-aprioristic knowledge104. As for the other criticism, (RM-W2d) Meinong replies simply 101 Meinong 1917: GA III, 390 (1972: 94). 102 Cf. Meinong 1917: GA III, 390-391 (1972: 95). 103 Meinong 1917: GA III, 391-392 (1972: 95). 104 Cfr. Meinong 1917: GA III, 391 (1972: 95)

Venanzio Raspa

77

that the productive activity of the subject arises in apprehending objects of higher order in very different ways: if such productive activity cannot be disregarded in the comparative apprehension of a difference, it is however not noticeable with certainty in the apprehension, for instance, of a melody105. Lipps made his criticisms from a point of view external to object theory, Witasek from an internal one; both acknowledge that fictional aesthetic objects have a kind of mental existence, which is denied by Meinong. For him fictional aesthetic objects do not exist, but they are disconnected from space and time (M1); besides, like all fictional objects, they are objectives which are apprehended by assumptions, or they require an objective, in that this is implicit in every complex representational objectum (M2); finally, they are specific objects of higher order, more precisely they are beauty-bearing complexes (M3). To conclude, it should be specified that, despite the differences we have examined, Meinong’s and Witasek’s considerations do agree on many points. The divergences are the result of the different focus of their respective discourses: Witasek has the tendency to consider as aesthetic objects not only the objects of literature, but also those of music and of the visual arts – as the few examples given imply –, he lacks however a broader perspective, which he infers when necessary from Meinong; Meinong, on the other hand, concentrates almost exclusively on fictional objects (with some excursions into music), which he tries to frame within the wide compass constituted by the object theory. Insofar as the two demands are unifiable, both Meinong and Witasek developed only a part of the work. References Albertazzi, L., D. Jacquette & R. Poli, (eds.) (2001), The School of Alexius Meinong, Aldershot/Burlington USA/Singapore/Sydney, Ashgate. Barbero, C., & V. Raspa (eds.) (2005), ‘Il pregiudizio a favore del reale. La teoria dell’oggetto di Alexius Meinong fra ontologia e epistemologia’, special number of Rivista di Estetica, n.s., 30, n. 3.

105 Cfr. Meinong 1917: GA III, 389 (1972: 93)

78

Fictional and Aesthetic Objects: Meinong

Borges, J.L., (1956), Ficciones, in Obras completas, Buenos Aires, Emecé [English translation in Borges 1962]. –– (1962), Labyrinths. Selected stories & other writings, ed. by D.A. Yates & J.E. Irby, Preface by A. Maurois, Norfolk (Conn.), Laughlin. Brentano, F., (18741/1924-19282), Psychologie vom empirischen Standpunkt, 2 Bde., Leipzig, Duncker & Humblot, 18741; 3 Bde., (ed.). von O. Kraus, Leipzig, Meiner, 1924-19282. Celan, P., (1952/1983), Mohn und Gedächtnis, Stuttgart, Deutsche VerlagsAnstalt, 1952; in Gesammelte Werke, Erster Band, Frankfurt am Main, Suhrkamp, 1983, pp. 7-78 [English translation in Celan 2001]. –– (2001), Selected Poems and Prose of Paul Celan, translated J. Felstiner, New York/London, Norton. ‘Death Fugue’ is available at http://osf1.gmu.edu/~lsmithg/deathfugue.html. Dölling, E., (1998), ‘Zeichen und Annahmen. Alexius Meinongs zeichenphilosophische Untersuchungen’, in Kodikas, 21, pp. 1-15. –– (2005), ‘‘... dieser Umweg führt über sprachliche Ausdrücke, durch die sich Annahmen verraten’: Eine semiotische Sicht auf Meinongs Annahmenlehre’, in Meinong Studies / Meinong Studien, 1, pp. 129-158. Findlay, J.N., (19632), Meinong’s Theory of Objects and Values, (19331) Oxford, Clarendon Press. Grossman, R., (1974), Meinong, London/Boston, Routledge & Kegan Paul. Haller, R., (1986), Facta und Ficta, Stuttgart, Reclam. Kant, I., (17811-17872), Kritik der reinen Vernunft, Riga, Hartknoch; in Ak. IIIIV. –– (1783), Prolegomena zu einer jeden künftigen Metaphysik, die als Wissenschaft wird auftreten können, Riga, Hartknoch; in Ak. IV, pp. 253-383 [English translation: Kant 1997]. –– (1910 ff.), Kants gesammelte Schriften, hrsg. von der Königlich Preußischen Akademie der Wissenschaften, Berlin, Reimer; Berlin/Leipzig, de Gruyter & Co., [abbreviated as Ak.] –– Bd. III: Kritik der reinen Vernunft ( 2. Aufl. 1787), 1911. –– Bd. IV: Kritik der reinen Vernunft (1. Aufl.). Prolegomena. Grundlegung zur Metaphysik der Sitten. Metaphysische Anfangsgründe der Naturwissenschaft, 1911. –– (1997), Prolegomena to any future metaphysics, with an introduction by L.W. Beck, Upper Saddle River (NJ), Prentice-Hall Inc. Jacquette, D., (1996), Meinongian Logic. The Semantics of Existence and Nonexistence, Berlin/New York, de Gruyter. Lambert, K., (1983), Meinong and the Principle of Independence. Its Place in Meinong’s Theory of Objects and its Significance in Contemporary Philosophical Logic, Cambridge, Cambridge University Press. Lindenfeld, D.F., (1980), The Transformation of Positivism. Alexius Meinong and European Thought, 1880-1920, Berkeley, University of California Press.

Venanzio Raspa

79

Lipps, Th., (1905), ‘Weiteres zur ‘Einfühlung’’, in Archiv für die gesamte Psychologie, 4, pp. 465-519. –– (1906), ‘Über ‘Urteilsgefühle’’, Archiv für die gesamte Psychologie, 7, pp. 1-32. Meinong, A., (1899), ‘Über Gegenstände höherer Ordnung und deren Verhältnis zur inneren Wahrnehmung’, Zeitschrift für Psychologie und Physiologie der Sinnesorgane, 21, pp. 182-272; in GA II, pp. 377-471 [English translation: Meinong 1978]. –– (1902), Ueber Annahmen, Leipzig, Barth (= Zeitschrift für Psychologie und Physiologie der Sinnesorgane, Ergänzungsband 2). –– (1904), ‘Über Gegenstandstheorie’, in Untersuchungen zur Gegenstandstheorie und Psychologie, hrsg. von A. Meinong, Leipzig, Barth, pp. 1-50; in GA II, pp. 481-530 [English translation: Meinong 1960]. –– (1905), ‘Über Urteilsgefühle: was sie sind und was sie nicht sind’, Archiv für die gesamte Psychologie, 6, pp. 21-58; in GA I, pp. 577-614. –– (1906), Über die Erfahrungsgrundlagen unseres Wissens, Berlin, Springer (= Abhandlungen zur Didaktik und Philosophie der Naturwissenschaft. Sonderhefte der Zeitschrift für physikalischen und chemischen Unterricht, I, 6, pp. 379-491); in GA V, pp. 367-481. –– (1906-1907), ‘Über die Stellung der Gegenstandstheorie im System der Wissenschaften’, Zeitschrift für Philosophie und philosophische Kritik, 129, pp. 48-94, 155-207; 130, pp. 1-46; in GA V, pp. 197-365. –– (1910), Über Annahmen, zweite, umgearbeitete Auflage, Leipzig, Barth; in GA IV, pp. 1-389, 517-535 [English translation: Meinong 1983]. –– (1915), Über Möglichkeit und Wahrscheinlichkeit. Beiträge zur Gegenstandstheorie und Erkenntnistheorie, Leipzig, Barth; in GA VI, pp. XV-XXII, 1-728, 777-808. –– (1917), ‘Über emotionale Präsentation’, Kaiserliche Akademie der Wissenschaften in Wien. Philos.-histor. Klasse, Sitzungsberichte, 183, 2. Abh.; in GA III, pp. 283-476 [English translation: Meinong 1972]. –– (1918), ‘Zum Erweise des allgemeinen Kausalgesetzes’, Kaiserliche Akademie der Wissenschaften in Wien. Philos.-histor. Klasse, Sitzungsberichte, 189, 4. Abh.; in GA V, pp. 483-602. –– (1921), ‘A. Meinong [Selbstdarstellung]’, in Die deutsche Philosophie der Gegenwart in Selbstdarstellungen, mit einer Einführung hrsg. von R. Schmidt, Leipzig, Meiner, Bd. 1, pp. 91-150; in GA VII, pp. 1-62 [Partial English translation: Meinong 1974]. –– (1960): ‘The Theory of Objects’, translated by I. Levi, D.B. Terrell, and R.M. Chisholm, in Chisholm, R.M. (ed.), Realism and the Background of Phenomenology, Glencoe (Ill.), Free Press, pp. 76-117. –– (1968-1978), Alexius Meinong Gesamtausgabe, hrsg. von R. Haller und R. Kindinger gemeinsam mit R.M. Chisholm, Akademische Druck- und Verlagsanstalt, Graz [abbreviated as GA].

80 ––

Fictional and Aesthetic Objects: Meinong

(1972), On Emotional Presentation, translated, with an introduction by M.L. Schubert Kalsi, with a foreword by J.N. Findlay, Evanston (Ill.), Northwestern University Press. –– (1974), ‘Meinong’s Ontology’, in Grossmann 1974: pp. 224-229. –– (1978), ‘On Objects of Higher Order and their Relationship to Internal Perception’, in A. Meinong, On Objects of Higher Order and Husserl’s Phenomenology, ed. by M.-L. Schubert Kalsi, The Hague/Boston/London, Nijhoff, pp. 137-200. –– (1983), On Assumptions, edited and translated, with an introduction by J. Heanue, Berkeley/Los Angeles/London, University of California Press. Morscher, E., (1973), ‘Meinongs Bedeutungslehre’, Revue Internationale de Philosophie, 27, pp. 178-206. Raspa, V., (2001), ‘Zeichen, ‘schattenhafte’ Ausdrücke und fiktionale Gegenstände. Meinongsche Überlegungen zu einer Semiotik des Fiktiven’, Zeitschrift für Semiotik, 23, pp. 57-77. –– (2005a), ‘Phantasie, Phantasieerlebnisse und Vorstellungsproduktion bei Meinong’, Meinong Studies / Meinong Studien, 1, pp. 95-128. –– (2005b), ‘Forme del più e del meno in Meinong’, in Barbero & Raspa (eds.) 2005, pp. 185-219. –– (2006), ‘La teoria dell’oggetto’, in Storia dell’ontologia, M. Ferraris (ed.), Milano, Bompiani (forthcoming). Russell, B., (1904/1973), ‘Meinong’s Theory of Complexes and Assumptions’, Mind, n.s. XIII, pp. 204-219, 336-354, 509-524; repr. in Essays in Analysis, ed. by D. Lackey, London, Allen and Unwin, 1973, pp. 21-76. Schuhmann, K., (2001), ‘Meinongian Aesthetics’, in Albertazzi, Jacquette & Poli (eds.) 2001, pp. 517-540. Simons, P.M. & E. Morscher, (2001), ‘Meinong’s Theory of Meaning’, in Albertazzi, Jacquette & Poli (eds.) 2001, pp. 427-456. Smith, B., (1996), ‘Pleasure and its Modifications: Stephan Witasek and the Aesthetics of the Grazer Schule’, Axiomathes, VII, 1-2, pp. 203-232. Twardowski, K., (1894), Zur Lehre vom Inhalt und Gegenstand der Vorstellungen, Wien, Hölder; repr. mit einer Einleitung von R. Haller, München/Wien, Philosophia, 1982 [English translation: Twardowski 1977]. –– (1977), On the Content and Object of Presentations, translated and with an introduction by R. Grossmann, The Hague, Nijhoff. Witasek, S., (1904), Grundzüge der allgemeinen Ästhetik, Leipzig, Barth. –– (1915), ‘Über ästhetische Objektivität’, Zeitschrift für Philosophie und philosophische Kritik, 157, pp. 87-94, 179-199.

Russell’s Descriptions and Meinong’s Assumptions1 Frederick Kroon 1. Introduction Russell is often portrayed as persistently misrepresenting the relationship of his theory of descriptions to Meinong’s theory of objects. It is sometimes pointed out that he misrepresented the origins of the theory as having its roots in his revulsion at Meinong’s acceptance of unreal objects (in fact, he had rejected Meinong’s view in an earlier Fregeinspired phase when he took definite descriptions to be ‘complex concepts’ that might denote nothing)2. And he misrepresented the content of Meinong’s theory of objects – the thing that explained his sense of revulsion – when he identified Meinong’s non-existent objects as objects that ‘must subsist in some shadowy Platonic world of being’ (Russell 1959, p. 84). One suspects that Russell was not guilty of deliberate misrepresentation when he represented his own work and that of Meinong in these terms. It is more likely that he had simply forgotten, or perhaps had grown to believe what others had said about the virtues of his theory and its relationship to Meinong’s work. Or perhaps he simply misunderstood, or conflated, various things he once knew or had heard. One place that may suggest an interpretation of this kind is a curious passage that occurs early on in Principia Mathematica. Suppose we say: ‘The round square does not exist.’ It seems plain that this is a true proposition, yet we cannot regard it as denying the existence of a certain object called ‘the round square’. For if there were such an object, it would

1 I am grateful for useful comments from participants at the Bergamo conference

‘On what (perhaps) there is’, as well as from members of an audience at the Australasian Association of Philosophy conference (New Zealand Division), University of Otago, December 2005. Special thanks to David Chalmers, Alberto Voltolini, and Markus Weidler, and to the Bergamo conference organisers, Andrea Bottani and Richard Davies. 2 See, for example, Ostertag (1998), pp. 5-6.

82

Russell’s Descriptions and Meinong’s Assumptions

exist: we cannot first assume that there is a certain object, and then proceed to deny that there is such an object.’ (Whitehead-Russell 1962, p. 66)3.

The first point to make is how little the first part of this quote engages with Meinong’s views. Meinong does not think that the round square has any kind of being, whether existence or subsistence. As a response to Meinong, then, the first part of the quote simply begs the question. But the more striking puzzle is this. Consider the way Russell invokes the idea that speakers first assume that there is an object, and then deny what they assume. If an unreal object such as the round square does indeed exist, however, then a speaker who ascribes nonexistence to this object is simply guilty of a de re inconsistent claim. It is hard to see how the speaker’s utterance involves first assuming that a certain object exists, and then denying the assumption. So the quote represents us with a hermeneutic puzzle: what led Russell to put matters in this misleading way? I want to offer a (speculative) solution to this puzzle, as well make a more substantial philosophical claim. The speculative solution, whose details will emerge in the course of the argument for the substantial claim, is that Russell remembered something Meinong once wrote. Meinong, it turns out, did indeed link negative existential judgements with affirmative existential assumptions, although not quite in the way Russell describes. But even if I am wrong in this speculative claim, there is a more important claim waiting in the wings. It is that Russell would have saved himself a lot of trouble had he himself accepted something like Meinong’s appeal to assumptions, and, moreover, that he could have accepted it without serious damage to his overall position. The paper is structured as follows. In the next section, I describe a problem for Russell’s account of the logical form of negative existentials involving descriptions, and suggest a Russellian solution. This solution is one that no one will care to adopt – it seems to turn negative existentials into self-contradictions – but I later argue that, properly interpreted, it constitutes a promising way of reconciling some of Meinong’s views about negative existentials with the kind of “robust sense of reality” that informed Russell’s own analysis. In section, 3 I

3 This passage occurs as part of Russell’s proof that definite descriptions are not

(logically) proper names but incomplete symbols.

Frederick Kroon

83

begin the task of articulating this reading of Meinong by describing Meinong’s Assumption View as articulated in the second edition of his On Assumptions (Meinong 1910). Because this view presupposes Meinong’s infamous commitment to non-existent objects, it would still offend Russell’s “robust sense of reality”, and so section 4 considers a weakened version of the view, one that retains the appeal to assumptions while giving up the appeal to non-existent objects. (Meinong defends a similar view in the 1902 edition of On Assumptions, which predates his discovery of non-existents.) Section 5 offers the finale: it shows how Meinong had himself tried to apply such a weakened Assumption View to the case of negative existentials, that Russell had known about the attempt (this arguably solves the first, hermeneutic puzzle), and that, properly interpreted, this way of understanding negative existentials provides Russell with a solution to the problem facing his theory of negative existentials. 2 Russell on Denoting There is a broad consensus among analytic philosophers about the central core of Russell’s theory of definite descriptions (Russell 1905; Whitehead-Russell 1962, *14). While there are issues that continue to excite lively debate, such as how to understand incomplete definite descriptions, it is widely agreed that definite descriptions are expressions whose logico-semantic role is to be understood in terms of quantifiers and predicates, even if not in quite the notational form preferred by Russell4. On Russell’s own version of the theory, the first clause of the theory, directed at predications involving definite descriptions, tells us that a sentence of the form ‘the F is G’ has the logical structure ‘there is exactly one x such that x is F, and this x is G’. (In the formalism of Principia Mathematica *14.01, [(ιx)(Fx)] G(ιx)(Fx) is defined as (∃x)((∀y)(Fy ≡ y = x) & Gx))5. Call this the predicative part of the theory.

4 Gareth Evans calls Russell’s way a ‘butchering of surface structure’ (Evans

1982, p. 57), and prefers a Russellian account on which descriptions are a form of binary quantifier (Evans 1982, §4.2). See also Neale (1990), chapter 2. 5 That is, G(ιx)(Fx) is definitionally equivalent to (∃x)((∀y)(Fy ≡ y = x) & Gx) so long as the scope of the definite description (ιx)(Fx) is the entire G(ιx)(Fx).

84

Russell’s Descriptions and Meinong’s Assumptions

The consensus continues to be impressive where existential statements are concerned. Russell was adamant that ‘exists’ cannot function as a genuine predicate, and took ‘exists’ in statements of existence to be a kind of dummy predicate, eliminable on analysis in favour of quantifiers and standard predicates. According to Russell’s formulation of the ensuing existential clause of the theory, ‘the F exists’ has the logical structure ‘there is exactly one x such that x is F’. (In the formalism of Principia Mathematica *14.02, E!(ιx)(Fx) is defined as (∃x)(∀y)(Fy ≡ y = x).) It follows straightaway from this second, existential part of Russell’s theory that a negative existential statement like ‘the golden mountain doesn’t exist’ is not, at the level of logical structure, a negated subjectpredicate statement, but a negated existentially quantified sentence: at the level of logical structure, ‘the golden mountain doesn’t exist’ says ‘it is not the case that there exists exactly one x such that x is a golden mountain’. Russell, of course, saw this analysis as proof that the truth of such a negative existential does not involve a commitment to such creatures of darkness as nonexistent golden mountains. Even philosophers not particularly wedded to Russell’s theory of descriptions tend to find such an analysis persuasive. Knowing and approving of Kant’s attack on the idea of existence as a determining predicate and of Frege’s claim that existence is a second-level, not a first-level, concept, they see Russell’s analysis as the natural culmination of such ideas. Of course, the broad consensus in favour of something like Russell’s theory fails where negative existentials involving names and other supposedly directly referential expressions are concerned. Given current scepticism about descriptivist theories of names, most philosophers now think that Russell’s theory gives the wrong account of negative existentials involving names but the right account, or close to the right account, of negative existentials involving definite descriptions (descriptive negative existentials). This is reflected in the fact that while much is being written about the former, there is virtually nothing being written about the latter. This is surprising, because the problems facing theories about the former are surprisingly close to the problems facing theories about the latter. It is often pointed out, for example, that we have every reason to construe a negative existential statement like ‘Hamlet doesn’t exist’ as a genuine subject-predicate statement that involves a referring use of

Frederick Kroon

85

the name ‘Hamlet’, which in turn suggests that in using such a sentence speakers are implicitly committed to there be something for the name to refer (uses of names satisfy a ‘commitment condition’). But the same is surely true of ‘the golden mountain doesn’t exist’. Even if (contra free logicians) ‘the golden mountain’ is not a genuine referring term but an expression that can only be defined in context in the manner of Russell, there is a good deal of evidence to suggest that the use of such a phrase in a negative existential brings with it commitment to there being a unique golden mountain, even as speakers deny that there is such a mountain. One piece of evidence involves claims like the following. Consider first the name-using complex negative existential: (1) Hamlet — you know, the Danish prince who apparently hates his uncle because he thinks the uncle killed his father — doesn’t in fact exist. He is a fictional invention. Here the description after ‘Hamlet’ occurs in a relative clause that serves to identify the individual who is the subject of the negative existential, a role that confirms a sense in which the speaker is committed to there being an individual for the name to refer to. At the level of logical syntax, this co-pairing of name and description seems structurally no different from the co-pairing of name and description in a sentence like ‘George W. Bush — you know, the once hugely popular 43rd President of the U.S. — is not in fact so popular anymore’. But such a pairing of names and descriptions in negative existentials doesn’t distinguish names from definite descriptions. We can similarly say: (2) The golden mountain — you know, the elusive mountain described in Smith’s book on famous journeys of exploration — doesn’t in fact exist. It is a mythical mountain. Here the relative clause containing the second description identifies the entity that is presented by the first description, and so the use of the first description once again appears to commit the speaker to there being such an entity. From the point of view of the commitment condition, there is no difference between the use of names and of definite descriptions in negative existentials. The problem that this presents for Russell is in some ways deeper than the problem posed by the supposedly non-descriptive nature of names, for it challenges a core part of Russell’s theory of descriptions,

86

Russell’s Descriptions and Meinong’s Assumptions

one that is independent of the question of the nature of names. The problem is how, given the commitments apparently carried by utterances of negative existentials like (1) and (2), his account of negative existentials can possibly provide the right analysis. The option that is surely most in the spirit of the “robust sense of reality” that for Russell inspired the theory is to regard the two definite descriptions ‘the golden mountain’ and ‘the elusive mountain described in Smith’s book on famous journeys of exploration’ as components of a larger description (‘the entity that is at once the golden mountain and the elusive mountain described in Smith’s book on famous journeys of exploration’), which is then to be eliminated from the context ‘… does not exist’ by using the existential part of Russell’s theory. This yields the reading: (2a) There does not exist a unique object that is at once uniquely a golden mountain and uniquely an elusive mountain described in Smith’s book on famous journeys of exploration. This is formally equivalent to: ~(∃x)((∀z)(Gz ≡ z = x) & (∀z)(Sz ≡ z = x)). But this reading is surely quite unintended. It simply loses the force of the identification, since the truth of (2a) is guaranteed by the non-existence of the golden mountain. But the truth of (2) shouldn’t be so easy to get. Consider a world in which there is no golden mountain and in which Smith has never authored a book. (2) is surely false in such a world (the embedded relative clause presents an intuitively incorrect way of identifying the golden mountain), even though (2a) it true. What Russell’s theory can’t do, it seems, is capture the fact that (2) identifies the golden mountain with the mountain described in Smith’s book, prior to declaring that this mountain doesn’t exist. Such an identification requires an affirmative identity sentence ‘The golden mountain is the elusive mountain described in Smith’s book on famous journeys of exploration’. Since for Russell such an identity sentence affirms that there is a certain kind of entity, while the negative existential claim denies there is such an entity, this provides us with a blatantly inconsistent claim as the analysis of (2): (2b) It is not the case that there is exactly one golden mountain; and there is exactly one golden mountain and exactly one elusive mountain described by Smith in his book on famous journeys of exploration, and the first is identical to the second.

Frederick Kroon

87

Formally: ~(∃x)(∀z)(Gz ≡ z = x) & (∃x)((∀z)(Gz ≡ z = x) & (∃y)((∀z)(Sz ≡ z = y) & x=y)). This analysis also suffers from another problem. Unlike (2a), it doesn’t capture the intuitive logic of the relative clause in (2). That clause portrays itself as qualifying a single occurrence of the description ‘the golden mountain’ (‘the golden mountain – you know, the elusive mountain …’), with ‘does not exist’ attached to this single occurrence; (2b), on the other hand, essentially posits two occurrences of the description, and severs all connection between these occurrences. Actually, there is a way in which Russell’s theory in ‘On Denoting’ could be adapted to capture this feature of the relative clause, while also retaining the fact that the relative clause posits an identity. But it is a way that goes directly against Russell’s official view of existence. We could insist, contrary to Russell’s explicit declaration, that ‘exists’ can after all function as a genuine first-level predicate, that it does so here, and that the descriptions in a negative existentials like (2) should be eliminated in accordance with Russell’s theory of descriptions for standard predicative contexts by giving the negation operator in ‘… does not exist’ narrow scope relative to the definite descriptions6. This would yield: (2c) There is exactly one golden mountain and exactly one elusive mountain described by Smith in his book on famous journeys of exploration to which it is identical; and it does not exist. Formally: (∃x)((∀z)(Gz ≡ z = x) & (∃y)((∀z)(Sz ≡ z = y) & x=y) & ~ (∃x). But even though (2c) seems to capture more of the apparent logical form of (2) than (2b), it again looks hopeless from Russell’s point of view. Not, I should stress, because the view that ‘exists’ stands for a 6 The claim is not that this would work for all descriptive negative existentials.

Certain descriptive negative existentials should no doubt be analysed in terms of a wide-scope negation operator. (Consider ‘The politician to beat Tony Blair doesn’t exist’ and contrast this with ‘The politician who beat Tony Blair doesn’t exist’, the latter uttered in response to a fake news-story about some alleged politician who defeated Blair in a leadership coup.)

88

Russell’s Descriptions and Meinong’s Assumptions

property of objects is straightforwardly incompatible with Russell’s official view of existence. There is, in fact, overwhelming evidence to suggest that for Russell ‘exists’ does have such a first-level use. Although Russell’s official view is that existence is a property of propositional functions, he sometimes explicitly uses ‘exists’ as a firstlevel predicate in order to motivate the view that existence as it features in negative existentials has a different role to play. When he argues this way, he makes it plain that he takes this first-level predicate to apply to every object there is: that it is, in short, a universal predicate7. Thus consider the passage encountered earlier: … we cannot regard [‘the round square does not exist’] as denying the existence of a certain object called ‘the round square’. For if there were such an object, it would exist. (Whitehead-Russell 1962, p. 66)

Similarly: If Romulus himself entered into [the statement ‘Romulus existed’], it would be plain that the statement that he did not exist would be nonsense, because you cannot have a constituent of a proposition which is nothing at all.” (Russell 1956, p. 242)

And If there were any fact of which the unicorn was a constituent, there would be a unicorn and it would not be true that it did not exist. (Russell 1956, p. 248)

It seems, then, that Russell himself is perfectly willing to use ‘exists’ as a first-level predicate. But since he clearly construes ‘exists’ so used as a universal predicate, one possessed by everything, it is plain that incorporating such a predicate and then assigning negation narrow scope in the manner of (2c) can only result in inconsistency: it simply cannot be the case that there is/exists a golden mountain which, among other things, does not exist. Hence (2c) seems every bit as obnoxious as (2b). 7 A large number of philosophers now endorse view that ‘exists’ is a first-level

predicate true of everything, among them Kripke (Kripke 1973), Peter van Inwagen (e.g., van Inwagen 1977), and Gareth Evans (Evans 1982). Even Frege had a version of this view (Frege 1979). It is worth pointing out that Kripke and Frege are both on record as thinking that this makes it misleading to call ‘exists’ so used a genuine predicate. (Perhaps this is also the reason Kant had for refusing to count existence as a genuine, i.e., a determining, predicate.)

Frederick Kroon

89

Call the narrow-scope reading ‘there is a unique F, and it doesn’t exist’ the narrow-scope analysis of the descriptive negative existential ‘the F doesn’t exist’8. Because of their problematic nature, Russell can scarcely afford to endorse such an analysis in the case of (2). But if he rejects the analysis, he seems to have no option but to regard (2) as a locution that stands in need of drastic paraphrase, even though the paraphrase fails to capture evident logical properties of the original. Either option looks unpalatable. 3. Meinong’s Assumption View9 By contrast, a view like Meinong’s seems tailor-made for dealing with such cases. Meinong would say that the committed use of terms in negative existentials is to be understood in terms of a commitment to unreal objects that, despite not existing, allow of being identified in different ways10. But of course such a response would scarcely have satisfied Russell, no matter how troubling the commitment problem he himself faced. For Russell famously thought that such a manoeuvre portrays “a failure of that feeling for reality which ought to be preserved even in the most abstract studies”, and that a “robust sense of reality is very necessary in framing a correct analysis of propositions about unicorns, 8 Gareth Evans explicitly considers the impact of taking ‘exists’ as a predicate

when formulating the existential part of Russell’s theory of descriptions, and claims that “[t]he scope ambiguity of [a descriptive negative existential] is difficult to perceive, because a reading on which the negation sign has narrow scope is obviously unintended”. (Evans 1982, p. 347) 9 This section and the following are based on Kroon (1992). 10 If we grant Meinong the distinction between nuclear and extranuclear properties, we can say that the golden mountain, specified in terms of its nuclear properties as golden and a mountain, can also be specified in terms of its extranuclear properties as the elusive object described in Smith’s book on great journeys of exploration (for a modern exposition of the distinction, see Parsons 1980). Alternatively, using the two-modes-of-predication view introduced by Meinong’s student Ernst Mally and more recently defended by Ed Zalta, we might say that the golden mountain encodes being golden and a mountain, but exemplifies being elusive and an object described in Smith’s book on great journeys of exploration (Zalta 1988).

90

Russell’s Descriptions and Meinong’s Assumptions

golden mountains, round squares, and other such pseudo-objects” (Russell 1919, pp. 47-8). My own view is that the problem Russell faces here is severe enough to show that he has to yield on something, but that what he should yield on is not, pace Meinong, his “robust sense of reality” but rather the way he sees this sense as helping to “frame a correct analysis of propositions about unicorns, golden mountains, and other such pseudo-objects”. What Russell should have said is that “a robust sense of reality” helps us to be confident that any apparent commitment to such entities is bound to be merely apparent, without prejudging how language-users manage to make a merely apparent commitment. Furthermore, I think he had a lot to learn from Meinong on this score. In particular, he had a lot to learn from Meinong’s Über Annahmen, a work that first came out in 1902 and was reissued in a second, revised edition in 1910 (references to the second edition will be to the 1983 translation On Assumptions [Meinong 1910]). One of the issues that confronted Meinong in the second edition of On Assumptions is what makes us certain that some given predication ‘O is F’ is true, where term O can stand for either an existent or a nonexistent object. Meinong’s distinctive answer goes as follows. Suppose we want to know what some object O is like. To discover O’s properties it is not enough that I have a representation of O, perhaps a mental image. I must in addition apprehend the object or, as Meinong puts it, intend the object, where intending an object is an active mental act which makes knowledge of an object’s properties possible11. Meinong’s account of intending comes in two parts. If I believe that object O has some form of being (existence or subsistence), then my intending O is the mental act of judging that O has that form of being. By thus intending O, I am then able to address the question of O’s properties, perhaps by exploiting the perceptual evidence that led me to believe that O exists, or, if O is a mathematical object, by constructing 11Note that Meinong has an almost Kantian view of why something like intending must

take place for knowledge to be possible. Representations as such are passive, ambiguous and often virtually subconscious (pp. 171-172). Only through the making of an appropriate judgment of being does one apprehend the object: ‘Representation needs, as something indispensable, the help of an apprehension that is mediate, one going through the objective [of being].’ (p. 174)

Frederick Kroon

91

proofs whose validity requires O’s subsistence (p. 174). But what about objects that we cannot intend in the way described because we do not positively believe that they are existent or subsistent. In particular, what about such objects as the golden mountain, the round square, phlogiston, Hamlet — objects that we know lack being? Meinong’s answer, which he calls the Assumption View, is as follows. Under conditions of agnosticism or disbelief we intend objects, and thereby gain a degree of cognitive access to them, by assuming that they have being. In Meinong’s own crisp summing-up: ...in order to give a thing some thought, a person “places himself in the situation in which there is such a thing” (p. 175).

Note that assuming a proposition or objective in this way carries no implications about belief, provisional or otherwise. Indeed, Meinong thinks that assuming is a sui generis intellectual attitude towards objectives, different in kind from judging in large part because of such an absence of belief. The example that perhaps most graphically bears this out is the case of assumption in the case of play-acting, drama and fiction: The intellectual attitude of the child at play is less than judgment, but it is more than representation; which it is to say, it is an attitude of assuming (p. 84).

Similarly “fiction is just assumption” (p. 86). What Meinong means is that in play-acting, drama and fiction a certain kind of pretense takes place. Thus ...the child at play “feigns” properties, situations, and so forth with regard to himself and others (p. 84),

while the dramatist will inevitably be confronted with the task of “placing” himself in not just one but, by turns, almost all the roles of the drama (p. 86).

For Meinong, then, assumption is frequently pretense. We often pretend that things are thus-and-so, without ever believing, even provisionally, that things are really this way. Now return to the problem of how to characterise intending when intending is directed at objects about whose ontic status the intender is either agnostic or sceptical. To give such an object some thought, says Meinong, one needs to assume that it has the appropriate form of being (existence or subsistence), not in the sense of provisionally accepting

92

Russell’s Descriptions and Meinong’s Assumptions

that it has being but in the sense of “placing oneself in the situation in which there is such a thing” (p. 175). Only in this way can one go beyond mere representation to a cognitive grasp or apprehension of the object. Meinong puts it like this in a crucial passage: … what if someone ... “intends” an objectum and thinks about it without therewith giving any consideration at all to the question of its existence or non-existence ... [A]ssumptions play a quite essential role in the apprehending of objecta in such cases. This is also in thorough accord with the testimony of direct observation. It tallies with our clear-cut experiences in many cases, that in order to give a thing some thought a person “places himself in the situation in which there is such a thing” (p. 175).

Meinong is advancing something like the following picture. Suppose a person is attempting to give the golden mountain some thought, but is not sure of its existence. Such a person intends the golden mountain by ‘placing himself in the situation in which there is such a thing’ (note how Meinong uses the very form of words he earlier used to characterise the imaginative activity of the dramatist), an act that in turn allows the person to acknowledge that ‘The mountain (the one I am now attending to) is golden’ from inside the scope of this assumption12. This shows that he successfully intends “the golden mountain”, recognising it as an object that is at once a mountain and golden: an object with a determinate, decomposable Sosein. (Meinong insists that simply contemplating the idea of the golden mountain would not get one this far.) Once we have grasped the object by engaging in such an assumption and thereby recognising the object’s make-up, we can move from assumption to judgement. In particular, we can then safely judge that the golden mountain is golden and that the golden mountain

12 Strictly speaking, Meinong appears to think that we intend descriptively pre-

sented objects like the golden mountain in terms of assumptions of so-being (for example, the assumption that the mountain is golden; p. 197). It is difficult, however, to make sense of such assumptions unless we at least take them to rest on assumptions of being. (Our assumption that the mountain is golden presumably rests on placing ourselves in a situation in which there is a golden mountain to which we are attending, thereby warranting the claim that this contextually determinate mountain is golden.) In other places, Meinong seems simply to identify the relevant assumption of being with assumptions of so-being.

Frederick Kroon

93

is a mountain, with the attributes “apply[ing] to the object analytically in the Kantian sense” (p. 197). Meinong didn’t just think about what grounds property ascriptions involving non-existent objects. He also thought about what grounds denials of existence involving non-existent objects, that is, negative existentials. But before we turn to this part of the story, let us see what Russell would have made of all this. 4. How Russell might have reacted Russell knew the first edition of On Assumptions well, and wrote a fair and laudatory review of the work (Russell 1904). In it, he acknowledges the importance of assuming as an attitude to propositions, and shows that he knows of its connections to pretense. What Russell would not have accepted, however, were the connections that Meinong later made to his theory of Objects. In particular, Russell would not have accepted the use to which Meinong put the Assumption View in the second edition, namely, its use as a tool for discovering properties of non-existent objects such as the golden mountain. For Russell was, of course, adamant that there were no such objects. Meinong, by contrast, thought that you couldn’t understand the sense in which agents have thoughts about the golden mountain, phlogiston, and so on, unless there were such objects. Of course, such an attitude meant that Meinong’s views were bound to remain anathema to Russell, since they conflicted with Russell’s “robust sense of reality” which found no room for such things. But it also suggests a certain way of bringing their views closer together. Meinong, we might say, maintained a kind of objectual conservatism about the kind of pretense that was invoked in the Assumption View: he thought that the relevant form of pretense had to be about, or conserve the identity of, particular objects that were available independently of the pretense. Such an objectual conservatism, however, raises a number of problems. One of these is that it is surely far from clear how the mere assumption that this object exists allows us to uncover its real properties: the properties it has outside of the scope of the assumption. The other, and more basic, problem is that the view seems in any case unnecessarily extravagant. We began the characterisation of Meinong’s Assumption View by saying that it tried to answer the question of how it was possible to know, of some object O whose

94

Russell’s Descriptions and Meinong’s Assumptions

existence was problematic, that it was F. But someone like Russell might well have asked why objects had to come into it at all. If Meinong’s real concern can be more neutrally described as a concern to discover an appropriate sense of truth in which utterances of sentences like ‘Phlogiston is the substance that causes combustion’, ‘Because it was made of solid gold, explorers were willing to risk death to find the golden mountain’, and ‘Hamlet was a depressed Danish prince’ count as true rather than false, there may be no need to go beyond a more thorough-going appeal to pretense. Perhaps such utterances are only true from the perspective of the pretense that the play Hamlet is recorded fact, that there is a unique mountain made solidly of gold, or that there is something satisfying the conditions of phlogiston theory. There may be no more to our ability to engage in thought and talk about phlogiston, the golden mountain and Hamlet than an ability to imagine that there are such things, and to describe the world from that perspective. (Such an ontologically deflationary account of our talk would also mean we don’t have to worry about how to settle what Hamlet, phlogiston, and the golden mountain are like outside of the pretense – a problem that Meinong faces. These objects are mere pretend-objects, so there is nothing more to settle.) This, in fact, is precisely what we find in certain versions of the pretense theoretic approach to the nature of talk about the non-existent, specifically in the kind of account championed by Ken Walton and his school (Walton 1990). What we may call pure pretense accounts of our talk of fictions, mythological objects, failed posits of science, and so on, try to do without special objects in places where Meinong imports an appeal to such objects. Talk of such objects is understood in terms of the thought that speakers may do as if there are objects of a certain kind, where this is then understood in terms of speakers’ implicit engagement in games of make-believe or pretense in which they make claims about such objects that count as true or false because of the norms governing such games (including claims that specify the intrinsic properties of such objects, such as “the golden mountain is golden”). On such a view, we don’t really intend to refer with such terms as ‘phlogiston’ and ‘Hamlet’, but we merely pretend to refer with these terms — we do as if our words pick our real substances and people that we can talk about, debate about, and even have emotional attitudes towards.

Frederick Kroon

95

Here, then, is a way in which Russell might have been able to meet Meinong halfway. Russell could have focused more on the possibilities of pretense that Meinong found so useful for epistemological-semantic purposes and that Russell might have found useful had they been presented to him in their pure pretense-theoretic form. For had the issue been put to him in that way, it is surely far less clear that Russell would have retained his opposition to Meinong. Had he been asked to imagine or assume that the golden mountain exists, and then say, from inside the scope of the imagining, what was true about it, even Russell would have agreed that “it” was of course golden (simply by applying the theory of descriptions from within the scope of the imagining). Similarly, Russell could have agreed that in that sense it was of course true that Hamlet was a prince of Denmark (indeed, that once upon a time there existed a prince of Denmark called ‘Hamlet’). What Russell would have resisted was the inference (one that Meinong made on more general grounds) to the conclusion that he was thereby imagining something about an unreal golden mountain or an unreal object of fiction called ‘Hamlet’, an object that somehow existed only “in the imagination”. Of course, all this is likely to seem quite anachronistic. And to a considerable degree it is. Meinong’s notion of assumption differs in some important respects from the notion of pretense used in modern pretense theories. What is not anachronistic, however, is the idea that appeals to assumption or pretense centering on “objects” like the golden mountain can be disassociated from the kind of objectual conservativism about pretense that Meinong championed in the 1910 edition of On Assumptions. Indeed, there is no such association in the first (1902) edition, which predates Meinong’s discovery of non-existent objects. That work had championed the view that, when we contemplate an idea to which no genuine object corresponds, what explains our sense that we are thinking about an object is not the actual directedness of the idea on to an object of some kind, but simply the assumption of existence – our ‘placing ourselves in a situation in which there is such a thing’ – that (inevitably) accompanies our thought. It is this version of a pure pretense view that finally underwent drastic transformation after Meinong’s “discovery” of non-existent objects, reappearing as Meinong’s objectually conservative Assumption View. I have suggested that this weakened Assumption View, once (re-)interpreted in terms of the theory of descriptions, might well have held

96

Russell’s Descriptions and Meinong’s Assumptions

attractions for Russell. Still, such a view doesn’t help us at all, it seems, when it comes to negative existential claims, which after all are denials of existence. How can affirmative assumptions of existence possibly have a role to play when denying existence? 5. Meinong, Russell and the role of pretense in negative existentials There is no doubt, however, that Meinong did think assumptions of existence had a role to play in the case of negative existentials. Here is what Meinong said about negative existentials (it occurs in the quotation that introduced us to the Assumption View; I add some adjoining text to make it easier to see what is going on). But what if someone intended something like “phlogiston,” in order to make this judgment about it: that there isn’t anything of that sort? ... [A]ssumptions play a quite essential role in the apprehending of objecta in such cases. This is also in thorough accord with the testimony of direct observation. It tallies with our clear-cut experiences in many cases, that in order to give a thing some thought a person “places himself in the situation in which there is such a thing” (p. 175).

At first glance, what Meinong says here looks barely intelligible: how can one apprehend an object like the golden mountain by “placing [one]self in a situation in which there is such a thing”, then contradict what this implies in the next breath by saying ‘phlogiston doesn’t exist’, thereby enabling one to assert that there is no such a thing after all? But in fact Meinong’s meaning is quite clear: to even apprehend the object the golden mountain in thought, a person has to “place himself in the situation in which there is such a thing”; he then judges that this object doesn’t exist, on the grounds that the properties he recognises the object as having through his apprehending the object this way (that is, its being golden and a mountain) are ones that are not instantiated in the real world. No existent object has such properties, so it doesn’t exist. But what if we have Russell’s “robust sense of reality”, and no inclination to accept a commitment to objects like the non-existent golden mountain? Knowing as he did the first edition of On Assumptions, Russell should have known that first-edition Meinong gave essentially the same account of negative existentials, but without invoking nonexistent objects. In his review of the second edition of On Assumptions, C.D. Broad describes Meinong’s first-edition view as follows:

Frederick Kroon

97

When we make a positive existential judgment we find on introspection the experience of grasping a mediate object, whether the judgment be true or false. Why not suppose then that this experience is always due to the existence of something like a judgment? When our judgment is a false affirmative or a true negative the experience of grasping an object is due to the existence of a positive Annahme [assumption]. Further we must suppose that we only experience an idea as presenting an object when it is followed by an Annahme that the object exists. If this be so it will explain how contradictions like round squares can apparently become mediate objects; for Annahmen are indifferent to contradictions13.

Earlier I posed a hermeneutic puzzle: how are we to understand Russell’s complaint, against the view that there are such things as the round square and the golden mountain, that ‘we cannot first assume that there is a certain object, and then proceed to deny that there is such an object’? My speculative answer is that Russell may have (dimly) remembered Meinong’s first edition claim that a negative existential involves an Annahme, or assumption to the effect that there is a certain object, which the negative existential then proceeds to deny. Russell put this together with what he knew to be Meinong’s commitment to objects like the golden mountain and the round square (Meinong 1904), and found the resulting doctrine incomprehensible. And this doctrine is indeed incomprehensible if one persists in thinking that these objects must exist if they are to be the subject of such assumptions. That was never Meinong’s view, of course. He was adamant that these objects do not exist, and that the assumption that they do was not tantamount to some kind of affirmative judgement but merely a pretense for purposes of giving such objects some thought. What about Meinong’s appeal to assumptions in the first edition of On Assumptions? There are reasons for thinking that Russell should have taken that kind of appeal more seriously in his work on negative existentials. For recall our problematic (2). That sentence should remind us that apparently true negative existentials often seem to carry a more or less explicit commitment to objects that are declared not to 13 Broad (1913), p. 99. See Meinong (1902), §24. This passage identifies the exis-

tence of (a) round square (‘die Existenz eines runden Viereckes’) with the beinground of the square (‘das Rundsein des Viereckes’), confirming the link between assumptions of so-being and assumptions of being that I mentioned in the previous footnote.

98

Russell’s Descriptions and Meinong’s Assumptions

exist while at the same time being described and identified in various ways. For someone like Russell whose “robust sense of reality” makes him repudiate Meinong’s realm of non-existent objects, Meinong’s first-edition way of appealing to assumptions holds out the promise that speakers might simply be engaged in the pretense that there are objects to be described and identified in these various ways, with the negative existential denying that there really are such objects. Unfortunately, it is hard to see how Russell could use Meinong’s first-edition view as it stands. For consider (2) again. If the thought is that the speaker engages in pretense in which there is a golden mountain that she and others can talk and write about, then in saying “the golden mountain doesn’t exist” she has expressed an implicit contradiction within the scope of the pretense — from the perspective of the pretense, there does exist such a mountain. Second-edition Meinong can explain the contradiction away by pointing out that the words ‘doesn’t exist’ are to be understood from outside of the pretense, where they are correctly applied to a certain non-existent object, the golden mountain, that is also identified (again, outside of the pretense) as the elusive mountain written about by Smith. The role of the pretense is to allow the speaker to grasp this object in thought — it serves a psychological-epistemic purpose, not a logical-semantic one. This is not a strategy available to first-edition Meinong. But there is another way in which someone sympathetic to the thought that negative existentials like (2) involve pretense might resolve the contradiction, a way that does not involve a commitment to non-existent objects and so is not an affront to Russell’s “robust sense of reality”. On this alternative way, implicit contradictions are often just the spur to systematic reinterpretation of what is being said, yielding a sense in which negative existentials are not literally true. (I should stress that there is nothing to suggest that first-edition Meinong himself would have accepted such a view. First-edition Meinong took negative existentials to be literally true, and it is partly his appreciation of the problems facing this view that explains his conversion to the 1904 Theory of Objects.) Consider an example like the following14. If you say, aiming to challenge your guide-book’s description of your hotel, 14 What follows adapts an argument in Kroon (2004) concerning negative exis-

tentials involving names to the case of descriptive negative existentials.

Frederick Kroon

99

(3) The ‘splendid and immaculate’ hotel you booked me into was neither immaculate nor splendid we all know what you mean. What you assert is something like: ‘When I said “splendid and immaculate” I was deliberately going along with the description given in the guide-book. Outside of the pretense that the hotel you booked me into is splendid and immaculate, it is not the case that it is either splendid or immaculate.’ Of course, your words don’t literally mean this, but you nonetheless manage to assert something like this by exploiting a certain interpretative tension that faces your audience: after all, you are unlikely to want to assert a contradiction. (Note that a familiar way of mitigating the appearance of contradiction in such a sentence is by adding the qualifier ‘really’: ‘[it] was not really immaculate or splendid’.) Here the pretense you engage in is set against your backgroundbelief that a certain modified definite description (the description minus what is denied) remains in force. Your communicative aim as far as the real world is concerned is to assert that the hotel you were booked into is neither splendid nor immaculate. Pretense is important because of subsidiary aims: the way you can display irony, for example, by deliberately going along with the faulty description used by the guide-book. It is tempting to offer another account of such locutions, one that regards the description ‘the “splendid and immaculate” hotel you booked me into’ as elliptical for something like ‘the hotel that you booked me into that was described to me as splendid and immaculate’. There are a number of reasons for resisting such an account. In the first place, there seems to be no compelling syntactic evidence for such a reading (indeed, speakers may not even know how they initially learned the faulty description). Secondly, it is easy to construct examples where the sense that the speaker is deliberately going along with someone’s else’s faulty descriptions is rather more obvious, making it easier to see that pretense is involved. Thus consider the following variant of (3): ‘The ‘splendid and immaculate’ hotel you booked me into – you know, the one where maids vacuum the floors every hour, the bar-fridge is checked every minute, and the guests are mostly Saudi princes – was neither immaculate nor splendid’. A third reason is that there are closely related examples that resist any such reading. Imagine, for example, that your taxi-driver thinks

100

Russell’s Descriptions and Meinong’s Assumptions

she has just spotted a large, grandiose hotel, but that the light is bad: although she is sure she sees the hotel, there is in fact no hotel there (there is only a parking-lot). After peering closely, you finally tell her: (4) You know the large hotel in the distance that you have been pointing out to me? I am pretty sure it is not a hotel at all — sadly (given my accommodation needs), it looks as if it is just a trick of the light. Given the context, it is implausible to suppose that you, the speaker, are declaring that a certain object that has been described as a large hotel is not a hotel. To see the real point of the statement, it is again best to invoke the role of pretense. In uttering (4), you are going along with the beliefs of your taxi-driver, pretending that you are successfully using the description to pick out a certain large hotel. What you assert by stating that the hotel in question is not a hotel is that, outside of the pretense, what is pretended doesn’t hold, in this case that outside of the pretense it is not the case that there is a large hotel in the distance. The last part of your utterance makes it clear that, unlike in the case of (3), you can’t be asserting more than this externally negated claim. Again using pretense, you manage to assert that there is not even a large object there, and that any thought to the contrary is caused by deceptive light conditions. We are now ready to tackle the golden mountain. Consider (5) The golden mountain doesn’t exist Russell should say that in uttering a sentence like (5) a speaker is likely to be engaging in a form of existential pretense to the effect that there is a (unique) mountain that is golden and, of course, exists (recall that for Russell everything exists). This pretense may well be fairly elaborate, and may even include the fact that this pretended golden mountain can be identified in a variety of different ways (for example, as the mountain written about by Smith). In the next breath, and in the scope of the pretense, this is the object that is then said not to exist: an implicit contradiction15. By analogy with the other cases, the very least 15 As in the case of (3), the appearance of contradiction can be mitigated by the inclu-

sion of the qualifier ‘really (‘The golden mountain doesn’t really exist’). I take ‘really’ to have a purely pragmatic role in such cases, serving to render salient what is really

Frederick Kroon

101

that the speaker succeeds in conveying with this contradiction is the externally negated claim that what is pretended doesn’t hold outside of the pretense; in this case that (5a) Outside of the pretense, it is not the case that there is a unique golden mountain and it exists. It is easy to see that the speaker can’t be reasonably interpreted as asserting anything stronger. In particular, since for Russell everything exists (presumably as a matter of a priori fact), the speaker can’t be reasonably interpreted as asserting the stronger internally negated claim that, outside of the pretense, there is a unique golden mountain and it doesn’t exist. (Unlike in the case of (4), we did not need context to rule out this reading; the logic of ‘exists’ alone suffices.) The only reasonable interpretation of (5), therefore, is the externally negated claim (5a). But since everything exists, we can reduce (5a) further by eliding the word ‘exists’, which finally yields: (5b) Outside of the pretense, it is not the case that there is a unique golden mountain, And this, modulo the appeal to pretense, is exactly what Russell’s own theory predicts. 6. Conclusion Whether or not I am right in my hunch that Russell’s rejection of the role of assumptions in a theory of negative existentials had its origins in his dimly remembering things that Meinong once wrote, the more important claim I have tried to advance in this paper is that acknowledging a role for something like Meinongian assumptions would confer significant advantages on Russell’s own theory. It would allow him to capture the way in which many apparently true negative existentials both deny existence and make claims about the “entities” whose existence is being denied, without his needing to yield on his “robust sense of reality”.

asserted, that is, what is asserted apart from the pretense. The presence of a contradiction is usually enough to render this salient, of course, which is why the word is so readily omitted. (Compare this to Evans 1982, chapter 10, which assigns a pivotal but problematic semantic role to ‘really’.)

102

Russell’s Descriptions and Meinong’s Assumptions

Indeed, something stronger can be said. In some ways the suggested pretense view of negative existentials comports rather better with the central core of Russell’s theory of descriptions than his own formulation. There always seemed to be something artificial about the way in which the part of the theory dealing with existential statements had to be specified in a separate clause. It turns out that if Russell accepts something like the suggested pretense view he does not need to invoke separate clauses. The first, predicative, clause can do the entire job. With ‘exists’ assigned a role as a genuine first-level predicate, there will even be predictable scope-ambiguities. In many cases, no doubt, assigning negation wide scope is the right way to understand negative existentials. But not always. Often enough, as in the case of (2), a narrow-scope analysis is more appropriate. Earlier in the paper, such a narrow-scope analysis struck us as bizarre because it turned negative existentials into contradictions. In the final part of the paper, I have tried to show that contradiction of this kind can be viewed as playing a benign and important role in Russell’s theory. Viewed from the perspective of a pretense stance, contradiction of this kind can be resolved in an entirely natural way, yielding exactly the readings Russell wants, and in a manner that doesn’t offend his “robust sense of reality”. References Broad, C. D., (1913), Critical Notice of A. Meinong, Über Annahmen (Leipzig, 1910), Mind, 22, pp. 90-102. Evans, G., (1982), The Varieties of Reference, Oxford, Clarendon Press. Frege, G., (1979), ‘Dialogue with Pünjer on Existence’, In Posthumous Writings, ed. Hans Hermes, Friedrich Kambartel, and Friedrich Kaulbach, Oxford, Basil Blackwell, pp. 53–67. Kripke, S., (1973), ‘Reference and Existence’, John Locke Lectures (unpublished). Oxford, Oxford University Press. Kroon, F. W., (1992), ‘Was Meinong only Pretending?’ Philosophy and Phenomenological Research, 52, pp. 499-527. –– (2004), ‘Descriptivism, Pretense, and the Frege-Russell Problems’, Philosophical Review, 113, pp. 1-30. Meinong, A., (1902), Über Annahmen (first edition), Leipzig, Barth. –– (1904), ‘The Theory of Objects’ In Realism and the Background of Phenomenology (ed.) R. Chisholm, pp. 76-117. Glencoe, Illinois, The Free Press, 1960. Translation of ‘Über Gegenstandstheorie’, In Untersuchungen zur Gegenstandstheorie und Psychologie, (ed.) Meinong. Leipzig, Barth, 1904.

Frederick Kroon ––

103

(1910), On Assumptions, Translated and edited with an Introduction by James Heanue, Berkeley, University of California Press, 1983. Translation of Über Annahmen, 2nd edition. Leipzig, Barth, 1910. Neale, S., (1990), Descriptions, Cambridge, Mass., MIT Press. Ostertag, G., (1998), Definite Descriptions, Cambridge, Mass., MIT Press. Parsons, T., (1980), Nonexistent Objects, New Haven, Yale University Press. Russell, B., (1904), ‘Meinong’s Theory of Complexes and Assumptions’, Mind, 13, pp. 204-19, 336-54, 509-24. –– (1905), ‘On Denoting’, Mind 14, pp. 479-493. –– (1919), Introduction to Mathematical Philosophy, London, George Allen & Unwin Ltd. –– (1956), ‘The Philosophy of Logical Atomism’, In B. Russell, Logic and Knowledge (R. C. Marsh, ed.), London, George Allen & Unwin, pp. 175281. –– (1959), My Philosophical Development, London, George Allen & Unwin. van Inwagen, P., (1977), ‘Creatures of Fiction’, American Philosophical Quarterly, 14, pp. 299-308. Whitehead, A. N. and B. Russell, (1962), Principia Mathematica to *56, Cambridge, Cambridge University Press. Walton, K., (1990), Mimesis as Make-Believe, Cambridge, Mass., Harvard University Press. Zalta, E., (1988), Intensional Logic and the Metaphysics of Intentionality, Cambridge, Mass., MIT.

McGinn on Existence Peter van Inwagen I I ask you to consider two oppositions: being and non-being; existence and non-existence. Before I read the chapter on existence in Colin McGinn’s book Logical Properties1, I should have said that there were five philosophical theories about the relation between these two oppositions. More exactly, I should have said that there were five such theories that were both germane to the issues addressed in McGinn’s theory and could be formulated in terms that analytical philosophers could make some sort of sense of. I have left out of the count those theories—the theories of Aristotle and Ryle, for example—according to which ‘is’ and ‘exists’ mean different things when they are applied to objects in different metaphysical or logical categories. I have left them out because the question whether ‘is’ and ‘exist’ are univocal is not very closely connected with the problems and questions with which McGinn is primarily concerned, although some of his incidental remarks show that he regards these words as univocal. I have also left out of my count certain theories of being and existence that can’t be translated into language acceptable to philosophers of our sort. In my view, various ancient and medieval theories have this feature; and it is certainly a feature of the theories of Heidegger and Sartre. I should have said that there were five theories about the two oppositions, but now I must say that there are six. I’ll begin by laying out briefly each of the five “pre-McGinn” theories—with the warning that my statements of them are far from complete. These statements concentrate on a few of the most important ideas of each theory and leave out many theses and qualifications that their proponents would regard as essential. I should say, too, that the count “five” is rather arbitrary: some of the theories overlap others to a significant extent, and it is possible to doubt whether the first two of the five statements I am about to 1 McGinn 2000.

106

McGinn on Existence

give express theories that differ in any philosophically important way. When I’ve finished my brief tour of the theories with which McGinn’s theory is in competition, I’ll go on to discuss his theory. (1) There is the theory of Quine, according to which the two oppositions are not two but one. Existence and being are the same. Existence or being is what is expressed by phrases like ‘there is’, ‘there are’, and ‘something is’. And, similarly, non-existence or nonbeing is what is expressed by phrases like ‘there is no’, ‘there are no’, and ‘nothing is’. Thus, ‘Universals exist’ means neither more nor less than ‘There are universals’, and the same goes for the pairs ‘Carnivorous cows do not exist’/‘Nothing is both carnivorous and a cow’ and ‘The planet Venus exists’/‘Something is the planet Venus’. This outline constitutes the essence of Quine’s philosophy of being and existence. Some have mistakenly supposed that the backwards-E quantifier plays a central role in his ontological views. But this symbol is, as Quine sees matters, only a formal replacement for the ordinary phrases ‘there is’, ‘there are’ and ‘something is’; its only superiority to ‘there is’ et al. is technical rather than philosophical: it lies in the fact ∃’ has been so contrived that this symbol explicitly that the syntax of ‘∃ interacts with variables—with third-person-singular pronouns—in a way that makes unambiguous cross-reference possible in structurally complex sentences. (2) There is the older theory of Kant and Frege and Russell, a theory that also treats the two oppositions as one. According to this theory, statements that appear to ascribe being or existence either to a particular object—thing, entity, item—or to objects of a kind (the kind being specified in the statement) are in reality statements that ascribe some relational feature to an abstract object—a concept, it may be, or a propositional function, or a property. (And similarly for statements that appear to deny existence or being to some object or to the members of some kind.) Thus, ‘Mermaids exist’ and ‘There are mermaids’ mean the same thing, and what they mean is something along the lines of ‘Something falls under the concept “mermaid”’ or ‘The propositional function “x is a mermaid” has instances’ or ‘Mermaidenhood is exemplified’.

Peter van Inwagen

107

No doubt theory (1) and theory (2) are at least very similar. Obviously, they are in some sense different theories: there is obviously some difference between the predicate ‘falls under the concept “mermaid”’ and the predicate ‘is a mermaid’. McGinn calls (2) the “orthodox” theory of existence, and says, “Nowadays, this Russellian position [by which he means, more or less, theory (2)] is routinely expressed by saying that existence is what is expressed by the existential quantifier,” and in saying this he comes close to identifying the two theories. Other philosophers maintain that (1) and (2) are sufficiently different to force their proponents to adopt importantly different analyses of “singular negative existentials”—sentences like ‘The planet Vulcan does not exist’—, a defensible but debatable thesis. And there is this point to consider: theory (2) is at least prima facie inconsistent with nominalism, and theory (1) is not. Finally, the two theories might seem to suggest, or even to imply, different answers to the vexed question whether existence is a property. Theory (2) was believed by its inventors to demonstrate that the idea that existence was a property was fallacious. But the Quinean who did not share the Master’s distaste for properties, properties of any description, could very well say something like this: Well, you can say that existence is a property if you want to. No harm in it. If existence is a property, it’s the property something has just in the case that it exists, just in the case that there is such a thing as it. That is, as wisdom is the property that corresponds to the open sentence ‘x is wise’, existence is the property that corresponds to the open sentence ‘x exists’, and that open sentence is logically equivalent to this one: ‘there is something that is x’. If you want to talk that way, however, you should keep it firmly in mind that existence, so conceived, is a wholly uninteresting property. Here’s why: ‘For every x, there is something that is x’ is a theorem of first-order logic with identity. Some will think that the theses ‘It is logically fallacious to think that existence is a property of any sort’ and ‘Existence can be thought of as a property, but only as a property that—for trivial, logical reasons— belongs to everything’ are importantly different theses. And some won’t. I am not going to say anything more about the question whether there is any significant difference between our first two theories. But

108

McGinn on Existence

our third theory is certainly different—vastly different—from both of them: (3) According to Meinong’s infamous theory, there are two modes of being, existence and subsistence. Concrete objects that have being exist, and abstract objects that have being subsist. Concrete objects that do not exist have no sort of being whatever, and abstract objects that do not subsist have no sort of being whatever. The Cheshire Cat does not exist, and, cats being concrete objects, there is no Cheshire Cat. The operation “division by 0” does not subsist (unlike the operation “division by 2,” which does subsist), and, therefore, although there is such a thing as division by 2, there is no such thing as division by 0. Nevertheless, the phrases ‘the Cheshire Cat’ and ‘division by 0’ have referents, and we can think about these referents and say true things about them, things like “The Cheshire Cat has a tail” and “Division by 0 plays a hidden role in several well-known fallacious proofs.” And these true things are true for this straightforward reason: each ascribes to an object a property that object has. The Cheshire cat counts having a tail among its properties and “playing a hidden role in several wellknown fallacious proofs” is one of the properties of division by 02. I will not discuss Meinong’s theory beyond the present remark. I confess I find it bizarre. For one thing, I don’t like the idea of “two modes of being.” But Meinong’s theory has a rather more important defect than its incorporation of the idea of modes of being, and that is that it’s self-contradictory—obviously self-contradictory. Here is one way of bringing out the contradiction in the theory: Meinongianism entails that there are things that participate in neither mode of being, things that have no being of any sort; but if there are such things, they obviously have being. For a thing to have being is for there to be such a thing as it; what else could being be? Now this defect in Meinong’s theory—its being obviously self-contradictory—is avoided by certain recent theories whose proponents describe themselves as Meinongians,

2 A note on terminology: in this exposition of Meinong, I have, anachronistically,

used the terms ‘concrete’ and ‘abstract’. This is an anachronism I am willing to defend.

Peter van Inwagen

109

philosophers such as Terence Parsons and Richard Routley, among others. I call these people neo-Meinongians, since, although their theories incorporate many Meinongian elements, they reject a component of Meinong’s theory of objects that I consider essential to it, the doctrine of Aussersein, a doctrine an immediate consequence of which is the self-contradiction that I have just called your attention to: that there are things of which it is true that there are no such things. NeoMeinongianism is the fourth theory in my list. (4) For the neo-Meinongians, as for Meinong, the two oppositions are indeed two. Unlike Meinong, however, the Neos happily apply the term ‘exists’ to abstract things, and they (fortunately) do not maintain that there are things that fall outside the realm of being. The essence of their position is that the things that exist are only some of the things that are, the things that are being simply “the things” full stop; everything that is is everything tout court. For the neo-Meinongians, there is such a thing as the hippo in this room that we can all see. If you look about the room you won’t see it, true, but that’s because the hippo in this room that we can all see doesn’t exist, and you can’t see non-existent things—not, at least, if you’re existent yourself. I think it’s fair to say that most of the technical work that has gone into neoMeinongianism has to do with blocking the deduction of contradictions from sets of statements like the one I have just presented. (5) There is the view of the possibilists. Here I’m not going to name any names; many philosophers hold, or have held, this view, however3. Possibilism has its roots in philosophical reflection on the standard Kripke-style semantics for quantified modal logic. Possibilism divides things, divides the things that are, into two exclusive classes: the things that actually exist and the things that don’t actually exist—that is, the things that might exist but don’t. The possibilists’ use of the word ‘actual’ to mark this distinction can be traced to the occurrence of this word in the phrase ‘the actual world’: actually existent things are the 3 I should mention that David Lewis was a Quinean, not a possibilist, despite the

fact that some people take him to have been not only a possibilist but the possibilist. Lewis was, in fact, the author of one of the most important polemics against neo-Meinongianism, the paper “Noneism or Allism?” (Lewis 1990).

110

McGinn on Existence

things that exist in the actual world and merely possible things (things that might exist but don’t) are the things that exist only in non-actual or merely possible worlds. But it is clear that, whatever may have been the possibilists’ reasons for using the word ‘actually’ in this context, the word is redundant, for things that don’t actually exist are just things that don’t exist, full stop—just as things that aren’t actually red are things that aren’t red, full stop. This point, to which we shall return, may be put as follows: it’s an easy logical step from ‘x might exist & x does not exist’ to ‘x does not exist’. There are important points of disagreement between the neo-Meinongians and the possibilists. Perhaps the most important is that the neo-Meinongians think that the realm of the non-existent contains impossible things (for example, things that, for some property, have both that property and its negation) and the possibilists don’t. But neo-Meinongians and possibilists agree on this point at least: that there are things that do not exist. II In the remainder of this paper, I’m going to use the term ‘neoMeinongianism’ to mean simply the thesis that there are things that do not exist. That is, I’m not going to call a philosopher a neo-Meinongian only if that philosopher accepts all the theses held in common by paradigmatic neo-Meinongians like Terence Parsons and Richard Routley. In this loose sense, the possibilists are neo-Meinongians, as is Colin McGinn. McGinn’s theory of existence and being, his treatment of the two oppositions I began by asking you to consider, is importantly different from that of any other philosopher who has held that there are things that do not exist4. That is why I said that there are now six theories about the two oppositions. The following three theses are, I believe, unique to McGinn’s theory. (1) Non-existent things are, one and all, man-made—or let us, to avoid accusations of both human chauvinism and male chauvinism, say mind-made.

4 The theory is presented in Chapter 2 (“Existence,” pp. 15-51) of McGinn 2000.

Peter van Inwagen

111

(2) Non-existent things are, one and all, necessarily or essentially non-existent, even those, and there are such, whose properties are logically consistent. (3) There are things that exist, but which nevertheless do not actually exist. I’ll say something about each of them. Thesis 1 Creatures with minds are capable of telling stories about things that do not exist, stories of various genres: myth, legend, fable, parable, fiction in the modern, literary sense—and that very important genre, the outright lie. When people tell such stories, they create (that is, bring into being) things that lack the property of existence. For existence is a property. But it not a property like self-identity that necessarily belongs to everything: existence is a property that belongs to some things and not to others. Conan Doyle, for example, brought Sherlock Holmes into being by performing certain mental acts; that he also put pen to paper and that, as a result, other people came to know about Holmes was not logically essential to this act of creation. And Holmes has many properties existent human beings have, just the properties of this sort that everyone supposes him to have: he is a detective and he does play the violin. Paganini and Holmes are, in exactly the same sense, members of the extension of the property plays the violin. But no existent person has the combination of properties that Holmes has; he is therefore, non-existent. In fact, his non-existence consists simply in this: that he was brought into being by a mental act. That is what it is to be nonexistent: to have been brought into being by a mental act. More cautiously, we might say that non-existent things are brought into being by a mental act of the sort by which Conan Doyle brought Holmes into being. The theists among us may want to say that one being, God, brings things into existence by mental acts. If that is true, however, those mental acts must be of a very different character from those by which human beings bring fictional characters into being. No doubt God, if there is a God, is as capable as we are of performing mental acts of the Conan Doyle sort—He can tell parables, for example—

112

McGinn on Existence

and, if He does, He thereby brings certain things into being without bringing those things into existence. Telling stories about what does not exist is not the only way for creatures with minds to bring non-existent things into being. Creatures with minds can make mistakes about what exists, and making mistakes about what exists, like telling stories about what does not exist, can bring non-existent things into being. For example: certain events in the mental lives of certain astronomers, events that constituted their coming to have the mistaken belief that there was (that is, that there existed) a planet inside the orbit of Mercury, caused the non-existent planet Vulcan to be. Thesis 2 Holmes and Vulcan, then, do not exist. But might they have existed? Holmes’s properties are, I believe, logically inconsistent, but that is due only to Conan Doyle’s carelessness. He might not have been careless about whether the set of properties he conferred on Holmes was logically consistent. Let us pretend, in order to avoid introducing an unnecessary complication into our example, that the properties he gave to his creation were logically consistent—and let us suppose that the astronomers who brought Vulcan into being did not inadvertently confer an inconsistent set of properties on Vulcan. No, says McGinn, Holmes and Vulcan couldn’t possibly have existed. And for just the reasons Kripke has called our attention to. In other possible worlds, there are existent objects (existent in those worlds at least) that have all the properties Holmes has. More exactly, they have all those properties of Holmes that an existent human being might have: they won’t, of course, have such properties as being non-existent or having been brought into being by Conan Doyle; and similarly for Vulcan. But, as Kripke has pointed out, there are many such objects. There are in other possible worlds vast numbers of existent objects that have just the sets of properties that Holmes and Vulcan have (given the appropriate restriction on properties). If Holmes or Vulcan might have existed, if they exist in other possible worlds, they must be two among those vastly many objects. But which two would they be? An object, Holmes or Vulcan or any object whatever, is, after all, one object, not two objects or many objects, and all those other-worldly objects are

Peter van Inwagen

113

equally good candidates for one or the other of the offices “being Holmes” and “being Vulcan.” Therefore, Holmes and Vulcan can’t be any of them, and therefore neither Holmes nor Vulcan exists in any possible world. It follows that it is false of each of them that it might have existed. Each is necessarily or essentially non-existent. If something is, like Holmes and Vulcan, brought into being by certain mental acts, nothing could possibly bring it into existence. Another argument for this conclusion would be the following: if something is brought into being by mental activity, that is a part of its essence (a version of Kripke’s essentiality of origins thesis); it is brought into being by mental activity in every world in which it comes to be, and the property being brought into being by mental activity is just the property non-existence: Holmes and Vulcan, by the essentiality of origins thesis, are non-existent in every world in which there are such objects as they. And here’s still another argument for this conclusion: if those properties Holmes has that an existent human being might have are consistent, then the whole set of his properties is consistent. He is in that sense “possible.” But some of his properties—the most obvious example is non-existence—are inconsistent with existence, and these properties are essential to him. That is why, despite his having a consistent set of properties, he can’t possibly exist. Thesis 3 But what about things that might exist but don’t? What about possibilia? What about the younger sister McGinn might have had or the possible fat man in the doorway? The possible sister? The possible fat man? Remember the multitude of descriptively identical Holmes/Vulcan-candidates—possibilia all—that figured in our discussion of the previous thesis. If there are any younger sisters McGinn might have had or any possible fat men in the doorway, mustn’t there be lots of such sisters and such men? Well, that’s a good question, but I won’t try to answer it. In the sequel, I’ll assume that there’s at most one possible sister and at most one possible fat man in the doorway—or assume at any rate that the arguments we shall be looking at do not depend in any essential way on how one counts possibilia. McGinn does not deny that there are mere possibilia. In his view, if I understand him, there being such a thing as the younger sister he might have had is guaranteed by the fact that it is not impossible that

114

McGinn on Existence

he should have had a younger sister. A similar modal fact ensures that there is such a thing as the merely possible fat man in the doorway. He recognizes, however, that the thesis that there are mere possibilia is in prima facie conflict with his analysis of non-existence, for it would have been possible for him to have a younger sister or for there to be a fat man in the doorway, no matter what mental acts anyone had performed: the modal status of a proposition is a mind-independent thing. But the sister and the fat man do not exist. Thus, there are non-existent things, mere possibilia, whose being does not depend on anyone’s mental acts. McGinn avoids the threatened inconsistency by grasping the nettle with his fist: the sister and the fat man do exist. What they don’t do, he says, is actually exist. They are merely possible, yes. They are nonactual, yes. But they do exist. McGinn, in fact, comes close to saying that even impossible objects, round squares and such, exist, although he stops short of affirming this thesis without reservation. It should be noted in this connection that although Sherlock Holmes and a round square can both be described as impossible objects, they are not impossible objects in the same sense. If there are indeed round squares, they are existent objects that—because of their inconsistent properties—couldn’t possibly be actually existent. And Holmes is a non-existent object that—despite the consistency of its properties— couldn’t possibly be existent. Now if McGinn is right about possibilia, if it is indeed true that there are possible things that exist but do not actually exist, there must be a mistake in an argument that occurred in my presentation of possibilism, the argument whose conclusion was that if something doesn’t actually exist, it follows that it doesn’t exist, full stop. Here’s the argument again: ... things that don’t actually exist are just things that don’t exist, full stop—just as things that aren’t actually red are things that aren’t red, full stop. This point, ... may be put as follows: it’s an easy logical step from ‘x might exist & x does not exist’ to ‘x does not exist’. Perhaps I was wrong to suppose that there is just one “point” here. Perhaps there are two. If there are two, the first of them might be put more fully this way:

Peter van Inwagen

115

All uses of ‘actual’ can be explained or paraphrased using only the adverb ‘actually’. And, with the exception of certain of its occurrences in modal contexts, an exception that need not detain us, ‘actually’ makes no semantical contribution to the sentences in which it occurs. Its function belongs to pragmatics, not semantics. Outside modal contexts, ‘actually’ behaves rather like ‘it is true that’: the result of prefixing ‘actually’ to a sentence is a sentence logically equivalent to the original. And if ‘actually’ is inserted into a sentence, if it is placed before the main verb of the sentence, it makes no contribution to the truth-conditions of the sentence in that position, either. (Using ‘actually’ in that way is analogous to using ‘now’ or ‘at present’ to modify a present-tense verb.) ‘Aristotle actually wrote the Nicomachean Ethics/actually died in 322 B.C./actually existed’ is equivalent to ‘Aristotle wrote the Nicomachean Ethics/died in 322 B.C./existed’. A thing that doesn’t actually exist, therefore, is simply a thing that doesn’t exist. If there were indeed two points in the original statement of the argument, here’s the second, the one in the parenthesis: The open sentences ‘x does not actually exist’, ‘x is non-actual’, ‘x is a merely possible object’, and ‘x is a mere possibile’ are logically equivalent to one another and to ‘x might exist & x does not exist’. Each of the sentences ‘x does not actually exist’, ‘x is non-actual’, ‘x is a merely possible object’, and ‘x is a mere possibile’, therefore, entails the sentence ‘x does not exist’. McGinn will obviously want to reject the salient premise of each of these two arguments: namely (i) that (outside modal contexts) ‘actually’ makes no semantical contribution to the sentences in which it occurs; and (ii) that ‘x does not actually exist’ and the other open sentences mentioned in the argument that contain possibilist terms of art are all equivalent to ‘x might exist & x does not exist’. Though McGinn will want to reject these premises, the philosophers I have called possibilists won’t: they’ll be happy to identify non-actual objects with objects that might exist but don’t. If I am right in supposing that McGinn does reject these two premises, I can only say that I don’t know what he means by the words and phrases I have called “possibilist terms of art”: ‘actually’, ‘actual’, ‘non-actual’, ‘merely possible’ and so on. I have no objection to his

116

McGinn on Existence

meaning by these words and phrases anything he likes. I have no proprietary interest in the definitions that occur in my argument. But I should like to see his definitions. In the absence of explicit definitions of ‘actually’ et al., I don’t really understand the third thesis. I don’t, I say, really understand it. Nevertheless, I do have a kind of formal understanding of it, an understanding of how the concepts it employs are formally related to one another, and I think my partial, formal grasp of the thesis is sufficient to enable me to see that McGinn has made, on one special point, a minor mistake in applying his own theory, a mistake that leads him to see a difficulty for that theory where there is no difficulty. Toward the end of the chapter, he affirms the following thesis: Existent things, or at least many of them, things like Venus (the planet, not the goddess) and Clinton (the former president, not the goddess) have the property existence only contingently, in contrast to non-existent things like Holmes, which, one and all, have the property non-existence essentially. McGinn concludes that this fact shows that existence and its negation are related in a way different from the way in which paradigmatic properties and their negations are related. The difference is this: The following generalization holds for typical properties: if they can be had contingently, their negations can also be had contingently. I myself would say that if it is indeed true that McGinn’s theory implies both that existence can be had contingently and that its negation must be had essentially, he’s in real trouble, for it’s demonstrable that if a property, any property whatever, can be had contingently its negation can be had contingently. But, fortunately for McGinn’s theory, it doesn’t really imply this. If McGinn’s theory is right, I would argue, in no case does an existent thing have the property existence contingently. It follows from McGinn’s theory that existent things are, one and all, essentially existent—in fact, necessarily existent. Bill Clinton, for example, exists in every possible world. He is not, as one might suppose, non-existent in those possible worlds in which he was never conceived. Given that the accessibility relation is symmetrical and that the actual world is therefore possible relative to those worlds, he exists in those worlds, on McGinn’s theory, as a mere possibile, since, in those worlds, it’s possible for him to be actual. If mere possibilia really exist, as McGinn says they do (and if the accessibility relation is symmetrical), then Clinton exists in every world (or at least in every world accessible from the actual world) in which he

Peter van Inwagen

117

puts in any sort of appearance whatever, in every world in which there is such a thing as Bill Clinton. That is to say, Clinton has the property of existence essentially (even necessarily). What he has contingently or accidentally is the property actual existence or actuality. And, in the worlds in which he does not actually exist, he has the complement or negation of this property (non-actuality) contingently. McGinn’s theory does not, therefore, imply that there are counterexamples to the thesis that if a property can be had contingently its negation can also be had contingently. An incidental remark: if McGinn’s theory is right, the property being necessarily existent is not one of the “perfections” that supposedly belong to the divine nature: the perfection would be being necessarily actually existent. This completes my exposition of McGinn’s theory. Some of my criticisms of the theory were mixed in with my exposition because it seemed best to me to present these criticisms while the points to which they were addressed were still fresh in the reader’s mind. I turn now to a more general critical task. What I will say in the remainder of this paper is directed at the idea of there being things that do not exist, at the thesis I have called neo-Meinongianism. I am, I confess, a follower of Quine in this and in most related matters. My criticism of neo-Meinongianism will be in every respect a Quinean criticism. But it will not, or so I hope, be a typical Quinean criticism of neoMeinongianism. III The typical Quinean critic of neo-Meinongianism opens his animadversions (as Quine might have called them) on neo-Meinongianism with some variant on the following confession: “I don’t see any difference, any difference at all, between ‘Unicorns do not exist’ and ‘There are no unicorns’”. And the neo-Meinongian generally ignores whatever follows this confession and says something along the lines of, “Thank you for sharing. I’m terribly sorry to learn that you’re ontologically challenged. But, whether you see it or not, there is such a distinction. It even exists.” And, if the neo-Meinongian goes in for psychoanalysis, he may add, “You do see the distinction between being and existence on some level, but you refuse to admit to yourself that you

118

McGinn on Existence

see it. You’ve somehow managed to convince yourself, at least when you’re in the philosophy room, that ‘there is’ means ‘there exists’. But the particular quantifier—the currency of the phrase ‘the existential quantifier’ testifies to the success of a century-long anti-Meinongian propaganda campaign initiated by Russell—and ‘there is’ and ‘il y a’ and ‘es gibt’ have a meaning more general than that of ‘there exists’: ‘There exists an F’ entails ‘There is an F’, but not vice versa.” Impasse. I’m going to try to break the impasse. I will make my case against neo-Meinongianism in such a way as not to provide any occasion for yet another unseemly dogfight over the usual bones of contention: the meaning of the backwards-E quantifier and the meaning of the ordinary-language phrase ‘there is’. It is mainly in this respect that the argument I’m going to present differs from the argument of David Lewis’s “Noneism or Allism?” (Lewis 1990). In that paper, Lewis made a certain point about the relation between the neo-Meinongian’s use of ‘exist’ and the backwards-E quantifier. I’m going to make a point that is similar to Lewis’s, but my point will be about the relation between the neo-Meinongian’s use of ‘exist’ and the negation of the universal quantifier. I begin with the idea of unrestricted universal quantification. It is a commonplace of the philosophy of language that when one uses the idiom of universal quantification, one often, one perhaps usually, has some restriction in mind. “We’ve sold everything,” says the sales clerk after a particularly busy day behind the counter, and we who hear this assertion do not protest that the number 510, the Taj Mahal, and the counter—a concrete object right there in the shop—remain unsold. We all, I believe, understand the uses we make of universal quantification in everyday life, and, even if the universal quantifications we assert in everyday discourse are mostly restricted, tacitly or explicitly, it does not require much philosophical instruction for us pass from an understanding of the everyday uses of the universal quantifier to an understanding of the pristine idea of unrestricted universal quantification. It seems to me that this idea, the idea of unrestricted universal quantification, is a pellucid and wholly unambiguous idea. It is the idea of “everything without qualification.” It is the idea we require if we wish to say that everything, absolutely everything, everything without qualification, is self-identical or conforms to the laws of logic or is either abstract or concrete.

Peter van Inwagen

119

I concede parenthetically that it has been denied that there is any such thing as what I am calling unrestricted universal quantification. Dummett has maintained that there cannot be such a thing because quantification presupposes a domain of quantification and there is no universal domain. If you want a long refutation of this position, see Timothy Williamson’s “Everything” (Williamson 2003). Here’s a shorter one: We can say, we just can, that everything, absolutely everything, everything without qualification, is self-identical or conforms to the laws of logic or is either abstract or concrete. But if Dummett’s argument were sound, we should not be able to say these things. Thus I refute Dummett. Now it seems to me that everyone, everyone including Quine and Kripke and Plantinga and Lewis and me (on the one hand) and McGinn and Parsons and Routley and Castañeda (on the other), means the same thing by the phrase ‘unrestricted universal quantification’ and everyone means the same thing by the unrestricted universal quantifier— that is, by the operator ‘Everything, absolutely everything, everything without qualification’, although we ordinary folk (on the one hand) and the neo-Meinongians (on the other) will certainly disagree about which sentences containing unrestricted universal quantifiers are true. When I say that everything exists and the neo-Meinongian denies that everything exists, we’re not talking past each another—not, at any rate, because we mean different things by ‘everything’. It is precisely because the neo-Meinongian knows that I mean just what he does by ‘everything’ that he indignantly rises to dispute my contention that everything exists. And, although the neo-Meinongian may suspect that we Quineans are guilty of tendentious fiddling with the meaning of ‘there is’, I do not think he will be tempted to entertain the corresponding suspicion as regards ‘all’. ∀’ to express absolutely unrestricted univerLet us use the symbol ‘∀ sal quantification (in other words, let us use this symbol in its usual sense). I say this: ∀x ~ x is a unicorn The neo-Meinongian says this: ∀x ~ x is a unicorn. ~∀ What am I to say to this? Let me sneak up on what I want to say. I begin my sneaking by laying out a faux naïf reaction to the assertion

120

McGinn on Existence

that not everything is not a unicorn. (By which I mean: this reaction would be my reaction if I didn’t know from experience what the neoMeinongian would say in reply.) I don’t see how what you say could be true. That not everything is a non-unicorn seems to me to be inconsistent with what I know about the world. If it is indeed false that everything is a non-unicorn, then, if I were made free of all space and all time, I ought to be able to find, encounter, or observe a unicorn. A unicorn, after all, must occupy a certain region of space-time. If anything is a unicorn, then, necessarily, that thing occupies a region of space-time. But, or so I believe, every region of space-time is such that it contains no unicorn: no magic carpet or starship or time machine can take me to a place where there is a unicorn. In other words: everything is a nonunicorn. I have spent enough time arguing with neo-Meinongians to know that few of them will respond to this argument by saying, “Golly, I never thought of that. I stand refuted.” A neo-Meinongian is rather more likely to respond to it by saying something like this: I know of course that the concept of a unicorn is the concept of a thing that has a location in space and time. And I know that neither you nor I can find or observe or discover a unicorn—and that we should be unable to do these things even if we could open a wormhole that would take us to any point in space and time. I am aware ∀x that unicorns are not, so to speak, “findable.” But the truth of ‘~∀ ~ x is a unicorn’ does not entail that unicorns are findable. Not everything is a non-unicorn (provided that this ‘everything’ is indeed an absolutely unrestricted ‘everything’)—and yet unicorns can’t be found. (Not by us, at any rate, not by existent people.) Even if superscience were to provide us with a starship or a time machine or a wormhole-opener, there are lots of places that even those things couldn’t take us to. They couldn’t take us to Middle-Earth, for example, or even to Barchester. In a word, they couldn’t take us to nonexistent places or non-existent times. And unicorns, being non-existent creatures, tend to be in non-existent places at non-existent times, although no doubt some of them are in existent places at existent times. But even if we go to one of the existent places at which there is a unicorn, and go there when the unicorn is there, we shall not see

Peter van Inwagen

121

it, for it is non-existent and of course one can’t see non-existent things. This, I maintain, is almost certainly what the neo-Meinongian will say in response to the naive argument—this or something very like it. If the neo-Meinongian says this, however, I must protest that either he contradicts himself or I do not understand him5. Consider the following sequence of statements. As the above reply to the naive argument shows (not that this needs much showing) the neo-Meinongian accepts the first statement in the sequence. In my view, on my understanding of existence, each statement (after the first) in the sequence is a consequence of—and is in fact equivalent to—the preceding statement in the sequence: Every unicorn is non-existent Every unicorn is such that everything [an unrestricted ‘everything’] is not it. ∀x (x is a unicorn → ∀y ~ y = x) ∀x ~ x is a unicorn. The second sentence follows from the first (in my view) because (in my view), if it means anything, ‘x is non-existent’ means ‘everything is not x’: I don’t see what else it could mean. Or, at any rate, the relation between the two sentences is something very like a meaningequivalence. Consider the two sentences ‘x is a circle’ and ‘x is a figure that, for some point y, comprises those and only those points that are equidistant from y’. Do these two sentences mean the same? Who can say? Perhaps the concept of sameness of meaning is not sufficiently well defined for this question to have an answer. However this may be, when I say that ‘x is non-existent’ means ‘everything is not x’, I mean no more than that the meanings of these two sentences are at least as intimately related as the meanings of the “circle” sentence and the “circle analysis” sentence.

5 If I were advising him on how to conduct his case, I’d recommend that he

respond to this protest as Roderick Chisholm once responded to a similar protest: “I accept the disjunction.”

122

McGinn on Existence

If the phrase ‘is non-existent’ means what I say it means, the neoMeinongian’s allegiance to ‘All unicorns are non-existent’ commits him to the thesis that everything is a non-unicorn. And, as we have seen, he contends that it is false that everything is a non-unicorn. It would therefore seem that—since the neo-Meinongian obviously does not mean to embrace a straightforward formal contradiction—, he must think that ‘is non-existent’ means something different from what I say the phrase means. But what does he think ‘is non-existent’ means? Well, we’d know how to answer that question if we knew how to answer the question, What does he think ‘exists’ means?, so let’s ask it: What does he think ‘exists’ means? I do not know. I say ‘x exists’ ∀y ~ y = x’; the neo-Meinongian denies this. Apparently, he means ‘~∀ takes ‘exists’ to be a primitive, an indefinable term, whereas I think that ‘exists’ can be defined in terms of ‘all’ and ‘not’. And I take definability in terms of ‘all’ and ‘not’ to be important, because the neoMeinongian means exactly what I do by ‘all’ and ‘not’—and thus he understands what I say is what ‘exists’ means, and he is therefore an authority on the question whether what I say ‘exists’ means is the same as what he says is what ‘exists’ means. Since the neo-Meinongian believes that ‘exists’ has a meaning that cannot be explained in terms of unrestricted universal quantification and negation, he therefore believes in two kinds of quantification ∀’ and ‘~∀ ∀~’. (Call where I believe in one. I have two quantifiers: ‘∀ these the unrestricted quantifiers.) Of course, I don’t usually write the second one that way, and I usually call it something that starts with ‘e’ and rhymes with ‘residential’. For subtle dialectical reasons, however, I won’t, for the duration of this argument, write the second unrestricted quantifier the way I usually do, and I won’t call it what I usually call it. The neo-Meinongian has four quantifiers: the two unrestricted quantifiers and two restricted quantifiers: ‘A’ and ‘E’6. The two restricted quantifiers (the “existentially loaded” quantifiers) may be defined as follows:

6 I should perhaps mention that one neo-Meinongian, Tererence Parsons, has told

me that he doesn’t like to be described as someone who “has” two and only two restricted quantifiers. He thinks that it would be as accurate to say that he “has”

Peter van Inwagen

123

Ax Fx =df ∀x (x exists → Fx) ∀x ~(x exists & Fx). Ex Fx =df ~∀ When I say that this is how the existentially loaded quantifiers may be defined, I mean that this is how they may be defined for the benefit of someone who knows what the neo-Meinongian means by ‘exists’. But not for my benefit, for, as I have said, I do not know what the neoMeinongian means by ‘exists’. Although I don’t know what the neo-Meinongian means by ‘exist’, and therefore don’t know what the existentially loaded quantifiers mean, I am at least in a position to dispute, as regards my own case, one accusation which neo-Meinongians frequently direct at Quineans like myself (McGinn is one of the neo-Meinongians who have made this accusation): that the Quineans have somehow convinced themselves that the only quantifiers are the existentially loaded ones. I say that, in my case at least, the charge is wrongly framed. I say that the charge the neo-Meinongians should bring against me is the following one: that I have somehow convinced myself that the only quantifiers are the unrestricted ones; that I have somehow talked myself into accepting the false thesis that the only meaning ‘exists’ can have, and the only meaning our use of the word requires it to have, is “not-allnot.” I don’t agree that the charge is accurate, of course, but it’s the right way for someone with their views to frame the charge they want to bring against me. My own description of our disagreement is the following. The neo-Meinongians and I mean the same thing by the unrestricted quantifiers; I think ‘exists’ means something they understand and mistakenly think is not what ‘exists’ means; they mean nothing by ‘exist’ or at any rate nothing I have been able to understand. There is one thing that it’s important to realize when we consider these two descriptions of the disagreement. You will misunderstand what I have been saying if you take me to have been saying that neoMeinongians (on the one hand) and I (on the other) mean two different quantifiers restricted to existent and non-existent cows as it would be to say that he “has” the existentially loaded quantifiers—that is, quantifiers restricted to bovine and non-bovine existents. But I think that most Neos, including McGinn, will be happy enough to say that they recognize four quantifiers, two of them unrestricted and two of them existentially loaded.

124

McGinn on Existence

things by ‘exist’. The neo-Meinongians and I have different theories about what ‘exist’ means: two different theories about what the English word ‘exists’ means. They understand mine and say it’s false; I don’t understand theirs, or at least I claim not to. The fact that the two theories are theories about the meaning of a word doesn’t mean that our disagreement is about a trivial matter. After all, it’s a philosophically very important word. But the English word means something, and at least one of the parties in this debate has got it wrong. If I have got it wrong, I have a mistaken theory about what ‘exists’ means, and I am just as wrong about what I mean by the word as I am about what anyone else means by the word. By way of comparison consider the following case. At one time, A.J. Ayer held a mistaken theory about the meaning of ‘knows that . . .’, to wit, that this phrase meant ‘believes that . . . and is right in so believing and is justified in so believing’; since Ayer was a competent speaker of English, he was wrong not only about what other people meant by ‘knows that’ but about what he meant by ‘knows that’. Of course, if I have got it right, something like “not-all-not” is what the neo-Meinongians mean by ‘exists’. The Neos, when they use the English word ‘exist’ of course mean by it what it means, and if that happens to be, as I say it is, “not-all-not,” they mean “not-all-not” by ‘exist’—although, according to their mistaken theory about the meaning of ‘exists’ that is not what they mean by it. What might the neo-Meinongian say in response? Whatever other neo-Meinongians may say, I would expect McGinn to say something like the following. Some terms have to go undefined, and it wouldn’t be surprising if a discourse-pervading term like ‘exists’ were one of them. I can’t give a formal definition of ‘exists’—just as you can’t give a formal definition of ‘all’ or of ‘not’. That is to say: I can’t give a formal definition of ‘exists’ that is independent of my theory of existence, a point to which I’ll return. Nevertheless, I can make the meaning of ‘exists’ clear to anyone who hasn’t embraced a false theory about its meaning. (There’s not much point in trying to help people who have embraced a false theory about the meaning of ‘exists’ to get clear about its real meaning: they don’t want help.) All I have to do is to ask that person to consider any of thousands of sentences like following one:

Peter van Inwagen

125

The planet Venus exists and the planet Vulcan doesn’t exist. The unprejudiced will see immediately that when we utter sentences like this one, we are predicating a certain property, existence, of a certain object, the referent of one of the terms we have used, ‘Venus’, and predicating its complement or negation, non-existence, of another object, the referent of the other term we have used, ‘Vulcan’. This exercise enables one to bring the properties existence and non-existence clearly before one’s mind by providing one with a perfectly clear case in which something that exemplifies each of these properties is contrasted with an otherwise similar object that exemplifies the other. Once one has done this, one will be in a position to consider my proposal about what property non-existence is, to wit, the property of having been brought into being by mental activity. So speaks McGinn, at least in my imagination. I don’t think I have much more to say to anyone who finds the imaginary speech convincing. That person and I have reached fundamental, ground-floor philosophical disagreement. It seems as evident to me that the ideas on display in the speech are fundamentally wrong as it does to McGinn that the Quinean identification of ‘x exists’ and ‘something is x’ is fundamentally wrong. I don’t dispute this: that someone who said (imagine that we know of an actual, historical occasion on which someone said these words in appropriate and felicitous circumstances), “Gentlemen, it’s time to be honest with ourselves and admit that we have been chasing a will-o’-the-wisp. Vulcan doesn’t exist,” would, in uttering the sentence ‘Vulcan doesn’t exist’, have said something true. But it doesn’t look to me as if, when the speaker said this, he referred to a planet or to an object of any sort or as if he had said anything of any description about any particular thing. That’s not how the case feels to me. McGinn can ascribe my reaction to this use of ‘Vulcan doesn’t exist’ to prejudice if he likes, but ascribing someone else’s failure to see things the way one sees them to prejudice has all the advantages of— well, you know the phrase. IV What I have just said is not and is not meant to be a refutation of the argument I have ascribed to McGinn. It is no more than a record of one

126

McGinn on Existence

philosopher’s failure to be convinced by that argument. But what I have said bears looking at because it shows that I am committed to a thesis that McGinn is suspicious of on grounds of philosophical methodology, and I will bring this paper to a close with a discussion of this thesis. I am committed to saying that our use of the grammatical predicate ‘doesn’t exist’ is a complicated, messy, gerrymandered affair. I have conceded that sentences like ‘Vulcan doesn’t exist’, sentences formed by prefixing a term (a proper name or definite description: a noun or noun-phrase that represents itself as denoting a particular object) to the predicate ‘doesn’t exist’, can be used to say true things. Let ‘N doesn’t exist’ be any true sentence of this form. (A “true sentence” being a sentence such that, if anyone uttered it in appropriate and felicitous and reasonably normal circumstances, that person would say something true.) For McGinn, the truth of ‘N doesn’t exist’ is a straightforward matter: the name ‘N’ denotes an object, and that object has the property expressed by the predicate ‘doesn’t exist’. The proper semantical analysis of this sentence, according to McGinn, is of exactly the sort that most philosophers would say was right for paradigmatic subject-predicate sentences like ‘The Taj Mahal is white’ and ‘Bill Clinton was once the governor of Arkansas’. But this straightforward semantical analysis is not available to a philosopher like me— that is, to a Quinean who concedes that ‘N doesn’t exist’ is true. It is not available to my philosophical allies and me for two reasons. (1) As my allies and I see matters, the name ‘N’ is without a referent, and the analysis implies that ‘N doesn’t exist’ is true only if ‘N’ has a referent. (2) If the predicate ‘doesn’t exist’ expresses a property, this property is, of course, the complement of the property existence. And, as my allies and I see matters, existence is the property identity-with-something or simply self-identity: a thing that is identical with something is identical with itself, and a thing that is identical with itself is identical with something. But the complement of the property identity-with-something (or self-identity) is the property identity-with-nothing (or non-self-identity or self-non-identity). This property is, of course, an impossible property, and if a property is impossible, nothing has it. The straightforward analysis implies that the sentence ‘N doesn’t

Peter van Inwagen

127

exist’ is true only if a certain object has the property expressed by ‘doesn’t exist’, and therefore implies that ‘N doesn’t exist’ is false if the property expressed by ‘doesn’t exist’ is a property nothing has. What, then, do I say is the right semantical analysis of ‘N doesn’t exist’? Alas, I must say that there is no right analysis—no one right analysis. I must say this because I believe that, as I said a moment ago, our use of the grammatical predicate ‘doesn’t exist’ is a complicated, messy, gerrymandered affair. In my view, there are at least three ways, three different ways, in which ‘doesn’t exist’ functions, and each of them requires a different semantical analysis. Here are three cases in which a speaker uses the predicate ‘doesn’t exist’, and uses it to say something true. (I’ve supplied some context for the first two; the third doesn’t need it: ‘→’ is Lewis’s counterfactual operator.) Only Sherlock Holmes could solve this case, and, unfortunately, Sherlock Holmes doesn’t exist. (Said by a frustrated police detective.) Lord bless you, sir, Sherlock Holmes doesn’t exist and he never did. He’s just a chap in a story made up by someone called Conan Doyle. (Words addressed by a classic, pre-war London bobby to a tourist who is trying to find 221B Baker Street.) Colin McGinn doesn’t exist → Philosophy is impoverished. Each of these uses of ‘doesn’t exist’ is, as I see matters, different from the other two—and in none of the three cases is the predicate used to attribute to something the property of non-existence. I’m not going to try to explain to you why I think ‘doesn’t exist’ has three uses. I’m not asking you to believe that ‘doesn’t exist’ has three uses; I’m simply confessing to you that that is what I believe. I’m confessing that I subscribe to a very complex theory of the way ‘doesn’t exist’ functions in our language, a theory according to which the proper semantical treatment of ‘doesn’t exist’ will be, so to speak, radically non-uniform. I don’t mean to imply that a theory that ascribes such complexity to ‘doesn’t exist’ is a logical consequence of Quineanism. I am one Quinean among many, and other Quineans may favor theories that ascribe a much simpler semantics to ‘doesn’t exist’—for

128

McGinn on Existence

example, the theory that ‘N doesn’t exist’ is true just in the case that ‘N’ doesn’t have a referent. (Good luck to them, but I think that that theory can be shown to be unsatisfactory.) But all Quineans must have a theory about how the predicate ‘doesn’t exist’ works that is at least somewhat more elaborate than McGinn’s. At a minimum, all Quineans who agree with me that ‘Vulcan doesn’t exist’ and other sentences of the same sort are true must dispute McGinn’s contention that sentences like ‘Vulcan doesn’t exist’ can be given a semantical analysis exactly parallel to the analysis that—almost everyone agrees—is appropriate for sentences like ‘Venus doesn’t have inhabitants’ and ‘Winifred doesn’t smoke’. McGinn regards a uniform treatment of subject-predicate sentences as a default position in philosophy: that is, he thinks that any philosopher who treats some subject-predicate sentences as requiring a semantical or philosophical analysis radically different from that of most other subject-predicate sentences is saying something that can be defended only by a very strong argument. I don’t see it that way myself. In fact, it seems to me to be very doubtful whether there can be such a thing as a standard semantical treatment of the subject-predicate sentences of natural language. To see just one example of the kind of thing that is behind my scepticism, consider the sentence ‘The shadow of the elm is moving slowly across the lawn and it will reach the garden path in about an hour’. This sentence is an ordinary enough subject-predicate sentence, and one that might be used in easily imaginable circumstances to say something true. Suppose, then, that someone uttered this sentence and said something true. Does the truth of what that person said imply that (on that occasion) ‘the shadow of the elm’ denoted something and that the thing it denoted then had the property expressed by the open sentence ‘x is moving slowly across the lawn and x will reach the garden path in about an hour’? I certainly hope not, for I can’t bring myself to believe that there really are shadows; and I find it even harder to believe that, if I am wrong and there are shadows, one and the same shadow can be in one place at one time and in another place at another time. Natural language and its grammar are very complicated things and they did not evolve (or they were not conferred on us by God or whatever the right story of their genesis may be) in order that the grammar of natural language might serve as a reliable guide in philosophical

Peter van Inwagen

129

speculation. Maybe natural (or “ordinary”) language is, as Austin said, the first word. It is certainly (as he conceded) not the last word. I don’t believe in shadows because I can’t give a coherent account of what properties shadows would have if there were any of them. And that is what I want to say about non-existent objects, too: the only property I know of that it seems right to me to apply the name ‘non-existence’ to is an impossible property. I therefore regard the grammatical structure of sentences like ‘The shadow is moving across the lawn’ and ‘Vulcan doesn’t exist’ as an untrustworthy guide in ontology. Even if I did believe that a “standard” (that is, a Tarskian) semantical analysis was the default semantical analysis for natural-language subject-predicate sentences, I’d say that the prima facie case for applying that analysis to the true sentences ‘The shadow is moving across the lawn’ and ‘Vulcan doesn’t exist’ was overridden by what I see as the semantical facts: in the one case, the fact that ‘the shadow’ has no referent and, in the other, the fact that ‘Vulcan’ has no referent and the fact that ‘doesn’t exist’ expresses a property nothing has*. References Lewis, D.K., (1990) ‘Noneism or Allism’, Mind, 99, pp. 23-31; reprinted in David Lewis, Papers in Metaphysics and Epistemology, Cambridge, Cambridge University Press, 1999, pp. 152-163. McGinn, C., (2000) Logical Properties: Identity, Existence, Predication, Necessary Truth, Oxford, Clarendon Press. Williamson, T., (2003) ‘Everything’, Philosophical Perspectives, 17, pp. 415465.

* This article is published simultaneously by The Philosophical Quarterly.

The Talk I Was Supposed to Give… Achille C. Varzi … I am not giving it. I am sorry. I have changed my mind and I am giving this talk instead—if everything works out as planned. But you may wonder: What is that it that I am not giving? What am I referring to when I apologize for not giving the talk I was supposed to give? Never mind the content of that talk— I can always give you a summary of what I meant to say, if you are interested. Indeed, I might even decide to give a talk with that very same content on a different occasion. The content, however, is not the talk, and if I gave a talk with that content on a different occasion I would not, on my reckoning, give the talk I was supposed to give. For the talk I was supposed to give is something that was supposed to take place today, and that is not going to happen. Hence the worry: How can we talk about it? How can we talk about events that fail to occur? Let us assume Donald Davidson was right in urging that events form a genuine metaphysical category, on a par with material objects1. Shall we say that a good inventory of the world ought to include “negative” events—failures, omissions, things that don’t happen—along with positive ones? Davidson himself was candid about this: “We often count among the things an agent does things he does not do”2, he wrote. Yet this may offend our ontological scruples. If I am not actually giving that talk, why should we include it in an inventory of the actual world? What would it be like? For instance, what smaller events (such as bodily movements and speech acts) would be part of it, and how would it relate to the other events of my life? Indeed, how would it relate to the other things that I am failing to do? Surely in failing to give that talk I am also failing to please the organizers. One failure or two? And surely there are many more things I could have done at this very same time but I am not actually doing: I am not drinking coffee, I’m not watering your plants, I’m not singing My Way. Would all of these nondoings deserve a place in our inventory of the world? 1 Davidson (1980). 2 Davidson (1985), p. 217.

132

The Talk I Was Supposed to Give...

Some might think so3, but I certainly don’t. In a way, speaking of non-occurring events is like speaking of non-existing objects, and my inclination is not to take these manners of speaking at face value4. We often talk as though there were such things, but deep down we may want our words to be interpreted in such a way as to avoid serious ontological commitment. For example, we may say that reference to Sherlock Holmes or to the winged horse is to be understood within the context of a fictional story, a pretense, a game of make-believe, and we may likewise say that reference to the talk I was supposed to give is to be understood within the context of a suitable fictional scenario. It is true that Sherlock Holmes lived in Baker Street insofar as that is true according to Conan Doyle’s narrative; similarly, it is true that the talk I was supposed to give was about mathematical fictionalism (say) insofar as that is true according to certain counterfactual speculations about how the world might or should have been. Unfortunately, non-occurring events appear to be more resistant to Ockham's razor than the analogy with non-existing objects might suggest. Not only do we often speak as though there were such things. We often speak in such a way as to suggest that reference to or quantification over non-occurrences is to be taken strictly and literally. Typical examples include statements such as ‘Beth saw Al not leave’ and ‘Al often doesn’t go jogging’, whose Davidsonian logical forms appear to require a straight commitment to actions that Al failed to perform. Or consider ‘Al’s failure to turn off the gas caused an explosion’. How can we deny that sometime we are causally responsible, not only for our actions but also for our omissions? And how can we provide a reasonable account of our talk about tryings and intendings without bringing in negative outcomes? I missed the deadline, even though I tried very hard not to. And Al didn’t call Sue, not because he got stuck doing something else, but because he intended not to call her. The aim of this talk (the one that I am giving) is to outline a way of resisting such difficulties on behalf of the view according to which the only events to be seriously countenanced are the positive ones—those 3 See e.g. Vermazen (1985). 4 The “ontological scruples” mentioned above parallel Russell’s (1905) scruples

concerning the reality of Meinongian possibilia, and the worries about ontological proliferation parallel Quine’s (1948) worries about their “identity conditions”.

Achille Varzi

133

that feature in the actual history of the world. This involves articulating two distinct but related ideas, each of which is independently motivated. First, there is the idea that, in many cases, a “negative” event (a non-leaving) just is an ordinary, positive event (a staying); it is a positive event under a negative description. Second, there is the idea that in some cases, when there is no positive event we could be talking about, what we say is strictly and literally false, albeit in a way that admits of true paraphrases. These two ideas are distinct, but they are closely related. I shall look at them in turn. Speaking of Events Concerning the first point, my reasons for putting it in terms of Anscombe’s controversial phrase, ‘under a description’5, stem from a general consideration concerning the semantics of our event talk, which in turn reflects a certain conception of what events must be if we accept a broadly Davidsonian framework. Events, on this view, are particulars6. They are unrepeatable entities located in space and time, and we can describe any such entity in various ways, just as we can describe any particular object in various ways. I say: “Al is the right person for Beth”. You say: “Tell me more about him”, and I elaborate: “He is smart, he likes traveling, and he is fond of jogging”. You say: “Al took a walk last night”. I say: “Tell me more about it”, and you elaborate: “It was a long and leisurely stroll in Central Park, before dinner, with his friend Sue, and Beth got really angry about it”. The pronoun ‘it’, here, just as the pronoun ‘he’ in the previous scenario, is all we need to carry the reference back to the entity we were talking about. Similarly, just as we can designate an object through a number of different descriptions, each of which relies on certain properties that the object possesses, so we can designate an event by means of descriptions that may vary depending on the context and the purpose of our discourse. We can refer to Al by his proper name or we can identify him by saying ‘Beth’s boyfriend’, ‘my smart colleague’, or ‘the president of the jogging club’, and we can likewise identify what he did last night by saying ‘Al’s walk in Central Park’,

5 See Anscombe (1957), esp. § 23. For elaborations and clarifications, see her (1979). 6 See Davidson (1970).

134

The Talk I Was Supposed to Give...

or ‘Al’s leisurely promenade with Sue’, or ‘the cause of Beth’s anger’. These phrases do not pick out different entities; they pick out the same entity under different descriptions. Admittedly, one may have different intuitions in this connection. Since Al could have walked alone, whereas he could not have gone for a promenade with Sue all by himself, some are inclined to say that ‘Al’s walk’ and ‘Al’s promenade with Sue’ must refer to distinct events by virtue of Leibniz’s law: those terms are not substitutable salva veritate7. Likewise, if Beth got angry, not because Al went for a walk, but because he went with Sue, some are inclined to say that ‘the cause of Beth’s anger’ does not refer to Al’s walk but to his promenade with Sue, which should therefore be treated as a distinct event8. I disagree with such intuitions. Surely not every description of what Al did last night may be equally adequate in every context, but that is not to say that he did many things at once: different descriptions may have different senses and may therefore serve different purposes, yet their actual referent may coincide. Thus, although it is true that in a possible world in which Al went for a lonely walk the description ‘Al’s promenade with Sue’ would not apply, it doesn’t follow that such a description fails to pick out Al’s walk in this world. (Compare: although it is true that ‘the number of planets’ could pick out a number other than 9 in different worlds, it doesn’t follow that it fails to pick out the number 9 in this world.) Likewise, although it would be silly to explain why Beth got angry by describing what Al did as a mere walk, it doesn’t follow that the walk is not the cause of Beth’s anger: causal explanations are language-sensitive but causal reports are semantically transparent9. (Compare ‘Al’s walk with Sue caused Beth’s anger’ with ‘My colleague’s walk with Sue caused Beth’s anger’: the former statement may be better suited as an explanatory report, especially if Al is known to be Beth’s boyfriend, but this has no bearing on what ‘Al’ and ‘my colleague’ refer to.) In short, I think we should be careful not to assume that the expressions we use to identify actions and other events tell the whole truth about those actions and events: just as an object’s description can be very partially informative, and even misleading for certain 7 This line of reasoning goes back to Goldman (1971) and Thomson (1971). 8 See e.g. Kim (1976). 9 See Davidson (1967b).

Achille Varzi

135

purposes, so an event’s description can be partially informative and misleading. As Jonathan Bennett has put it, there is no way we can “read off” the nature of an object or the nature of an event from the words we use to refer to them10. Positive versus Negative Descriptions With this picture in place, it is obvious that among the many ways in which we can pick out an event, there are some that may be broadly classified as negative descriptions. Al took a walk in Central Park. Really, he was supposed to go jogging with his friend Tom, and he would have done so had Sue not joined him. That was the plan. So, depending on the context, we might want to refer to Al’s walk, not as a walk but as a non-jogging. It was a walk, so it was not a jogging, and that may very well be the bit of information about Al’s doing that we might want to convey in our description of it. His walk caused Beth’s anger owing to Sue’s presence? Then we’d better refer to it explicitly as a promenade with Sue. Al’s walk caused Tom’s complaint because they were supposed to go jogging together? Then we might want to refer to it, not as a walk or a promenade with Sue, but (negatively) as a non-jogging. To be sure, there is linguistic noise here. When using negative descriptions, we might be inclined to go for expressions that do not qualify as perfect nominals: in English, the cause of Tom’s complaint is best described, not as Al’s non-jogging, but as Al’s not jogging, and this latter expression is a gerundial. It has a verb that is still “alive and kicking inside it”, as Zeno Vendler famously put it11. And if Vendler is right, and Bennett with him, gerundial nominals do not strictly speaking pick out events: they typically pick out facts, or states of affairs. It is indeed unclear whether English admits of any grammatically decent examples of negative event descriptions that pass the Vendler-Bennett test12. But never mind that: to the extent that we intend to refer to an event, the point is clear enough: by speaking of Al’s not jogging we are not referring to a negative action; we are referring to what Al actually 10 See Bennett (1988). For a more detailed formulation of my views on these

issues, I refer to Varzi (2002). 11 Vendler (1967), Ch. 5. 12 See Bennett (1988), p. 148.

136

The Talk I Was Supposed to Give...

did—his walk—by mentioning a salient property that it lacked. A negative description has a negative sense, not a negative referent. I suggest that precisely this sort of account applies to the putatively problematic cases mentioned above, which involve hidden quantification rather than explicit event reference13. Suppose I say (1) Beth saw Al not leave. According to a standard analysis, due to Jim Higginbotham14, naked infinitive perceptual reports require a Davidsonian analysis. In particular, a statement such as (1) would be analyzed as (1') ∃e (non-leaving (e, Al) & saw (Beth, e))15. Does this mean that by uttering (1) I am committing myself to a negative event that Beth saw, i.e., a non-leaving, a leaving that Al did not perform? Some people think so,16 but I surely don’t. Surely Beth could only see what Al actually did, unless she was hallucinating. For instance, suppose Al rushed upstairs rather than leaving. Then that is what Beth saw: she saw him rush upstairs. And I could report this fact by uttering (2) instead of (1): (2) Beth saw Al rush upstairs. In that case the correct analysis would be (2') ∃e (rushing-upstairs (e, Al) & saw (Beth, e)). That I appeal to a negative characterization in (1) and a positive characterization in (2) doesn’t mean that the variable e must pick out two distinct events in the two cases. It picks out the same event under two different descriptions, and which description is more appropriate is a pragmatic question. For instance, suppose everybody expected Al to leave. Then (1) is much more effective as a report of what happened than (2). It is more effective because the negative attribute ‘non-leaving’ conveys 13 Here I elaborate on Tovena and Varzi (1999). 14 See Higginbotham (1983) and the discussion in Neale (1988). 15 Strictly speaking, since Beth’s seeing is itself an event, the fully spelled out log-

ical form of (1) should involve two bound variables: ∃e∃e'(non-leaving (e, Al) & seeing (e', Beth, e)). For simplicity, here and below I will ignore such details if they are not relevant. 16 Here I disagree with Przepiórkowski (1999), for instance.

Achille Varzi

137

a salient piece of information, whereas ‘rushing-upstairs’ does not. Sometimes we do exactly the same when it comes to speaking of objects. If I say (3) Beth saw a non-Meinongian, then I am characterizing what Beth saw with the help of a negative attribute. Surely that doesn’t mean that Beth saw a negative entity. In fact, if I knew more I could be more explicit, as in (4) Beth saw a 55 year-old gentleman wearing glasses, a blue tie, and a hat. Yet this need not be the best way of reporting what Beth saw. If, as we may suppose, Beth was attending a Meinong conference, then the negative attribute ‘non-Meinongian’, though less informative, picks out a feature of the person Beth saw that is much more salient than his age and look, so in that context (3) may well be more effective than (4). The case of (1) versus (2) is exactly parallel. And the parallel is confirmed by the fact that in both cases we can go on and ask: Tell me more. Tell me more about this non-Meinongian. “Well, he looked like a gentleman, about 55 years old, wearing glasses, a blue tie, and a hat”. Tell me more about what Beth saw. “Well, she saw Al rush upstairs, on the sly, without saying a word, while nobody else was looking”. Just as with the walk that was not a jogging, we can always switch from a negative characterization to a positive one, and vice versa: we choose our words depending on the message we wish to convey. Indeed, one further reason to reject the thought that a negative attribute picks out a negative event is that often there is no clear-cut criterion for saying whether an attribute is negative. Consider (5) Al stayed whereas Tom left. The connective ‘whereas’ suggests that staying and leaving are to be understood as opposite. But which is positive and which negative? If everybody else left with Tom, then what Al did could be described equally well (if not better) as a non-leaving. However, if everybody else stayed with Al, then it is what Tom did that could be described negatively as a non-staying. Obviously both options are equally plausible, so the opposition between staying and leaving cannot be construed ontologically, as a conflict between being and non-being. Nor

138

The Talk I Was Supposed to Give...

can we choose between the two options on purely semantic grounds. The choice depends on the context, so it is a function of pragmatic factors at large. At this point it should be clear how other problematic cases can be handled along similar lines. Consider: (6) What happened in the end is that Sue didn’t kiss Al. Henriette de Swart takes statements such as this to provide evidence for the existence of negative events—in this case, a non-kiss by Sue17. Presumably this is because, on a standard analysis, the logical form of (6) implies the existential statement (7) ∃e (non-kissing (e, Beth, Al)). But so what? On the face of it, what happened at the end, e, must be something perfectly positive, ontologically speaking. It must be something that did happen. Al didn’t tell us, so we don’t know. Perhaps Sue just said “Good night”, or “Thanks for the pleasant walk”. Perhaps she hopped into a taxi cab without saying a word. Who knows? What we know, based on what Al told us, is that whatever she did, it was not a kissing: (7') ∃e (~kissing (e, Beth, Al)). And since we are starting to feel judgmental about this whole story, we are happy to accept this negative characterization as an adequate characterization of what happened at the end, uninformative as it may be in all sorts of respects. Here is another typical case where we seem to be quantifying over negative events: (8) Al often doesn’t go jogging. Stockwell, Schachter, and Partee18 take statements such as (8) to provide counterexamples against Lakoff’s claim that “one cannot assert the frequency of an event that does not occur”19, I see why one might 17 See de Swart (1996), p. 229. 18 Stockwell et al. (1973), ex. (49b). Compare Horn (1989), p. 54, de Swart

(1996), p. 230, and Przepiórkowski (1999), p. 241. 19 Lakoff (1965), p. 172.

Achille Varzi

139

be tempted to view it that way, but I disagree. That is, I agree with Lakoff. For one thing, it seems to me that statements such as (8) should not to be construed as quantifying over events at all, whether positive or negative. Consider first (9) Al often goes jogging. On the face of it, in (9) we are quantifying over certain times—over evenings, say—and we are saying that there are many such times during which Al goes jogging. In other words, we are quantifying over the open formula: (9') Evening (t) & ∃e (jogging (e, Al) & during (e, t)). But then (8) should be handled similarly. In (8) we are again quantifying over certain times—evenings—and we are saying that there are many such times during which Al does not go jogging: we are quantifying over the formula (8') Evening (t) & ~∃e (jogging (e, Al) & during (e, t)). There is nothing here that should suggest the repeated occurrence of a negative event. On the other hand, if we took such statements to be truly quantifying over events, then again I see no reason to suppose that the quantification in (8) ranges over negative events. What did Al do yesterday evening? He went for a walk with Sue. What did he do the evening before? He went to the bookstore with Beth. And on Tuesday he went to see the Yankees play. All of those things are things that Al actually did, and they all took place in the evening. We may describe them directly, as I have just done, or we may describe them indirectly as doings that are not joggings—as non-joggings. That is not very informative. But for someone like Tom, who loves jogging, it may be enough. And it may be perfectly all right for Tom to summarize the picture by saying that Al often doesn’t go jogging. Al is the president of the jogging club, so in the evening he shouldn’t engage in all those non-joggings— whatever they are. Causation and Causal Explanations What about putative cases of causation by omission? “Every causal situation develops as it does as a result of the presence of positive factors

140

The Talk I Was Supposed to Give...

alone”, says David Armstrong20, and I agree. Yet our ordinary way of speaking—and much philosophical talk, too—suggests that the driving forces of nature also include omissions, failures, things that did not happen. Can we resist that suggestion by exploiting the idea that prima facie negative causes are just positive causes under a negative description? To some extent, the answer is straightforward. I have already hinted at the idea above, but let me spell it out more explicitly. On a Davidsonian semantics, singular causal statements are best construed as statements involving a causal relation between events, not as statements involving a causal connective or other propositional devices (on pain of absurdity)21. For example, the statement (10) Al’s walk with Sue caused Beth’s anger is to be construed as having the following form (modulo uniqueness): (10')∃e ∃e' (walk (e, Al, Sue) & anger (e', Beth) & caused (e, e')). Accordingly, consider now a statement that appears to elicit a negative cause, such as (11) Al’s non-jogging caused Tom’s complaint. The corresponding logical form would be (11')∃e ∃e' (non-jogging (e, Al) & complaint (e', Tom) & caused (e, e')). One might be inclined to maintain that e, the cause, must be a negative event insofar as it must answer to a negative predicate, but we have seen that that is a non-sequitur. We know what Al did last night: he took a walk with Sue. So that is what triggered Tom’s complaint—a perfectly positive cause. The reason why we may not want to put it that way is that the statement (12) Al’s walk with Sue caused Tom’s complaint is likely to convey the wrong message. The statement would be true, because Al’s walk with Sue and Al’s non-jogging with Tom are one and the same event under two different descriptions, and causal reports are semantically transparent. So the logical form of (12), 20 Armstrong (1999), p. 177. 21 See Davidson (1967b).

Achille Varzi

141

(12')∃e ∃e' (walk (e, Al, Sue) & complaint (e', Tom) & caused (e, e')), would be materially equivalent to that of (11). But as Bennett pointed out, “in general, truths about causes will be assertible only if they report causes that are salient—that is, stand out as notably significant, surprising, or the like”22. In general, when we offer a causal report we are not just trying to say something true; we are trying to explain why something happened. And the adequacy of an explanation does not depend exclusively on the truth of what is said: as with every speech act, it depends also, and to a great extent, on the relevant background of shared knowledge and presuppositions. (Grice’s Cooperative Principle: “Make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged”23.) Thus, if the only nonMeinongian at the conference was a gentleman wearing glasses, the statement (13) Seeing a non-Meinongian surprised Beth is true if and only if the statement (14) Seeing a gentleman wearing glasses surprised Beth is also true. But if we want to explain why Beth was surprised, given that she was attending a Meinong conference, then (13) is likely to be more effective. Likewise, then, if we want to explain why Tom complained, we should better describe what Al did in a way that brings out the fact that it didn’t measure up to Tom’s expectations. Tom expected Al to go jogging, and it is the fact that Al’s action turned out to be something else than a jog that explains the complaint. That is why (11) is better than (12). In fact, that is why (11), but not (12), translates naturally into the familiar jargon of causal explanations: (11'')Tom complained because Al did not go jogging. 22 Bennett (1995), p. 133. 23 Grice (1975), p. 45. This is not to say that the value of a causal explanation is

always a matter of pragmatics. For instance, as Hempel (1965) famously emphasized, scientific explanations are meant to be as objective and non-pragmatic as the laws and theories on which they rest. But our concern here is with ordinary explanations, hence the usual Gricean standards apply.

142

The Talk I Was Supposed to Give...

(12'')Tom complained because Al went for a walk with Sue. That Al’s action was a walk with Sue—or that it took place in Central Park, before dinner, and so on—may be more informative than its not being a jog, but that information explains nothing when it comes to Tom’s complaint. So here is how the account would go in a nutshell. We sometimes construe a causal report as a causal explanation, but in order for the explanation to be successful we must use a description of the cause that conveys the right sort of information. Exactly what counts as the right sort of information depends on the context. And in certain contexts, negative descriptions are just more informative, causally, than positive descriptions24. They are in the negative, but they nonetheless pick out a perfectly “positive” cause. There are, however, cases in which this straightforward account does not apply. Consider the example mentioned at the beginning: (15) Al’s failure to turn off the gas caused an explosion. If Al tried to turn off the gas—for example, if he tinkered with the gas knob with the intention of turning off the gas, and the knob broke off— then (15) is no different from the cases discussed above: ‘Al’s failure to turn off the gas’ is just a negative description of his unsuccessful fiddling with the knob. But suppose Al didn’t try. Suppose he just forgot that the gas was on. Then there is no good candidate for the analysis offered above. As Higginbotham has emphasized25, there would be no point is claiming that we have a negative description of a positive cause, for we have no clue as to what that cause might be. We are not told what Al actually did. Yes, he left the gas on, but that’s just another way of saying that he failed to turn it off; for all we know, no truly positive action of Al (reading a book, going for a walk, talking to Beth

24 If you like facts, you may want to say that whereas causation is a relation

among events, causal explanations relate facts. And when it comes to fact descriptions, differences in logical power create differences of referent. But one does not need an ontology of facts to cash out the distinction: one can just say that causal explanations relate statements, or sets of statements. And when it comes to statements, what you see is what you get. 25 See Higginbotham (2000), p. 74.

Achille Varzi

143

on the phone) may be directly involved in the causal history of the explosion. Any Davidsonian analysis of (15) along the lines of (15') ∃e ∃e' (non-turning-off (e, Al, gas) & explosion (e') & caused (e, e')) would therefore seem to commit us to the existence of a cause e that is strictly and literally negative: a failure, an omission, a non-doing by Al. Here is where the second idea advertised at the beginning enters the picture. Granted, in cases such as (15) we cannot just call upon straightforward pragmatic considerations. The problem is not that we tend to construe such causal reports as causal explanations, with the additional constraints that this involves; the problem arises with their being causal reports in the first place. However, precisely this may be questioned. For what grounds do we have to say that (15) is genuinely a causal report? Of course its surface grammar is akin to that of (11), (12), and the like: it features the word ‘caused’ explicitly. But we know that the surface grammar of a statement may be misleading. What reasons are there to suppose that every statement of the form ‘a caused b’ should be taken strictly and literally, and subjected to Davidson’s analysis? In a companion paper26, I contend that we have no good reasons. There is a natural disposition to think that whenever we engage in causal talk, we do so by speaking of causes and effects, but this is wrong. As we have just seen, often we are interested in providing a causal explanation of why something happened; and although typically a causal explanation is a certain way of reporting the existence of a causal nexus—a contextually salient way of citing a cause of what happened—it need not be so. As Helen Beebee convincingly argued, “the explanans of a causal explanation need not stand to the explanandum as cause to effect”27. Indeed, a causal explanation need not even mention a cause in order to fit the bill. If we know what triggered a certain event, and we know it under a suitable description, then it is natural to formulate an explanation on such grounds, as with (11'') (and unlike (12'')). But the converse need not hold. If I say (16) Tom complained because nobody went jogging with him, 26 Varzi (2006). 27 See Beebee (2003) (the quote is from p. 301). Beebee’s argument, in turn, rests

on the theory of explanation of Lewis (1986), esp. section III.

144

The Talk I Was Supposed to Give...

I am offering a perfectly good explanation of Tom’s complaint— indeed a causal explanation. Nonetheless (16) is silent about the actual cause of the complaint: there is no mention, direct or indirect, of the particular event that brought it about, which is to say Al’s walk. To put it differently, (16) does not inform us about the causal history of Tom’s complaint by telling us what event(s) belong to it. Rather, it informs us about that causal history by telling us that it does not include any event of a certain type—any event of someone jogging with Tom—and that suffices for (16) to qualify as a good causal explanation. Sometimes we produce a causal explanation of a certain effect e by mentioning e’s cause(s). Sometimes we do not: we just point out that one sort of event that was supposed to occur, and whose occurrence would have prevented a certain effect, did not actually occur. In such cases, we have a causal explanation that cannot b matched by a genuine causal report. In the case of (16), we just have a causal explanation. Well, then: I think this is exactly what we should say about (15), too. Our statement reads like a causal report, to be analyzed as in (15'), but it isn’t. It is just a causal explanation in disguise: (15'') There was an explosion because Al didn’t turn off the gas. In some cases ‘because’ goes proxy for ‘cause’ and Davidson’s analysis applies. Such is the case of (11) and (11''), or (12) and (12''), and we could say the same with respect to statements such as: (17) Beth’s turning on the light caused an explosion. (17'') There was an explosion because Beth turned on the light. In other cases, like (15) and (15''), it’s exactly the other way around. And although we can always switch from the ‘cause’ language to its ‘because’ counterpart, the converse does not hold. Every causal report translates directly into a causal explanation, but not vice versa. So what should we say about (15) as it stands? I think we should just say that strictly speaking this statement is false, or at least not true. It is not true because the subject term, ‘Al’s failure to turn off the gas’ (unlike the term ‘Beth’s turning on the light’), has no referent. Why are we inclined to assert (15) if this statement is not true? The answer, I think, lies once again in a fact about our conversational practices, though a general fact that goes beyond Gricean considerations and takes us back to our starting point. When we say, for example,

Achille Varzi

145

(18) Holmes lived in Baker Street, we say something which, strictly speaking, is not true (or so I would argue). Strictly speaking, what is true is a contextualized claim of the form: (18')According to Conan Doyle’s narrative, Holmes lived in Baker Street. When we say (19) The average star has 2.4 planets, again we say something which, strictly speaking, is not true. There are no such things as average stars. Strictly speaking, what is true is a statement along the following lines: (19') There are 12 planets and 5 stars, or 24 planets and 10 stars, or...28. When speaking with the vulgar, we often say things that, strictly speaking, are not true, and we do so because we count on a charitable reading of our statements, or simply because we don’t care about their ontological underpinnings. When we do care, however, we must pay attention and, if necessary, clarify what we mean by producing an ontologically adequate paraphrase. Likewise with (15), I submit. When speaking with the vulgar we may be inclined to assert that the explosion was caused (among other things) by Al’s failure to turn off the gas. But this is only speaking with the vulgar. If pressed, we should speak differently. If pressed, we should either revert to a genuine causal report, as in (17), or else we should abandon the ‘cause’ language in favor of the ‘because’ language and assert (15'') instead. As Davidson himself put it, the ‘caused’ of a sentence such as (15) (unlike that of a sentence such as (17)), “is not the ‘caused’ of straightforward singular causal statements, but is best expressed by the words ‘causally explains’”29. Trying Not To So much for the complications arising from familiar cases of alleged “causation by omission”. Let me conclude with some brief remarks 28 The example is from Melia (1995), p. 224. 29 See Davidson (1967b), pp. 161f.

146

The Talk I Was Supposed to Give...

concerning the last worry mentioned in the introduction, which is more intimately related to the philosophy of action broadly understood. We have seen that sometimes the use of a negative description goes hand in hand with the outcome of a certain action. If Al actually tried to turn off the gas (he tinkered with the gas knob), and if his trying did not yield the desired result (the knob broke off), then speaking of his failure is just speaking of his unsuccessful trying, i.e., of his tinkering with the gas knob. The notion of trying, however, is itself troublesome. For we can try to do something just as we can try not to do something. In the first case, the something we are trying to do is an action of some sort (turning off the gas, for instance). But what about the second case? Shall we say that when we try not to do something, our trying is directed towards a negative action of some sort?30 To be sure, the first example I mentioned at the beginning is easy enough: I tried hard not to miss the deadline if and only if I tried hard to meet it—one goal under two different descriptions. Or consider (20) Al tried not to move. There is no pressure to posit Al’s goal as a negative event: a non-moving is just a staying put (under a negative description), i.e., we can rephrase (20) as (21) Al tried to stay put. And the trying itself may be described as an action, or a complex of actions, that are perfectly standard: Al worked hard to master urges to move his body, possibly with the help of mental self-control techniques. But consider the following: (22) Al tried not to call Sue. In this case, there is no obvious way of describing what Al was trying to do in positive terms. As with (15), no good candidate seems available for a simple analysis in terms of negative re-descriptions of positive events. For what could the positive goal of Al’s trying be? Writing a letter instead? Crawling into bed and falling asleep? Leaving for 30 Apart from the ontological scruples that motivate the present work, this sort of

worry arises naturally in the philosophy of action—for instance, it afflicts those theories according to which one who is not trying to do anything at all is not sentiently directing one bodily motions at anything. See e.g. Mele (2003), sec. 6.4.

Achille Varzi

147

Cuba? We have no clue. And even if it turns out that he did one of those things, and that he did it as a result of his trying to do it, it would be hard to maintain that this is what (22) is all about. It is one thing to try to fall asleep, quite another to try not to call someone, even if one thing implies the other. I think this is another case where our intuitions and linguistic practices are seriously misleading. On the one hand, the difficulty is compounded by the fact that the language of trying, unlike the language of perceptual reports or the language of singular causation, admits of no obvious logical analysis. Surely it is not just a matter of implicit quantification. Just like the statement (23) Al sought a lion cannot be analyzed as (23') ∃x (lion (x) & sought (Al, x)) (as Quine famously argued31), likewise a statement such as (21) cannot be analyzed along the lines of (21')∃e (staying-put (e, Al) & tried-for (Al, e)), for (21) may be true even if no event of staying put (with Al as an agent) took place32. At least this is obvious on the generic, or de dicto, understanding of such statements, which in the case of (21) is the only reasonable reading; what would it mean to say that Al was aiming at (i.e., trying to realize) a particular action of staying put, as opposed to an action of that sort?33 Accordingly, when we say that a trying is directed towards a putatively negative action, as in (22), we should not automatically construe our statement as involving a compromising hidden quantifier. For such a statement may be true even if Al’s trying did not succeed, i.e., even if he did end up calling Sue. 31 Quine (1956). 32 For simplicity, I ignore the fact that the predicates ‘sought’ and ‘tried-for’ may

in turn be analyzed in Davidsonian fashion (see note 15). 33 Of course, staying put would be a pretty boring action (perhaps not even an action), but it would still qualify as an occurrence of some sort. Following Bennett (1988), § 60, and contra Lombard (1986), I don’t see any interesting structural difference between “dynamic” occurrences (i.e., events that involve change) and “static” property instances.

148

The Talk I Was Supposed to Give...

Indeed, the analogy with seeking is telling: on a de dicto reading, there is no significant difference between (23) and (24) Al sought a unicorn, so one could hardly take the truth of the latter to provide evidence in favor of an ontology of Meinongian non-existents34. On the other hand, one could rejoinder that the analogy between seeking and trying fires back. On a standard analysis35, the correct logical form of such statements as (23) and (24) (on their de dicto reading) is given by: (23*) Al strove that ∃x (lion (x) & finds (Al, x)). (24*) Al strove that ∃x (unicorn (x) & finds (Al, x)). This is somewhat controversial, because it leaves it open whether ‘Al strove that’ should in turn be subjected to a Davidsonian, event-based analysis, but never mind: there is a clear sense in which (23*) and (24*) are on the right track. The trouble is that a corresponding analysis of (21) and (22), along the lines of (21*) Al strove that ∃e (staying-put (e) & agent (e, Al)) (22*) Al strove that ∃e (non-calling (e, Sue) & agent (e, Al)), would seem to require that we treat Al’s non-calling Sue on a par with his his staying put—and this is bad news. We have seen that there is no plausible way of treating ‘non-calling’ as a negative description of something positive Al was trying to do. Thus, if (22*) is correct, we seem to be stuck with a negative event after all. Nor can we avoid this conclusion by reading the negative particle ‘non’ as a propositional ‘not’, as in (22**) Al strove that ∃e (~ calling (e, Sue) & agent (e, Al)). Surely (22) does not merely say that Al tried to do something else than calling Sue. For it goes without saying that Al tried to do very many such things in the relevant interval of time: he tried (and managed) to put on his pants; he tried (and failed) to turn off the gas, and so on. 34 This is acknowledged, for instance, by Parsons (1980), § 2.4. 35 I am thinking of Quine’s own analysis in (1956). The recent literature has been

influenced greatly by the work of Montague (1969).

Achille Varzi

149

Perhaps here is where the analogy between ‘seeking’ and ‘trying’ breaks down, on pain of countenancing negative events, and surely the literature offers a variety of alternatives36. Perhaps so. But I would rather say that it is the analogy between trying to do something and trying not to do something that breaks down. For when we try to do something, we are striving for there to be some event of a certain kind. When we try not to do something, however, our endeavors admit of two different construals: one can push the analogy and say that we are again striving for there to be some event of a certain (negative) kind; but one can also say that we are striving for there to be no event of a certain (positive) kind. The first option is reflected in (22*). But I see no good reason to favor that option over its alternative. On the contrary, the second option strikes me as much more plausible, regardless of any ontological considerations. If so, then, the analogy between the two sorts of tryings breaks down: (21*) gets the quantifier right, but (22*) does not. The correct way of representing the logical form of (22) is this: (22***) Al strove that ~∃e (calling (e, Sue) & agent (e, Al)). And surely enough, (22***) does not commit us to any dubious ontological creatures. As I said, this may not be the end of the story insofar as the phrase ‘Al strove that’ may call for further analysis. After all, it is this phrase that now gives expression to Al’s trying, and we may want to say that the latter is itself an action or a complex of actions that Al performed. But this is now a matter of detail and we need not worry: if this is what tryings are, they have a perfectly positive make-up. So, however we continue from here, I think we are now on the right track. The thought that tryings may involve negative outcomes rests on a false analogy.

References

Anscombe, G.E.M. (1957), Intention, Oxford, Basil Blackwell. –– (1979), ‘Under a Description’, Noûs, 13, pp. 219–233. Armstrong, D.M. (1999), ‘The Open Door’, in (ed.) H. Sankey, Causation and Laws of Nature, Dordrecht, Kluwer, pp. 175–185.

36 See for instance Ginet (2004).

150

The Talk I Was Supposed to Give...

Beebee, H. (2003), ‘Causing and Nothingness’, in (eds.) J. Collins, H. Hall, and L. A. Paul, Causation and Counterfactuals, Cambridge (MA), MIT Press, pp. 291–308. Bennett, J. (1988), Events and Their Names, Oxford, Clarendon Press. –– (1995), The Act Itself, Oxford, Clarendon Press. Davidson, D. (1967a), ‘The Logical Form of Action Sentences’, in N. Rescher (ed.), The Logic of Decision and Action, Pittsburgh, University of Pittsburgh Press, pp. 81–95; reprinted in Davidson (1980), pp. 105–122. –– (1967b), ‘Causal Relations’, Journal of Philosophy, 64, 691–703; reprinted in Davidson (1980), pp. 149–162. –– (1970), ‘Events as Particulars’, Noûs, 4, pp. 25–32; reprinted in Davidson (1980), pp. 181–187. –– (1980), Essays on Actions and Events, Oxford, Clarendon Press. –– (1985), ‘Reply to Bruce Vermazen’, in (eds.) B. Vermazen and M.B. Hintikka, Essays on Davidson: Actions and Events, Oxford, Clarendon Press, pp. 172–176. de Swart, H. (1996), ‘Meaning and Use of not … until’, Journal of Semantics, 13, pp. 221–263. Ginet, C. (2004), ‘Trying to Act’, in (ed.) J.K. Campbell, Freedom and Determinism, Cambridge (MA), MIT Press, pp. 89–102. Goldman, A. (1971), ‘The Individuation of Actions’, Journal of Philosophy, 68, pp. 761–774. Grice, H.P. (1975), ‘Logic and Conversation’, in (eds.) P. Cole and J. L. Morgan, Syntax and Semantics: Speech Acts, New York, Academic Press, pp. 41–48. Hempel, C. (1965), ‘Aspects of Scientific Explanation’, in his Aspects of Scientific Explanation and Other Essays in the Philosophy of Science, New York, Free Press, pp. 331–496. Higginbotham, J. (1983), ‘The Logic of Perceptual Reports: An Extensional Alternative to Situation Semantics’, Journal of Philosophy, 80, pp. 100–127. –– (2000), ‘On Events in Linguistic Semantics’, in (eds.) J. Higginbotham, F. Pianesi, and A.C. Varzi, Speaking of Events, New York, Oxford University Press, pp. 49–79. Horn, L.R. (1989), The Natural History of Negation, Chicago, Chicago University Press. Kim, J. (1976), ‘Events as Property Exemplifications’, in (eds.) M. Brand and D. Walton, Action Theory, Dordrecht, Reidel, pp. 159–177. Lakoff, G. (1965), Irregularity in Syntax, New York, Holt, Reinhart & Winston. Lewis, D.K. (1973), ‘Causation’, Journal of Philosophy, 70, pp. 556–567. –– (1986), ‘Causal Explanation’, in his Philosophical Papers: Volume II, Oxford, Oxford University Press, pp. 214–240. Lombard, L.B. (1986), Events: A Metaphysical Study, London, Routledge & Kegan Paul.

Achille Varzi

151

Mele, A.R. (2003), Motivation and Agency, Oxford, Oxford University Press. Melia, J. (1995), ‘On What There’s Not’, Analysis, 55, pp. 223–229. Montague, R. (1969), ‘On the Nature of Certain Philosophical Entities’, The Monist, 53, pp. 159–194. Neale, S. (1988), ‘Events and “Logical Form”’, Linguistics and Philosophy, 11, pp. 303–321. Parsons, T. (1980), Nonexistent Objects, New Haven (CT), Yale University Press. Przepiórkowski, A. (1999), ‘On Negative Eventualities, Negative Concord, and Negative Yes/No Questions’, in (eds.) T. Matthews and D. Strolovitch, Proceeding of Semantics and Linguistic Theory 9, Ithaca (NY), CLC Publications, pp. 237–254. Quine, W.V.O. (1948), ‘On What There Is’, Review of Metaphysics, 2, pp. 21–38. –– (1956), ‘Quantifiers and Propositional Attitudes’, Journal of Philosophy, 53, pp. 177–187. Russell, B. (1905), ‘On Denoting’, Mind, 14, pp. 479–493. Stockwell, R., Schachter, P. and Partee, B. (1973), The Major Structures of English, New York, Holt, Rinehart & Winston. Thomson, J.J. (1971), ‘The Time of a Killing’, Journal of Philosophy, 68, pp. 115–132. Tovena, L.M., and Varzi, A.C. (1999), ‘Events and Negative Descriptions’, in Sinn und Bedeutung 1999. Fourth Annual Conference of the Gesellschaft für Semantik, Düsseldorf, Heinrich-Heine Universität, pp. 71–73. Varzi, A.C. (2002), ‘Events, Truth, and Indeterminacy’, The Dialogue, 2, pp. 241–264. –– (2006), ‘Omissions and Causal Explanations’, forthcoming in (eds.) J. Quitterer and F. Castellani, Agency and Causation in the Human Sciences, Paderborn, Mentis Verlag. Vendler, Z. (1967), Linguistics in Philosophy, Ithaca (NY), Cornell University Press. Vermazen, B. (1985), ‘Negative Acts’, in (eds.) B. Vermazen and M.B. Hintikka, Essays on Davidson: Actions and Events, Oxford, Clarendon Press, pp. 93–104.

Two Interpretations of “According to a Story” Maria E. Reicher I The general topic of this paper is the ontological commitment to socalled “fictitious objects”, that is, things and characters of fictional stories, like Sherlock Holmes and Pegasus. Discourse about fiction seems to entail an ontological commitment to fictitious entities, a commitment that is often deemed inconsistent with empirical facts. Consider, for instance, the following sentence: (1) Pegasus is a flying horse. This seems to be a true sentence about a character of Greek mythology. In case you are reluctant to accept it as true, compare the following: (2) Pegasus is a flying dog. Probably most people (even those who have a certain inclination to raise doubts about the truth of the first sentence) share the intuition that the first sentence is closer to the truth than the second. Given the principle of bivalence, the most obvious way to do justice to this intuition is to say that the former sentence is true while the latter is false. According to the principle of existential generalization, (EG) Fa → There is at least one F. the sentence (1) Pegasus is a flying horse. implies (3) There is at least one flying horse. Furthermore, according to a principle that one might call the “predication principle”, (PP) Fa → a exists. (1) implies that there is something that is identical with Pegasus, or

154

Two Interpretations of “According to a Story”

(4) Pegasus exists, for short. Thus it seems that a person who accepts the sentence (1) cannot consistently deny that Pegasus exists and that there is at least one flying horse. Yet most people who are inclined to accept (1) as true at the same time believe that there are no flying horses and that Pegasus does not exist and never existed. This is, in a nutshell, the problem of discourse about fiction. Basically, the problem consists in the fact that a system of beliefs seems to be inconsistent. Three strategies to get rid of this inconsistency suggest themselves: 1. One might change one’s overall system of beliefs to the effect that sentences like (1) are rejected as false – at least if they are taken at face value. (See Ryle 19331.) 2. One might change one’s logic to the effect that the principle of existential generalization and the predication principle are rejected as invalid. (See Leonard 1956, Hintikka 1959, Leblanc/Hailperin 1959, Rescher 1959, Lambert 1983.) 3. One might change one’s ontology to the effect that one accepts that flying horses in general, and Pegasus in particular, exist. (See in particular van Inwagen 1977, Thomasson 1999.) In this paper, I want to discuss two solutions that have been proposed and discussed in the literature. The first one is a special case of the first strategy, that is, it amounts to rejecting sentences like (1) as false, at least if they are taken at face value. This solution is called the “storyoperator strategy”. It consists in paraphrasing sentences like “Pegasus is a flying horse” by means of the expression “according to a story” or “according to the story s” (or some synonym to these expressions)2 such that, for instance, (1) is interpreted as an elliptic formulation of 1 Ryle argues that sentences that seem to be sentences about fictitious objects are

in fact either “about” nothing at all (and neither true nor false) or about (allegedly) metaphysically innocuous entities like books, sentences or names. 2 “According to a story” is an indeterminate story operator (“according to some story or other”), in contrast to the determinate story operator “According to the

Maria Reicher

155

(1a) According to a story, Pegasus is a flying horse. (For a story-operator strategy see Künne 1990 and 1995.) I shall argue that the story-operator strategy is unsatisfactory. The second solution to be discussed here falls under the third strategy, that is, it is assumed that Pegasus and other fictitious objects exist. I shall argue in favour of this solution, and I shall spell it out in some detail. The realist ontology of fictitious objects that will be defended here is not entirely new. The uncommon aspect is, however, that in order to explicate it, I make use of the story operator, but I use it in a way that is quite different from its use within the so-called “story-operator strategy”3. Within the story-operator strategy, the story operator is supposed to block the inference from (1) Pegasus is a flying horse. to (3) There are flying horses. and to (4) Pegasus exists. story s”, where “s” stands for some particular story, for instance the Greek mythology. Both sorts of operators are used in the literature. Whether one of them is preferable to the other (and if so, which one) depends on the particular context. In this paper, I use the indeterminate operator throughout. The main reason for this decision is that the interpretation of the story operator that I am going to propose works more smoothly for the indeterminate operator – though, in principle, it can be applied to determinate operators as well (see note 9). 3 One may interpret Kit Fine’s distinction between a “literal” and a “story-relative copula” as an expression of a similar idea. The story-relative copula is said to hold “between an object, property and story when the object has the property in the story” (Fine 1982, 108). However, Fine’s aim is not a re-interpretation of the story operator and he does not endorse a realist ontology of fiction at all (at least not in this paper). Furthermore, there are differences between Fine’s interpretation of the “non-literal” copula and my own: I do not interpret it as expressing a relation between an object, a property, and a story. (More will be said on the “non-literal” copula below in sections IV and V.)

156

Two Interpretations of “According to a Story”

From (1a) According to a story, Pegasus is a flying horse. we cannot infer that there are flying horses; neither can we infer that there is something that is identical with Pegasus. At most, we can infer the following: (3a) According to a story, there are flying horses. and (4a) According to a story, Pegasus exists. This, however, does not contradict the empirical fact that Pegasus does not (“really”) exist and that there are no flying horses (“in reality”). Therefore, a person who accepts the sentence (1a) as true is not bound to accept the existence of flying horses in general and Pegasus in particular. Thus, the ontological commitment to flying horses in general and Pegasus in particular is avoided. Or so it is argued. There is, however, an ambiguity in the story operator that usually goes unnoticed. In principle, the story operator can be interpreted in two ways: either as a sentence operator or as a predicate operator. If it is used as a sentence operator, the sentence as a whole is within its scope. According to this interpretation, the sentence (1a) has the logical structure “S (Fa)”, (1as) According to a story [Pegasus/is a flying horse], i.e., the structure of a predication within the scope of an operator. (The slash indicates a way in which the bracketed sentence may be “cut up” into its logical parts.) If, however, the story operator is interpreted as a predicate operator, the sentence has the structure “Fa”, (1ap) Pegasus/is according to a story a flying horse, i.e., the structure of a predication. In this case, the story operator functions as a “predicate modifier”. The rendition (1ap) makes it explicit that the expression “according to a story” belongs to the predicate4.

4 In the remainder of the paper, I shall use renditions of the kind of (1as) and (1ap)

whenever I wish to make it explicit that the story operator should be interpreted as a sentence operator or as a predicate modifier, respectively. In some cases,

Maria Reicher

157

Usually, the story operator is interpreted as a sentence operator, for it is in this interpretation only that it can block the two unwelcome inferences. For if (1) Pegasus is a flying horse. implies that Pegasus exists, then (1ap) also implies that Pegasus exists. The greater complexity of the predicate in the latter sentence does not, of course, prevent this inference. Nevertheless, in what follows I shall argue that the story operator should be interpreted as a predicate modifier. My argument is twofold: 1. As a sentence operator, the story operator does not fulfil its intended task, that is: contrary to what has been said so far, the story-operator strategy does not help to avoid the ontological commitment to fictitious objects. 2. As a predicate modifier, the story operator can play an important role within a consistent and plausible realist ontology of fiction. II Why does the story operator as a sentence operator fail to prevent the ontological commitment to fictitious objects? The reason is that the story-operator strategy works well for some cases but it does not work at all for others. Consider the following sentence: (5) Pegasus is a character from Greek mythology. If we put a story operator in front of this, we get: (5a) According to a story [Pegasus is a character from Greek mythology]. Unfortunately, this story-operator paraphrase is not truth preserving. Given that Greek mythology is the only story in which Pegasus occurs, (5a) is false, despite the truth of (5). For according to Greek mythology, Pegasus is not a character, but an animal of flesh and blood; and there is no Greek mythology within Greek mythology.

however, I shall not commit myself to one interpretation or the other but leave the story operator intentionally ambiguous. In these cases, I use the standard formulation “According to a story, …”.

158

Two Interpretations of “According to a Story”

Greek mythology is the “truthmaker” of sentence (5), but it is not the truthmaker of (5a)5. Similarly, while it is true that (6) Miss Marple has been impersonated by Margaret Rutherford, it is (probably) not true that (6a) According to a story [Miss Marple has been impersonated by Margaret Rutherford]6. And while it may be true that (7) Sherlock Holmes is more famous than Hercule Poirot, it is (to my knowledge at least) not true that (7a) According to a story [Sherlock Holmes is more famous than Hercule Poirot]7. We may distinguish two kinds of sentences about fictitious objects: internal and external ones. Internal sentences about fictitious objects are true “within” or “according to” a story. (1) Pegasus is a flying horse. is an internal sentence. External sentences about fictitious objects are straightforwardly and literally true (if they are true). Examples of external sentences are: (5) Pegasus is a character from Greek mythology. (6) Miss Marple has been impersonated by Margaret Rutherford. (7) Sherlock Holmes is more famous than Hercule Poirot.

5 It is, of course, possible that there is a story s (and s may even be a fictional

story) such that, according to s, there is something called “Pegasus” which is a character from Greek mythology, in which case the Greek mythology would be a story within a story and (5a) would be true. But this is, evidently, not sufficient to make (5a) an adequate paraphrase of (5). 6 (6a) is not true unless there is a story s such that it is part of s that there is an actress named “Margaret Rutherford” and a fictitious character named “Miss Marple” and that the former has impersonated the latter. 7 That is, to my knowledge, there is no story in which both Sherlock Holmes and Hercule Poirot occur.

Maria Reicher

159

Thus, we may say that the story-operator strategy, as a strategy to avoid ontological commitment to fictitious entities, works for internal sentences only; it does not work for external sentences. Unfortunately, for a proponent of the story-operator strategy, it cannot be enough to have a good paraphrase for internal sentences only. For the acceptance of external sentences too entails an ontological commitment to fictitious entities. According to the predication principle, (5) Pegasus is a character from Greek mythology, implies that Pegasus exists; (6) Miss Marple has been impersonated by Margaret Rutherford, implies that Miss Marple exists; and (7) Sherlock Holmes is more famous than Hercule Poirot, implies that both Sherlock Holmes and Hercule Poirot exist. Proponents of the story-operator strategy must either simply declare all external sentences about fictitious objects to be false (which is an extremely implausible move), or they have to find some ad hoc paraphrase strategies for them (which is not much better). Thus, the story operator fails as a means to avoid the ontological commitment to fictitious entities. III Here is an alternative proposal: There is no satisfactory way to avoid the ontological commitment to fictitious entities. But this is not a reason to worry, since there is no compelling reason to search for a way to avoid the ontological commitment to fictitious entities. To the contrary, there are good reasons to accept the ontological commitment to fictitious entities. The good reasons are provided by the fact that reference to and quantification over fictitious entities is a part of both everyday and scientific and even of juridical discourse. Talk about fictitious objects is not only part of literary theory and film criticism, but even part of copyright. (Fictitious characters are explicitly protected by particular copyright regulations, just like novels, paintings, and film works.) Thus, the existence of fictitious objects can and should be accepted as an empirical fact. It is an empirical fact that there are fictitious

160

Two Interpretations of “According to a Story”

characters of novels and plays, fairy tale princes, and movie heroes, since it is an empirical fact that there are fictional novels and plays, fairy tales and fictional movies in which these characters occur. Outside philosophical circles this fact is generally accepted. But many philosophers find it hard or even impossible to accept the existence of fictitious objects. This calls for an explanation. Often (though not always) the reason for the denial of the existence of fictitious objects is a sort of category mistake. This category mistake consists in the (often tacit) assumption that fictitious objects, if they existed, would be spatio-temporal objects. If Pegasus existed, the reasoning goes, it would be a material object, a living being of flesh and blood, an object that is accessible to the senses and located in space. As a matter of fact, there is no living being which has all (or most) of the properties ascribed to Pegasus in the relevant story. Thus, Pegasus does not exist. But this reasoning is mistaken. Fictitious objects are not spatio-temporal objects. They belong to the same category as novels, plays, and fairy tales. Just like novels, plays, and fairy tales, they owe their existence to particular mental processes of authors. They are products of human imagination, “mental creations”, to use a term taken from copyright. Just like novels, plays, and fairy tales, they are contingent abstract objects, that is, they are neither physical nor mental, neither accessible to the senses and located in space nor dependent on particular acts of consciousness, once they have been created, though, they have been brought into being through particular acts of consciousness. Fictitious objects are part of our cultural reality, just like novels, symphonies and scientific theories. Thus, acceptance of the existence of fictitious objects is not a symptom of a dramatic loss of sense of reality. The belief that there is a fictitious flying horse called “Pegasus” does not entail the belief that Pegasus (or at least remnants of it) could be discovered, in principle, by a zoologist or an archaeologist. The belief that there is a fictitious detective called “Sherlock Holmes” does not entail the belief that it is or was possible at any time to shake hands with “him”. Fictitious objects are just not the kind of objects to be discovered by archaeological research or to meet and shake hands with. But a fictitious object can be famous, it can be a character from a particular story, and it can be impersonated by an actress. Therefore, the sentences

Maria Reicher

161

(5) Pegasus is a character from Greek mythology, (6) Miss Marple has been impersonated by Margaret Rutherford, and (7) Sherlock Holmes is more famous than Hercule Poirot, can be literally true. Note that some predicates, like “being a character from Greek mythology” and “having been impersonated by somebody”, that are applied to fictitious objects in discourse about fiction apply to abstract objects only. Others, like “is famous”, “came into being in the year 1899”, “is the subject of a doctoral thesis” apply both to abstract and to concrete objects. At any rate, “the abstractness theory” (as one might call it) is consistent with all external sentences about fictitious objects. However, there is a serious objection against the abstractness theory, namely: It works well for external sentences about fiction, but it seems that it does not work for most internal sentences. Consider again: (1) Pegasus is a flying horse. Horses (flying or not) are not abstract objects but living beings located in space. In other words, the predicate “is a flying horse” can be truthfully applied to physical objects only. An abstract object cannot be a horse, let alone a flying horse. Thus, the abstractness theory seems to fail, since it does not provide an explanation for the truth of internal sentences like (1). Similar considerations hold for (8) Hamlet hates his stepfather. To hate somebody is a mental property; and whatever the correct metaphysics of mental properties looks like, it is clear that abstract objects are not the kind of objects that can exemplify mental properties. There are two possible reactions to this: one might, in the light of this objection, reject the abstractness theory; or one might reject the assumption that internal sentences like (1) and (8) are true. I opt for the second alternative. At this point, one might object that the apparent truth of (1) was the starting point of this paper. I have argued that there is an intuitive difference between

162

Two Interpretations of “According to a Story”

(1) Pegasus is a flying horse, and (2) Pegasus is a flying dog. If the assumption that (1) is true is rejected, one has to find an alternative explanation for this intuitive difference. To explain this difference, one may use the story operator: (1a) According to a story, Pegasus is a flying horse, is true, while (2a) According to a story, Pegasus is a flying dog, is false. Generally speaking, the explanation is as follows: external sentences about fiction are literally true. Internal sentences, by contrast, are literally false. We may, however, interpret internal sentences as elliptical renderings of “according to a story” sentences. Indeed, this is the usual understanding of such sentences in discourse about fiction. Now the question arises: how shall we interpret the story operator in this context? Is it to be interpreted as a sentence operator or as a predicate operator? I opt for the second alternative. I would like to put forward two arguments for this proposal. The starting point of the first argument is the fact that (1ap) Pegasus/is according to a story a flying horse. implies that Pegasus exists. Furthermore, it implies that there is at least one entity that is according to a story a flying horse. By contrast, (1as) According to a story [Pegasus is a flying horse]. neither implies that Pegasus exists nor does it imply that there is an entity that is according to a story a flying horse. The question is: is it desirable that we can derive that Pegasus exists and that there is something that is according to a story a flying horse from our paraphrase of (1) Pegasus is a flying horse, or is it not desirable that we can derive these things? I would like to argue that a theory that allows us to do this is preferable to a theory that does not: intuitively, (1) is a sentence about

Maria Reicher

163

Pegasus, a sentence that is used to describe Pegasus in a certain way, to express that Pegasus has a certain property (if we want to talk in terms of properties)8. The predicate operator interpretation does full justice to this fact, while the sentence operator interpretation does not. To avoid a possible misunderstanding, it should be noted that the property that is ascribed (truthfully) to Pegasus is, of course, not the property of being a flying horse, but rather the property of being a flying horse according to a story. An abstract object cannot exemplify the property of being a flying horse, but it can exemplify the property of being a flying horse according to a story. Moreover, properties like being a flying horse according to a story or being a detective according to a story etc. can be exemplified by abstract objects only. The sentence (1ap) Pegasus/is according to a story a flying horse. is an external sentence about Pegasus. It is literally true that Pegasus has the property of being a flying horse according to a story. One may call sentences like this “quasi-internal sentences” in order to distinguish them from those external sentences that do not contain a story operator, and one may call the latter “purely external sentences”. Quasi-internal sentences play a crucial role for the description of fictitious objects. Intuitively, there is no reason to deny that quasi-internal sentences are “about” their subjects in exactly the same sense in which “purely external sentences” are about their subjects. For instance, intuitively (1ap) is a sentence about Pegasus in exactly the same sense in which (5) Pegasus/is a character from Greek mythology. is a sentence about Pegasus. Thus, if the rules of logic allow us to derive that Pegasus exists from (5) then we should also be allowed to derive that Pegasus exists from (1ap). Therefore, I opt for the predicatemodifier interpretation of the story operator. For a second argument to the same conclusion, consider the following two sentences: 8 For explanatory purposes, I put aside my reservations against an ontological

commitment to properties in this and the following section, though, in principle, I strive to avoid it (see note 11).

164

Two Interpretations of “According to a Story”

(9) Sherlock Holmes is a detective. (10) Hercule Poirot is a detective. Intuitively, from these two sentences, taken together, it follows: (11) There is something that Sherlock Holmes and Hercule Poirot have in common (... namely that each of them is a detective). In a half-formal way, this may be rendered as follows: (11')∃F (Sherlock Holmes is F & Hercule Poirot is F). The “F” in this sentence is a predicate variable, that is, a variable that stands for predicates, not for singular terms; here it stands for “is a detective”. Those who have animadversions against predicate variables may instead quantify over properties: (11'')∃x (Sherlock Holmes exemplifies x & Hercule Poirot exemplifies x). The variable “x” stands here for the abstract singular term “the property of being a detective”. As has been argued earlier, the sentences (9) Sherlock Holmes is a detective. (10) Hercule Poirot is a detective. are both false, if taken literally. Only their story-operator paraphrases are true. But the inference from these story-operator paraphrases to the sentence (11) There is something that Sherlock Holmes and Hercule Poirot have in common. is valid only if the story operator is interpreted as a predicate modifier. To see this, consider: (9as) According to a story [Sherlock Holmes is a detective]. (10as)According to a story [Hercule Poirot is a detective]. From these, we cannot derive (11as) According to a story [There is something that Sherlock Holmes and Hercule Poirot have in common], for this would be true only if there were a story in which both Sherlock Holmes and Hercule Poirot occur, which does not follow from the

Maria Reicher

165

claims that Sherlock Holmes is a detective and that Hercule Poirot is a detective. Consider, by contrast: (9ap) Sherlock Holmes/is according to a story a detective. (10ap) Hercule Poirot/is according to a story a detective. From these, we can derive (11') ∃F (Sherlock Holmes is F & Hercule Poirot is F), where the predicate variable “F” here stands for the predicate “is according to a story a detective”9. IV The basic point of the predicate-modifier interpretation of the story operator is that for discourse about fictitious objects two kinds of predicates are needed. Most realist ontologies of fictitious objects involve a modes-of-predication distinction, although these distinctions come in different varieties. Most modes-of-predication distinctions fall into one of two categories. I call them property-modifier theories and relation-modifier theories, respectively. Property-modifier theories and relation-modifier theories have two things in common: First, all of them use some sort of predicate modifier; and second, all of them involve quantification

9 At this point, my decision to use an indeterminate story operator instead of

determinate ones becomes relevant. If determinate story operators were used, the predicate in (9ap) would be different from the one in (10ap), since Holmes and Poirot originate in different stories. In this case, the paraphrases would be (9ap') “Sherlock Holmes/is according to the story s1 a detective” and (10ap') “Hercule Poirot/is according to the story s2 a detective”. These two sentences do not imply (11'). However, in principle the point of the argument can be made with determinate operators as well: One may interpret (9ap') and (10ap') as expressing relations between fictitious characters and stories, and this allows to derive (11''') “∃G (Ghs1 & Gps2)”, where “G” stands for the two-place predicate “_is a detective according to the story_”, “h” stands for “Holmes”, and “p” stands for “Poirot”.

166

Two Interpretations of “According to a Story”

over properties. The main difference between property-modifier theories and relation-modifier theories concerns the question of what exactly the predicate modifier is used to modify. Consider: (1b) Pegasus exemplifies the property of being a horse. A predicate modifier can be used to modify either the relation term “exemplify” or the abstract singular term “the property of being a horse”. The difference can be illustrated by means of the story operator: (1bpm) Pegasus/exemplifies according to a story/the property of being a horse. (1brm) Pegasus/exemplifies/the property of being a horse according to a story. The former sentence, (1bpm), expresses that Pegasus stands in a particular relation to the property of being a horse, namely in the relation of exemplifying according to a story. The latter sentence, (1brm), expresses that Pegasus stands in the usual exemplification relation to a particular property, namely to the property of being a horse according to a story. The standard version of the property-modifier theory in the literature involves a distinction between so-called nuclear and extranuclear properties. Thus, one might say: Pegasus has the nuclear property of being a horse. But: Pegasus has the extranuclear property of being a fictitious object. (See Findlay 1963, Meinong 1972, Parsons 1975 and 1980, Routley 1979 and 1980, Jacquette 1996.) The relation-modifier theory can be found in the literature in a number of varieties, among others in the following ones (for the sake of simplicity, I use the standard Pegasus-example for all cases): Pegasus is determined by the property of being a horse. But: Pegasus satisfies the property of being a fictitious object. (See Mally 1912.) The property of being a horse is ascribed to Pegasus. But: Pegasus has the property of being a fictitious object. (See van Inwagen 1977.)

Maria Reicher

167

Pegasus encodes the property of being a horse. But: Pegasus exemplifies the property of being a fictitious object. (See Zalta 1983 and 1988.) Structurally similar are Nicholas Wolterstorff’s distinction between having a property essentially within and having a property (see Wolterstorff 1980), Hector-Neri Castañeda’s distinction between consociation and consubstantiation (see Castañeda 1979 and 1990), Kit Fine’s distinction between the story-relative copula and the literal copula (see Fine 1982, 108f., and 1984) as well as Roman Ingarden’s distinction between having a property assigned and inherently containing a property10 (see Ingarden 1960, § 20). The common insight behind most versions of predicate-modifier theories is that discourse about fiction involves two kinds of assertions, “internal” and “external” ones (and even if some predicate-modifier theorists do not make this distinction explicitly, one nevertheless may use their theories in order to make this distinction explicit). All versions of predicate-modifier theories help to resolve certain paradoxes that would arise without them and would provide serious reasons to reject the ontological commitment to fictitious objects. The most common objection against all kinds of predicate-modifier theories is that their distinctions are neither intuitively clear nor defined. Often, the critics suspect that predicate-modifier theories are just a formal trick, consisting in the introduction of an artificial terminology that is applied on an ad hoc basis in order to avoid contradictions. Although this criticism is wrong, it is (at least in some cases) understandable that predicate-modifier theories arouse this impression. Terms like “nuclear” and “extranuclear”, “consubstantiation” and “consociation”, “encoding” and “being determined by” are either wholly artificial or used in a more or less unfamiliar way. Unfortunately, it seems to be impossible to provide definitions for them. It is not surprising, then, that many people are suspicious and rather try to “paraphrase away” the reference to fictitious objects instead of accepting an ontology that is based on mysterious distinctions. 10 Ingarden’s 1960 is written in German. The translations of the two technical

terms mentioned here are mine.

168

Two Interpretations of “According to a Story”

But most people, even those who claim that they are unable to understand terms like “encoding”, “consociation”, “being determined by”, and so forth, seem to have an intuitive understanding of the locution “according to a story”. Thus, the story operator may serve as a tool to introduce the modes-of-predication distinction in a way that is intuitively easily acceptable. V In this final section of the paper I shall show how the modes-of-predication distinction resolves (or better: prevents from arising) a number of paradoxes that would arise without this distinction. Paradox I: “The incompatibility paradox” The point of this paradox is that fictitious objects seem to have properties that are incompatible with their ontological status as abstract objects. Instances of this paradox have already been mentioned before, among others: 1. Pegasus is a horse. 2. Horses are located in space. 3. Thus, Pegasus is located in space. (1,2) 4. Pegasus is an abstract object. 5. Abstract objects are not located in space. 6. Thus, Pegasus is not located in space. (4,5) The solution is clear: Premise 1. is false. However, it is true that 1'. Pegasus/is according to a story a horse.11 From this and the premise that horses are located in space, it does not follow, however, that Pegasus is located in space. Therefore, the paradox does not arise.

11 Note that this sort of predicate modification, in contrast to the ones mentioned

in the foregoing section, does not involve reference to or quantification over properties.

Maria Reicher

169

Paradox II: “The incompleteness paradox” The point of this paradox is that fictitious objects seem to violate the principle of excluded middle, since a story never determines all of the properties of a fictitious object. Here is an instance of this paradox: 1. Either “Pegasus’ eyes are blue” or “It is not the case that Pegasus’ eyes are blue” is true. (This is an instance of the principle of excluded middle). 2. “Pegasus’ eyes are blue” is not true (since there is no passage in the relevant story that implies that Pegasus’ eyes are blue). 3. “It is not the case that Pegasus’ eyes are blue” is not true (since there is no passage in the relevant story that implies that Pegasus’ eyes are not blue). 4. Thus, neither “Pegasus’ eyes are blue” nor “It is not the case that Pegasus’ eyes are blue” is true. (2,3) Solution: The third premise is false. It is indeed true that it is not the case that Pegasus’ eyes are blue, because Pegasus is an abstract object, and abstract objects don’t have eyes. However, the following two sentences are true: 2'. “Pegasus’ eyes are according to a story blue” is not true. 3'. “Pegasus’ eyes are according to a story not blue” is not true. But since the latter sentence is not the negation of the former, this does not violate the principle of excluded middle. Paradox III: “The contradiction paradox” The point of this paradox is that it seems that fictitious objects may have contradictory properties and thus may violate the law of contradiction. Imagine a fictional story that implies the following two sentences: (12) John once owned a red car. (13) It is not the case that John once owned a red car. (Suppose that the author simply made a mistake, or that she wanted to test the reader’s attentiveness.)

170

Two Interpretations of “According to a Story”

The solution is: The second sentence is true. It is indeed not the case that John ever owned a red car, since the fictitious object John is an abstract object, and abstract objects are not the kind of things that can own cars. However, the following two sentences are both true: (12ap) John/according to a story once owned a red car. (13ap) John/according to a story never owned a red car. But since the latter sentence is not the negation of the former, this does not violate the law of contradiction. Summary and conclusion: Story operators are ambiguous. They may be interpreted either as sentence operators or as predicate modifiers. As sentence operators, they may be used to block the inference from internal predications about fictitious objects to existential sentences about fictitious objects. However, external sentences about fictitious objects cannot be paraphrased by means of story operators; and thus story operators as sentence operators do not rid us from the ontological commitment to fictitious objects, if we take discourse about fiction seriously. I opt for a realist ontology of fictitious objects. If the story operator is used to modify “internal predicates” of fictitious objects, the assumption that fictitious objects exist can be consistent with logical as well as with basic metaphysical principles and with empirical facts*. References Barbero, C., (2005), Madame Bovary: Something Like a Melody, Milan, Edizioni Albo Versorio. Castañeda, H.-N., (1979), ‘Fiction and Reality: Their Fundamental Connections. An Essay on the Ontology of Total Experience’, Poetics, 8, pp. 31–62. –– (1990), ‘Fiction, Perception, and Forms of Predication (Reply to Künne)’, in (eds.) K. Jacobi, H. Pape, Thinking and the Structure of the World. HectorNeri Castañeda’s Epistemic Ontology Presented and Criticised, Berlin, de Gruyter, pp. 268–284.

* The work on this paper was supported by the Austrian Fonds zur Förderung der

wissenschaftlichen Forschung (project P16981-G03).

Maria Reicher

171

Findlay, J.N., (1963), Meinong’s Theory of Objects and Values, London, Oxford University Press. Fine, K., (1982), ‘The Problem of Non-existents. I. Internalism’, Topoi, 1, pp. 97–140. –– (1984), ‘Critical Review of Parsons’ Nonexistent Objects’, Philosophical Studies, 45, pp. 94–142. Hintikka, J., (1959), ‘Existential Presuppositions and Existential Commitments’, Journal of Philosophy, 56, pp. 125–137. Ingarden, R., (1960), Das literarische Kunstwerk, Tübingen, Niemeyer. [English translation, The Literary Work of Art, Evanston, Northwestern University Press, 1973.] van Inwagen, P., (1977), ‘Creatures of Fiction’, American Philosophical Quarterly, 14, no. 4, pp. 299–308. Jacquette, D., (1996), Meinongian Logic. The Semantics of Existence and Nonexistence, Perspectives in Analytical Philosophy, 11, Berlin–New York, de Gruyter. Künne, W., (1990), ‘Perception, Fiction, and Elliptical Speech’, (eds.) K. Jacobi, H. Pape, Thinking and the Structure of the World. Hector-Neri Castañeda’s Epistemic Ontology Presented and Criticised, Berlin, de Gruyter, pp. 259–267. –– (1995), ‘Fiktion ohne fiktive Gegenstände: Prolegomenon zu einer Fregeanischen Theorie der Fiktion’, in (eds.) J. Brandl, A. Hieke, P. Simons, Metaphysik. Neue Zugänge zu alten Fragen, Sankt Augustin, Academia, pp. 141–161. [Reprint in (ed.) M.E. Reicher, Fiktion, Wahrheit, Wirklichkeit. Philosophische Grundlagen der Literaturtheorie, Paderborn, Mentis, 2006.] Lambert, K., (1983), Meinong and the Principle of Independence. Its Place in Meinong’s Theory of Objects and Its Significance in Contemporary Philosophical Logic, Cambridge, Cambridge University Press. Leblanc, H. and Hailperin, T., (1959), ‘Nondesignating Singular Terms’, Philosophical Review, 68, pp. 239–243. Leonard, H.S., (1956), ‘The Logic of Existence’, Philosophical Studies, 7, pp. 49–64. Mally, E., (1912), ‘Gegenstandstheoretische Grundlagen der Logik und Logistik’, Zeitschrift für Philosophie und philosophische Kritik, 148, Ergänzungsheft. Leipzig. Meinong, A., (1972), Über Möglichkeit und Wahrscheinlichkeit, in (eds.) R. Haller, R. Kindinger, Alexius Meinong Gesamtausgabe VI. Graz, Akademische Druck- und Verlagsanstalt. [Originally published in 1915.] Parsons, T., (1975), ‘A Meinongian Analysis of Fictional Objects’, Grazer Philosophische Studien, 1, pp. 73–86. –– (1980), Nonexistent Objects, New Haven, Yale University Press. Rescher, N. (1959), ‘On the Logic of Existence and Denotation’, Philosophical Review, 68, pp. 157–188.

172

Two Interpretations of “According to a Story”

Routley, R., (1979), ‘The Semantical Structure of Fictional Discourse’, Poetics, 8, pp. 3–30. –– (1980), Exploring Meinong’s Jungle and Beyond. An Investigation of Noneism and the Theory of Items, Canberra. Ryle, G., (1933), ‘Symposium: Imaginary Objects’, Proceedings of the Aristotelian Society, Supp. vol. XII, pp. 18–43. Thomasson, A.L., (1999), Fiction and Metaphysics, Cambridge, Cambridge University Press. Wolterstorff, N., (1980), Works and Worlds of Art, Oxford, Clarendon. Zalta, E.N., (1983), Abstract Objects. An Introduction to Axiomatic Metaphysics, Dordrecht, Reidel. –– (1988), Intensional Logic and the Metaphysics of Intentionality, Cambridge, MIT Press.

Madame Bovary as a Higher-Order Object* Carola Barbero 1. What fictional entities are Fictional entities are complex entities corresponding, as object-correlates, to sets of properties. The set of properties constituting them is made up of internal and external properties (and the distinction is based on a difference of attribution), with corresponding ways of givenness. The relation subsisting between the set of properties and the corresponding fictional object (object-correlate) is not a relation of identity. If it were, only what is explicitly said by the author in the novel would be ascribable to a fictional entity, but we all know that we often make inferences from a text such as the following: (a) Madame Bovary is a naïve and stupid woman. What is asserted in (a) is absolutely true, even if Flaubert never says anything like it. But, if so, then it cannot be true that what a fictional entity is, is identical with the set of properties assembled by the author and corresponding to it. A fictional entity is more than the corresponding set of properties: actually it is an object of higher order1. Nevertheless this not only means that the whole is more than the sum of its parts (as the Gestalt saying claims), but also that we can preserve the fictional entity even when varying its components, and this is because the fictional entity as a whole depends generically on its constituents, and it remains one and the same fictional entity only when its form (i.e. structure) has not changed. A fictional entity like Madame Bovary is an object of higher order made up of both internal and external properties. We speak about internal and external properties, even if this may sound somehow misleading, because what we want to underline is the element of creation, the

* I would like to thank M. Ferraris, F. Kroon, K. Mulligan, M. Reicher, P. van

Inwagen, A. Varzi and A. Voltolini for helpful comments on and objections to a previous version of this work. 1 Following A. Meinong (1899), I define ‘higher order object’ an object built on other objects, which therefore are its basis, though it is not reducible to them.

174

Madame Bovary as a Higher-Order Object

origin of the properties’ attribution. Let’s try to give more details about this terminological choice. As is well known, the most interesting neoMeinongian theories are on the one hand those maintaining a double mode of predication, one for abstract objects and another for concrete ones, and on the other hand those that make use of two different kinds of properties, constitutive and non-constitutive ones. Rapaport (1978) is a supporter of the first strategy, whereas Parsons (1980) of the second one. By introducing a distinction between internal and external properties, it seems that what we mean to hold is a two-property position; nevertheless, since we use just one kind of property, because we maintain that what change simply are the ways these properties are attributed, we have to admit that the theory standing here in the background properly is the two-modes-of-predication one (the modes of attribution could be interpreted as a sort of specification of predication). It is to underline this element of attribution that we speak about internal (the properties attributed by the author) and external properties (attributed by readers and critics). This distinction makes it possible to explain how an object may have incompatible properties: an entity like Madame Bovary, for instance, has at the same time – even if not in the same way – the property of being concrete like a woman (as far as its internal properties are concerned) and abstract like a law (as far as its external properties are concerned), remaining nonetheless one and a same object. We introduce two different modes of attribution in order to underline the element of creation present in fictional entities. Therefore internal and external properties should not be interpreted as ‘properties according to the story’ and ‘properties in the external world, real properties’, they should on the contrary be seen on the one hand as the properties used by the author to build Madame Bovary as a human being, and on the other hand as the properties readers and critics used to recognize and identify it as an abstract object. It is clearly not a problem, from such a point of view, if both author and critics attribute a same property. For example, if we had a novel starting ‘once upon a time there was a little abstract object…’ in this case the property of ‘being abstract’ would be both an internal (since the author of the novel has attributed it to its creature) and an external property (because we, as readers, identify the fictional character as an abstract object). Nevertheless it is important to remember that in both cases we have

Carola Barbero

175

one and only one kind of property, simply predicated or attributed from different sources. 2. What is ‘Madame Bovary’? By ‘Madame Bovary’ I mean a complex object, and one and the same object in these different cases: in a literary comment on the book and on the character whose name is ‘Madame Bovary’2 and in a description of Emma waiting for Charles the day after their engagement3. There is one and only one object, simply having different properties. Fictional entities are composed of a collection of internal properties ascribed by the author to the entity within the novel and by the combination of external properties attributed to it by readers and critics. Fictional entities are, as far as their internal properties are concerned, arbitrary entities4: Flaubert could, in fact, have changed Madame Bovary’s (internal) properties. For instance, while writing the 2 “Madame Bovary is utterly modern. It’s the first sex and shopping novel, after

all. Emma Bovary, desperately seeking satisfaction and fulfilment, employs retail therapy as an escape from an unhappy marriage, experiments with adultery, is obsessed with money and with material objects as substitutes for happiness. She’s convinced it’s men who’ve got it all: power and freedom and choice and ease of movement. She can’t buy herself the male privileges she envies, so she spends money on blue glass vases and new curtains instead”. Preface by Michèle Roberts to G. Flaubert, Madame Bovary, Penguin Classics, 2003, p. VII. 3 “Next day, by nine o’ clock, he was at the farm. Emma blushed when he came in, though she gave a forced little laugh, to keep herself in countenance. Père Rouault embraced his future son-in-law. All discussion of money matter was postponed; there was, after all, plenty of time, since the marriage could not decently take place before the end of Charles’ period of mourning, that is to say, in the spring of the following year”. G. Flaubert, Madame Bovary, Penguin Classics, 2003, part one, ch. 3, p. 24. 4 On arbitrary objects and on their ontological status see K. Fine (1985), In order to give arbitrary objects an ontological status, Fine introduces two kinds of predication: “Two questions may be discerned here: are there any arbitrary objects?; and what are they like? To deal with the first, it is necessary to distinguish, in a way that is familiar from the philosophical literature, between two uses of the phrase ‘there are’. […] In one use of the phrase, he is concerned to deny that there are numbers; for that is just his position. But in the other use, he may be prepared to admit that there are numbers; […]” pp. 6-7.

176

Madame Bovary as a Higher-Order Object

novel, he could have ascribed to her the property of being a blonde or a fat woman, or that of having a son instead of a daughter: being the creator of this entity, Flaubert could indeed have decided nearly anything about it. It is nonetheless important to remember that this ascription is made possible by language, and this is clearly not arbitrary. Fictional entities can nevertheless legitimately be considered as arbitrary objects only during their creation, during the ascription process undertaken by the author who can arbitrarily decide how to shape them. But once the novel is finished, the fictional entity is no longer arbitrary: in fact it generically depends on the set of properties combined by the author in order to create it. Actually once the author has finished writing and has brought the novel to a close, then – in readers and critics – a hypostatising use of language5 takes place and, it is thanks to this process that the novel itself and its characters become real objects of our world, individuated as abstract objects. Afterwards we leave behind this hypostatising use and we turn back to the novel, focusing on the characterizing use6, i.e. the author’s use of language to build characters. Exactly at this point we must acknowledge that fictional entities are no longer arbitrary: the set of internal properties and the object of higher order arising from them, to which Madame Bovary corresponds, form its (her) structure and we cannot change it any longer, at the risk of losing the final product of Flaubert’s work of imagination. Madame Bovary could therefore have been different from what it is, if Flaubert’s pretending use of language7 – i.e. the author’s use of language while writing the novel – had been different, attributing different properties to that entity (or the same properties but shaped in a different way) from those it actually has.

On the contrary, F. Nef (1998), pp. 95-7, maintains that fictional entities are not arbitrary entities while, for instance, scientific or theoretical objects are. Arbitrary entities are general entities according to him, whereas fictional entities are individual ones. For a theory of fictional entities as general objects see M. Santambrogio (1992). 5 S. Schiffer (1996). 6 This particular use of language is not from Schiffer, as in the previous case, but it is an idea of my own. 7 S. Schiffer (1996).

Carola Barbero

177

Of course we could be interested, for instance, in moving Madame Bovary from Flaubert’s novel and bringing it to another novel we wanted to create8. In that case we would borrow a character from Flaubert’s novel and we would use it in our own, attributing new properties to it. Nevertheless the character Flaubert created as Madame Bovary is fixed once and for all in its structure from the beginning of the hypostatising use of language on. We could consequently define fictional entities as arbitrary in terms of genesis, but as objectively fixed as far as their structure is concerned. In the wide philosophical debate on fictional entities, focused on establishing what they are and what they aren’t, there is certainly a point that very few people would deny, and that is that, for instance: • Madame Bovary is a fictional entity •

Madame Bovary is the main feminine character in Flaubert’s novel of that title

•

Madame Bovary is the symbol of human dissatisfaction and desperation.

And what we mean by saying this is that there is a book, the book Madame Bovary, and that this book concerns a fictional entity, which is the symbol of human dissatisfaction and desperation and so on. 8 This is exactly what for instance – continuing to use Madame Bovary as an

example – W. Allen does in his “The Kugelmass Episode”, in Side Effects, 1986, Ballantine Books.

178

Madame Bovary as a Higher-Order Object

Madame Bovary therefore certainly has (it is true that she/it has) those external properties deriving from the hypostatising use of language. Nevertheless, what it is now important is to answer the following question: did Flaubert create a woman? 3. What did Flaubert create? Flaubert created a woman with beautiful eyes which “though they were brown, they seemed to be black because of the lashes, and they met your gaze openly, with an artless candour […] Round her neck was a white collar, turned down. Her black hair […] was parted in a delicate central line that traced the curve of the skull; and just revealing the tip of the ear, …”9.

Then, and this is absolutely important, Flaubert created a woman10, and this same woman, is a character (we have said: is the main feminine character) and a symbol (symbol of human dissatisfaction and desperation). We could therefore define Madame Bovary as an object created by an author and accepted by readers and critics. We are willing to admit that a fictional entity is created by an author. There is nevertheless an important difference between creating a statue 9 G. Flaubert, Madame Bovary, p. 15, Part One, Ch. 2. 10 Flaubert attributed to Madame Bovary the property of being a woman and, in

fact, Madame Bovary undoubtedly has that property. The author makes it true that a character has a certain property by creating the character with this property. K. Fine (1982) maintains a similar view of creation in fiction saying that fictional entities “[…] come into being as the result of [the appropriate activity of the author], in much the same way as a table comes into being as the result of the activity of a carpenter”. He says that fictional entities are created and that, as far as fictional entities are concerned (but not as far as any artifact is concerned), to

Carola Barbero

179

(Canova, Italic Venus) and creating a fictional literary entity (Flaubert, Madame Bovary). In fact, whereas in the former case the author creates something that exists ‘out there in the world’, in the latter case the creation does not exist like you and me, but in a different, more abstract way: it exists in books, in stories. To create a fictional entity actually means to give it a positive way of givenness (to make it be, somehow) thanks to an act of stipulation (this characteristic of fictional entities allows them to be legitimately included in the wider domain of social objects11). 4. Fictional entities are dependent entities Fictional entities generically depend on their constitutive elements (i.e. the corresponding sets of properties) and on the relations subsisting among them. This can be defined as generical structural dependence. This means that a fictional entity – once it has been accepted in our ontology thanks to the hypostatising use of language – depends on the characterising use of language of the author, i.e. it depends on an author’s specific use of language. If Flaubert, for instance, had characterised Madame Bovary as the object-correlate corresponding only to that specific set of properties: {having a stupid husband, being beautiful, having lovers}, then these, and the relations subsisting among them, would be the constituents of Madame Bovary. create does not mean to bring into existence, but to bring into Being. Therefore he introduces a double mode of predication: to be and to exist. A. L. Thomasson also investigates the creation process of fictional entities, underlining the similarities with the creation of other artifacts such as tables and chairs, tools and machines, all requiring creation by intelligent beings. She nonetheless also points out a crucial difference between fictional entities and all other artifacts: “[…] the way in which fictional characters are created does make them strange, for although one cannot simply create a table, toaster or automobile by describing such an object, fictional characters are created merely with words that posit them as being a certain way” (Thomasson 1999, p. 12). 11 J. Searle (1995) has noticed the similarity between cultural entities and institutional entities, in their both being brought into existence simply by being represented as existing. His favourite example is money: “‘This note is legal tender for all debts public and private’. But that representation is now, at least in part, a declaration: It creates the institutional status by representing it as existing. It does not represent some prelinguistic natural phenomenon” (p. 74).

180

Madame Bovary as a Higher-Order Object

It is important here to distinguish between two different kinds of explanations, one concerning the genesis and the other concerning the description of fictional entities considered as structures. On the one hand the genesis requires a two-level theory: the basis is made up of sets of properties and the higher-order object is the complex emerging from them (the sets of both internal and external properties have a structure on a second level). On the other hand the description concerns the fictional entity as a whole, once the novel is finished. Now the fictional entity directly is a higher-order object, i.e. a whole whose parts are themselves determined as being such that they can exist only as parts of a whole of this given kind. Clearly what is at stake with higher-order objects is, as Husserl remarked in his Logical Investigations, especially in the Third, a problem of unity. Let’s consider these dots:

Where does this kind of unity come from? As is well known, Husserl maintained that unity can come about in two distinct ways. We can have: • a whole not needing any additional object (e.g. the dots of the drawing, and here it is apparent that they do not need any additional object in order to fit together to make a unified whole); •

a whole which needs an additional object (e.g. two pieces of wood that need to be joined together, i.e. they are not in themselves sufficient to make a unity but can be unified only given the presence of some additional object). Fictional entities are wholes of this second kind, and in fact they need some unifying objects, because the properties constituting the set

Carola Barbero

181

are not in themselves sufficient to make a unity but can be unified only given the presence of some additional object. Unifying objects may be, still according to Husserl, of two sorts: on the one hand they may be independent objects (like glue), capable of existing independently from the whole of the given sort; on the other hand they may be dependent objects, capable of existing only in consort with the objects they serve to unify. The structure Madame Bovary consists in seems to be a unifying object of the second kind, but only generically and not rigidly (the structure needs some of the properties of the corresponding sets, even if not exactly those properties at the basis of the original structure) dependent on its original properties. What is fundamental for Madame Bovary is not only to be built on some structured properties, but also to be branded by its author, Flaubert, who assembled the properties making it possible for the structure to emerge. Yet, as we have already noticed, this dependence on the properties and the relations subsisting among them is not rigid, but generic. Madame Bovary in fact rigidly depends on its structure, and only generically on its properties and on the relations subsisting among them, in the same way as a melody depends rigidly on its shape and generically on its pitch, on its notes and on its intervals. In the same way as melody is a succession of notes, but is nevertheless not reducible to a succession of notes, so a fictional character such as Madame Bovary, in order to remain the same character, has to preserve its structure and its form, but not its exact constitutive properties12. Fictional entities as higher-order objects are defined by two formal criteria: (1) a fictional entity is a structure and therefore cannot be understood as the mere sum of its constituting properties (i.e. fictional entities are not additive agglomerates); (2) a fictional entity is a transposable structure, where ‘to transpose’ means ‘to preserve the structure independently of its original constitutive elements’. This second criterion is of course important only when we want to understand the modalities of transposition of a character, asking if, for

12 Even if its original structure, i.e. the structure originally shaped by Flaubert to

create Madame Bovary, clearly depends exactly on the constitutive elements we can find in the novel of that title.

182

Madame Bovary as a Higher-Order Object

instance, Allen’s Bovary in “The Kugelmass Episode” is the same as Flaubert’s. The character will be identified as the same if and only if the structure of the character has been preserved. Fictional entities depend genetically on the existence of a specific author for the attribution of internal properties, and on the existence of readers and critics for the ascription of external properties. The author is obviously necessary for their creation: it is through the pretending use of language first, that is subsequently revealed to be a characterising use, that fictional entities are created. This reflects a very common intuition according to which a fictional entity is the author’s creature; the author is like a father to it, and this is why in a world in which Gustave Flaubert had never existed, we can reasonably maintain that Madame Bovary would not have existed either. On the other hand, we have to consider the dependence of fictional entities on readers and critics: first because in an illiterate world Madame Bovary would not even have been a fictional entity and secondly because it is thanks to their practices of reading and criticizing (hypostatising use of language) that the fictional entity in question first came to be included in our ontological domain. Properties and sets of properties are structured parts, those parts constituting the structure the fictional entity consists in. But how are we then to find fictional entities, for instance in the work of literature we have to hand? Let’s explain it by following these points: 1- the sequence of words must have a sense for the reader; 2-

there must be some properties genuinely characteristic of the fictional entity x;

3-

the fictional entity must be a relatively self-contained whole and therefore not part of any other continuation or sets of properties;

4-

the fictional entity must be non-decomposable: its parts (the properties or sets of properties) must be dependent entities, not themselves capable of existing as separate fictional entities in their own right. Fictional entities therefore are structures, and the structure has to be intelligible, where ‘intelligible’ here means not only that the structure

Carola Barbero

183

has to be recognized as such, but also that the dominant, leading characteristics of the fictional entities need to be individuated. This approach to fictional entities is a sort of resolution between holism and elementarism: we have fictional entities as wholes, but these entities are constituted of particular properties, particular sets of properties, and the wholes as such are, when considered from a genetical point of view, the result of a production, a creation of an author. Melodies too are structures – or rather they are structures par excellence – first because the sequence of notes that is a melody possesses a certain whole-property; and secondly because each single note (and segment) possesses its own part-property (role, function, tonic, dominant, leading note, cadence, trill, etc.). Fictional entities seen as structures present whole- and part-properties in this same sense. 5. Fictional entities are higher-order objects Fictional entities are higher order objects, and this means that the properties or the set of properties used by the author to characterize them do not exhaust the whole essence of these entities. Indeed, fictional entities are not merely identical to their corresponding sets of properties; they are something more and different, therefore not reducible to a relation of identity with the corresponding set. The best way to explain why fictional entities can be defined as higher order objects is by the analogy between fictional entities and melodies. What happens with melodies? I can play a familiar melody of eight tones, then afterwards I can employ eight new tones, and yet you will still recognize the melody despite the change. How is this possible? Is there something that is more than the mere sum of eight tones, i.e. a ninth something, which is the form-quality, the Gestaltqualität13, of the original eight? This ninth factor could in fact be identified with the element which enabled you to recognize the melody despite the fact that it had been transposed. Or we could maintain that, in addition to the eight tones, what we have are intervals (which are relations) and 13 This term was used for the first time by C. von Ehrenfels (1890); its importance

is mainly terminological: he gave a name to a common phenomenon. For a historical overview of objects of higher order, or Gestalten, see also A. Meinong (1891), (1899), and E. Husserl (1891), (1900-1901).

184

Madame Bovary as a Higher-Order Object

these are what remain constant through the transpositions. We could assume not only elements, as we did in the previous proposal, but also relations between these elements as additional components of the total complex14. Or, again, we could propose the thesis that to this total of eight or more tones there come certain higher process which operate upon the given material to produce unity. All these considerations may analogously be appropriate for fictional entities during their creation, while the author is writing the novel. We have properties, sets of properties, and we have to understand what kind of relation subsists between these sets of properties and the fictional object – e.g. Madame Bovary – itself. The set of properties is not one and the same as its object-correlate: what then is the difference between Madame Bovary and its constitutive properties? It is the very same kind of difference subsisting between the eight tones and the melody. We therefore have to suppose (1) that for fictional characters there is also something like a form-quality, a shape; or (2) that what makes the difference between properties and character are the relations subsisting among the different properties in the set; or, (3) finally, that there needs to be a higher process operating upon the set of properties to produce unity. What happens while the author is writing the novel can clearly be explained using one of these keys. This mirrors however only what happens when the author is writing, i.e. how a set of different properties can be transformed into Madame Bovary or Anna Karenina. This is nevertheless solely the genetic side of the question (i.e. how, from a set of properties, Madame Bovary emerges). We have now to face the structural side of the matter. How, once the novel is finished, does the character appear as a whole? Clearly we do not make a sum of properties while reading the novel. Here again we have the analogy with melodies. The radical question raised by the Gestalt theorists was the following: is it really true that when I hear a melody I have a sum of individual tones which constitute the primary foundation of my experience? Is it not perhaps the reverse of this that is true? In fact, what I really have, i.e. what I really hear, are not simply the individual notes, actually what I hear is a part which is itself 14As far as melody is specifically concerned, this second option nevertheless fails

to account for the phenomenon, because in some cases we still have the same melody even if its original relations have been altered.

Carola Barbero

185

determined by the nature of the whole. What is given to me by the melody does not arise as a secondary process from the sum of the pieces as such. On the contrary, what happens in each single part of the melody already depends upon what the whole is. Obviously the task, for Gestalt theorists, was (and is) to investigate these whole-conditions and discover what influences they exert upon experience. Let us now see how this works with fictional entities. Once created by their author and once the novel is finished, fictional entities can correctly be considered to be wholes. This is why we find the characterising properties of these entities when reading, though we already individuate them as parts, aspects of the whole character which is, somehow, already there. The form or shape characteristic of the fictional entity can therefore, exactly as in the case of melody, be transposed to a different story, even with different constitutive properties15, and still remain the same entity, as happens, for instance, to Madame Bovary in Allen’s story16, or, to give a different example, to Maigret’s wife, Louise, who writes her autobiography after her husband’s death, in 15 Madame Bovary could have been different, but nevertheless still remained the

same object of higher order if, for instance, Flaubert had attributed to her the property of having a son instead of a daughter, or if he had endowed her with different properties of this sort. In that case the shape would clearly have been preserved. And what if Flaubert had attributed to her the property of being a rabbit? In that case, of course, we would always have an entity named ‘Madame Bovary’, but it would be a different one, only having the same name the previous object had (because the object of higher order would clearly not be the same). On problems of this sort, see A. L. Thomasson (1999), pp. 56-69. 16 One could clearly simply maintain that, in order to recognize a fictional entity as that particular fictional entity, what we need is just a name: Madame Bovary is that entity (woman, character, x) whose name is ‘Madame Bovary’. For instance in “The Kugelmass Episode”, the main character (Professor Kugelmass) is literally thrown into the novel written by Flaubert and consequently it is apparent why the woman he meets there certainly is Madame Bovary. Nonetheless, maintaining that the fictional entity and its name are one and a same thing, how could we explain the case of Faust? We find much of this same character, with different names, in different stories from different authors, and therefore it can not be true that a character is one and a same thing with its name. We have Goethe’s Faust and we have Marlowe’s Doctor Faustus, which clearly are different characters. Yet, although they are different characters, they are not totally different characters. We could in fact easily suppose that there is a sort of Faust-nucleus (which

186

Madame Bovary as a Higher-Order Object

Barbara Notaro Dietrich’s My Husband Maigret17. Louise Maigret still remains the same character she was in Simenon’s novels, even if, of course, one that is much richer in properties: in Simenon’s works Louise was merely Maigret’s wife, whereas in Notaro Dietrich’s she is the main character and tells the reader everything about her life with the famous French detective inspector, clearly from her own point of view. Therefore Madame Bovary can legitimately be considered as a higher-order object. Its structure is made up of Madame Bovary’s leading properties (i.e. the most representative among the properties constituting the sets at its basis) with a specific whole function. For instance, Madame Bovary’s structure could be: ‘being married to a failed doctor, having some lovers, being the symbol of women’s dissatisfaction and unfaithfulness and committing suicide’. That this could be seen as its structure is evident from the fact that if we read these characteristics somewhere else, then we immediately think of Madame Bovary. These elements give important details about the way the properties constituting the structure work. The most important among both internal and external properties are those properties constituting the structure. For a complex to be a structure means to have certain special characteristics; the complex must possess certain structural qualities, which are not supernumerary entities existing alongside or above the separable fundamenta. The structural qualities are intrinsic to the structure and are special kinds of whole-properties, whereas structures are special kinds of wholes. is already an object of higher order, a shape) they both include, for instance the object-correlate of the set containing the property of having made a pact with the devil. This Faust-nucleus is the object of higher order upon which both Goethe’s and Marlowe’s Fausts are based. Here we have something like a character (Faustnucleus) included in another character (Goethe’s and Marlowe’s), and this explains why even if we are not willing to accept that Goethe’s and Marlowe’s are exactly the same object, we wish to know why they are not totally different or why they can legitimately be considered as different interpretations of a common object. In this case we have an object of higher order (Goethe’s and Marlowe’s) built on another object of higher order: both Faust and Doctor Faustus are characters founded on the basic Faust-nucleus, which is itself a character. On the various problems raised by Faust, see A. Thomasson (1999), p. 59. 17 B. Notaro Dietrich, Mio marito Maigret, Edizioni e/o, 2004.

Carola Barbero

187

6. Open problems Our proposal is to see Madame Bovary as the object-correlate of both internal and external properties, and therefore to see it/her as a higher order object. Nevertheless, in order to maintain that fictional entities are higher order objects, we should somehow be able to make explicit which are the Gestalt Laws in the case of fictional entities. As is well known, for perceptual higher order objects this is almost an easy task: there are in fact specific laws governing perceptual grouping, as for instance: Similarity: Objects with similar properties (e.g. shape, color) Proximity: Nearby objects Good Continuation: Objects that define smooth lines or curves Symmetry: Objects that form symmetrical patterns Periodicity: Objects that form periodic patterns These same laws happen to apply in spatial perception, even if not always in temporal ones. The problem for us now clearly is: are we able to find laws governing fictional entities’ identity? How can the fictional entity survive despite changes in its internal properties? How is this compatible with the idea that between these properties there subsists a sort of internal relation (i.e. what specifically makes it possible for the object to be more than a mere aggregate of the set of properties whose object-correlate it is)? Answering these questions is important also in order to understand how we are to consider what is involved in our apparent grasp of complex formations such as fictional entities. What can account for the unity they seem to involve? What do we need to apprehend fictional entities as such? Clearly, we can grasp this structure only if we are somehow able, by a cumulative process involving the operations of memory, to unify everything in one intellectual glance, and a discursive process of this sort is indispensable if we are to grasp a structure which needs, as its basis, a succession of elements. This is an important aspect of the matter. In fact consider the situation where someone is able to read, for instance the book Madame Bovary, as a succession of words, but is not able to unify them in one intellectual glance and then have characters, plot and so on, in that case we would not say that this person has understood what kind of object

188

Madame Bovary as a Higher-Order Object

Madame Bovary is. Let us recall the conditions that must be satisfied for a sequence of words to possess a particular structure and therefore to be a fictional entity. Such a sequence must have a sense, there must be some characteristic properties and it must be a relatively self-contained and non-decomposable whole. Nevertheless, whether we view fictional entities as complexes of elements (as mere sets of properties), or as elements bound together by a unifying relational moment, or as a whole within which elements and relations are distinguishable only abstractly, is in the end a matter of convenience18. Some sorts of examples will call for one description, some for another, in the light of given contexts and purposes. This should not, though, be understood as implying that there is no truth of the matter in any given case, as though the structures of fictional entities could vary in reflection of the way in which the theorist chooses to describe them. Rather, each of the given alternatives is merely a different way of formally articulating strictly identical material. Consider for example the distinction between two descriptions of a ballet, one of which lists the separate dancers and describes the total effect of their dancing together following the given choreography, the other of which lists the steps, jumps and pirouettes and describes how each contributes to the total effect. These different descriptions analyse different aspects of one and the same entity, but they do not change what that entity is. 7. Conclusions We do not claim to have provided the called-for ontological clarification of fictional entities as higher-order objects (or structures) here. There is a lack of ontological clarity about many of the notions we have employed. Probably if the appropriate ontological clarifications of structures were available, then the present proposal could be stronger and notions like those we have used would play a significant role in inquiries into fictional entities and the like. But we are not sure that such clarifications can be provided. Besides these problems concerning the concept of structure or of higher-order object, there is a specific problem concerning artistic objects such as fictional entities and melodies. Is it really possible to 18 On the relevance of different ontological and metaphysical perspectives and

their connections with convenience, see A. Varzi (2001), (2002).

Carola Barbero

189

define the shape or the structure of an artistic object such as Madame Bovary? Is it really enough, and not a sort of reductive move, to sketch Madame Bovary’s structure as ‘being married to a failed doctor, having some lovers, being the symbol of women’s dissatisfaction and unfaithfulness and committing suicide’? It is almost true that if we read these characteristics somewhere else than in Flaubert’s novel, then we immediately think of Madame Bovary, but is this a sort of sufficient condition to be ‘Madame Bovary’? Can we find examples of good shapes or structures in other literary cases? Let’s consider Anna Karenina by Lev Tolstoi. What is Anna’s shape? We could say ‘being passionate, being abandoned, being the symbol of women’s love, desperation and mistakes, and falling under a train’, because these are the most important among both its internal and external properties, and therefore constitute the shape. But how are we to decide which are the most important among internal and external properties? How are we to define or draw any fictional entity’s shape? The question is still open. References Allen W., (1986), “The Kugelmass Episode”, in Side Effects, New York, Ballantine Books. Ehrenfels von C., (1890), “Über Gestaltqualitäten”, Vierteljahrsschrift für wissenschaftliche Philosophie, 14, pp. 242-92. Fine K., (1982), “The problem of non-existents”, Topoi, 1, pp. 97-140. Flaubert, G., Madame Bovary, London, Penguin Classics, 2003. Husserl E., (1891), Philosophie der Aritmetik. Psychologische und logische Untersuchungen I, Halle, Husserliana, XXI ed. I. Strohmeyer, 1983. –– (1900-1901), Logische Untersuchungen, Halle, Husserliana, XIX ed. U. Panzer, 1984. Meinong A., (1891), “Zur Psychologie der Komplexionen und Relationen”, Zeitschrift für Psychologie und Physiologie der Sinnesorgane, 2, pp. 245265. –– (1899), Über Gegenstände höherer Ordnung und deren Verhältnis zur inneren Wahrnehmung; English translation in On Objects of Higher Order and Husserl’s Phenomenology, (ed.) M.-L. Schubert Kalsi, The HagueBoston-London, M. Nijhoff, 1978. Notaro Dietrich B., (2004), Mio marito Maigret, Rome, e/o Edizioni. Parsons T., (1980), Nonexistent Objects, New Haven, Yale University Press. Rapaport W.J., (1978), “Meinongian Theories and a Russellian Paradox”, Noûs, 12, pp. 153-180.

190

Madame Bovary as a Higher-Order Object

Santambrogio, M., (1992), Forma e oggetto, Milan, Il Saggiatore. Schiffer S., (1996), “Language-Created Language-Independent Entities”, Philosophical Topics, 24, 1, pp. 149-167. Searle J. R., (1995), The Construction of Social Reality, New York, The Free Press. Thomasson A. L., (1999), Fiction and Metaphysics, Cambridge, Cambridge University Press. Varzi A. C., (2001), Parole, oggetti, eventi, Rome, Carocci. –– (2002), “Words and Objects” in A. Bottani et al. (eds.), Individuals, Essence, and Identity. Themes of Analytic Metaphysics, Dordrecht, Kluwer, pp. 4975.

Identity across Time and Stories Francesco Orilia 1. Introduction There are sentences that assert that a certain object that exists at a given time t is identical with another object that exists at another time t'. For example, let us suppose that at time t' Tom finishes painting with red his only car, which had been white up to t, and sells it to Mary. It is legitimate in this case to assert: (1) the white car owned by Tom at time t is the red car owned by Mary at time t'. In making this assertion, we seem to presuppose that the very same ordinary object “survives” in time, in some objective sense, despite the change in its properties. In current terminology, this can be expressed by saying that an object1 persists in time, where this persistence can be understood in two ways, as either an endurance or a perdurance, as we shall see. I shall call sentences like (1) sentences of diachronic identity (or of reidentification across time, or the like). The truth of many such sentences is presupposed in much, perhaps most, of what we say about ordinary objects. For instance, when we say that Botticelli’s Primavera was a dingy painting (at time t), before restoration (completed at a time t'), we presuppose that Botticelli’s Primavera before restoration (at time t) is Botticelli’s Primavera after restoration (at time t'). And examples of this kind could be multiplied ad nauseam. There are also sentences which are very similar in structure to diachronic reidentification sentences, but which seem to assert not so much that an object “survives” in time but that a fictional character “survives”, as we “move” from a story to another. Here is an example, which intuitively is about Ulysses: (2) The Achaean who devised the trick with the Trojan horse, according to the Odyssey, is the Achaean who guided the expedition past the Pillars of Hercules, according to Dante’s Inferno. 1 Unless otherwise indicated, the word “object” is used here and in the following

to talk about ordinary objects such as tables, chairs, stones, trees, cats, people.

192

Identity across Time and Stories

Let us call such sentences sentences of intertextual identity or of reidentification across stories or the like. Clearly, there is a structural similarity between these two kinds of sentence, which suggests an Analogy Constraint (AC), as we may call it. AC demands, as far as possible, a similar ontological analysis for both diachronic and intertextual reidentification sentences. Ontological analysis is, in the tradition of Russell 1905, a paraphrase and/or translation in a canonical formal language, which aims at depicting the truth conditions of philosophically notable sentences in a way that perspicuously presents their ontological commitments. I shall use ambiguously “ontological analysis” (or “analysis”, for brevity's sake) to indicate any such paraphrase or translation, the proposition expressed by it, or even the process of providing such a paraphrase or translation. The context will disambiguate, when important. Note that an ontological analysis need not be what we may call a semantic analysis. This is an analysis that aims at showing, as in Montague grammar, how the various meanings of the constituents of a sentence are compositionally combined so as to constitute the meaning of the sentence. Arguably, however, the legitimacy of an ontological analysis depends in part on how plausibly it can be connected to a semantic analysis. There will be an opportunity below to clarify this distinction and the relevance of the latter for the former, by means of a particular example. Clearly, ontological analysis is a tool for ontology (or metaphysics)2. But one may ask: which ontology? Indeed, as is well known, Strawson 1959 distinguished descriptive and prescriptive metaphysics. The former aims at revealing something about our conceptual scheme and the latter something about very general features of the real world. I take it for granted that ontological analysis is a tool for descriptive ontology. But I also think that it is relevant for prescriptive ontology. For after all our conceptual scheme serves us rather well in our survival in a hostile world and therefore we have default reasons to think that some part of our conceptual scheme may embed a (by and large) correct theory about the world (Castañeda 1980). These reasons can however be overcome by more empirical and scientific considerations and thus the extent to which ontological analysis reveals something about the world 2 For present purposes, I make no significant distinction between ontology and

metaphysics.

Francesco Orilia

193

and conversely the extent to which it must be superseded and/or supplemented by empirical/scientific investigation must be decided on a case by case basis. Here, in the light of AC, I want to pursue a parallel ontological analysis of both diachronic and intertextual reidentification sentences. I shall then briefly consider what we can learn from this analysis from the standpoint of prescriptive metaphysics. 2. Some preliminaries In the light of what we saw above, we can assume that certain sentences of philosophical interest have a superficial grammatical form, to which one can associate a formula and/or paraphrase constituting a corresponding analysis (with an explanation of how the symbols in the formula or paraphrase are to be understood). For example, sentences of identity through time are superficially of the form (ITT) The F at time t is the G at time t'. Moreover, sentences of identity across stories are of this superficial form: (ITS) The F in story s is the G in story s'. (“In story s”, “according to story s” and the like will be regarded as synonymous.) These sentences crucially involve definite descriptions. They may of course also involve proper names and (at least in the case of diachronic reidentification) indexicals and demonstratives. I argued elsewhere (Orilia 2000a, 2003), however, that, contrary to widespread referentialist opinions, by combining semantic and pragmatic considerations, such singular terms can ultimately be viewed as definite descriptions. Thus, roughly, we can view proper names such as “Ulysses” and “Socrates” as, respectively, “the fictional character called Ulysses” and “the philosopher called Socrates” and for present purposes I shall assume this in the following. And I shall also take for granted that definite descriptions should be understood à la Russell, in the sense that, roughly, a sentence of form (DD) The P is Q should be given the following ontological analysis: (DDa) ∃x (∀y (Px → x = y) & Qy),

194

Identity across Time and Stories

which we abbreviate as (DDb) ∃1x (Px & Qy). In saying this, I mean simply that the use of the definite article in (DD) commits us to just one entity that exemplifies property P, where this entity could be abstract or concrete, particular or universal, etc. Thus, in taking advantage of analysis (DDa), I use variables of quantification as in principle unrestricted, although I may use letters (such as “t”, “s” or “E”) that are meant to suggest certain classes of entities as domains of intended values (e.g., times, stories and properties that function, in a sense that we shall see, as essences). In particular, I am not committed to the Russellian idea that definite descriptions are incomplete symbols which cannot be assigned a meaning when taken apart from a sentential context. Following Montague 1974 and Cocchiarella 1982, I take them to mean properties (of properties), which I call definite denoting concepts, and which can be represented by lambda abstracts. For example, “the P” can be taken to mean the definite denoting concept [λf ∃1x (Px & fy)], which I abbreviate as [THE P]. Consequently, (DD) should be given an “intermediate” analysis such as (DDc) [THE P]Q, wherein the denoting concept [THE P] is predicated of the property Q. (DDc) is, by λ-conversion3, logically equivalent to the “final” Russellian analysis (DDa). This intermediate analysis can be considered a semantic one, for it provides a meaning for the constituent “the P” (i.e., [THE P]), another meaning for the constituent “Q” (i.e., Q) and tells us that their syntactic concatenation in (DD) corresponds to the composition of such meanings by means of the predication relation (in such a way that [THE P] is the predicated entity and Q the argument 3 The logical principle of λ-conversion asserts, for the monadic case, that [λx A]d

↔ Ax/d, where Ax/d is the formula that results from A by replacing each free occurrence of x in A with d (provided d is free for x in A). For more details on λconversion and the use of λ-abstracts in representing the meaning of definite descriptions and other noun phrases, cf. Partee et al. 1990. As I see it, λ-conversion and λ-abstracts should be deployed in the context of a type-free logic that avoids Russell's paradox and related problems without type restrictions at the grammatical level (Orilia 2000, 2003).

Francesco Orilia

195

of the predication). The ontological analysis (DDa) follows from (DDc) by laws of logic, which shows that in this case the ontological analysis is well-connected to the semantic one. Although the emphasis here is on ontology rather than on semantics (and thus I shall ignore analysis (DDc) as far as possible), I dwell on this point here, because there will be reasons to appeal to definite denoting concepts. With these reasons in mind, let me point out that it is appropriate to say that a definite description “the P” (singularly) refers to an entity x (or denotes x, if you wish), when the meaning, [THE P], of the description is such that it determines x, that is, it is such that P is univocally exemplified by x4. 3. The endurantist analysis Turning again to reidentification across time, the most natural option is of course to read sentences of form (ITT) at face value, which means that the “is” connecting the definite descriptions should be seen as expressing identity, an identity which links to itself one single object that happens to exemplify F with respect to a time t, and G with respect to another time t'. By following this line, we come to this ontological analysis: (ITTa) ∃1x (at-t(x exemplifies F) & ∃1y (at-t'(x exemplifies G) & x = y)5. This analysis can aptly be called endurantist. For, from its standpoint, our conceptual scheme appears to be committed to endurantism, according to which objects are three-dimensional continuants that “move” through time while remaining self-identical, i.e., endure in time, in spite of changing their properties. This view is naturally associated with presentism6, according to which to exist is to presently 4 P is univocally exemplified by x when ∃1y(Py & x = y). 5 Here we treat “at-t(x exemplifies F)” and similar expressions as the combination

of “at-t”, understood as a modal operator, and a sentence (possibly open). We could alternatively write “at(t, [x exemplifies F])”, where “at” is a predicate that connects a term t and a nominalised sentence purporting to denote a proposition (similar considerations apply to (ITSb), below). We need not be fussy about these distinctions for present purposes, but, as I see it, whatever format one chooses, we give expression to relations between a time and the proposition that x exemplifies property F. 6 See, for instance, Loux 1998, ch. 6. The thesis that endurantism and presentism

196

Identity across Time and Stories

exist. It is compatible with presentism to admit that there are (now) past and future times, for times (values of variables t, t', etc.) can be understood as abstract entities at which propositions can be true or false (where truth simpliciter is truth at the present moment)7. And it is also compatible with this doctrine that there is an ordinary object x that existed at some past time and/or will exist at some future time, as long as x also exists at the present moment. But it is not compatible with it to say that x fails to exist at the present moment, although x existed or will exist at some past or future moment or that x exists eternally8, rather than at a specific (present) time. That is, as we may put it, it is not compatible with presentism that there is an entity that fails to exemplify all existence-entailing properties9 of ordinary objects at the present moment (it is neither a chair, nor a rock, nor has it a certain color or shape and so on and so forth), while it exemplifies some such properties at some past or future time, i.e., ∃x∃t∃t' (x exists at t & x does not exist at t' and t is past or future and t' is present). Moreover, it is not compatible with presentism that that there is an entity (a “time slice”, as we shall see) that has existence-entailing properties of ordinary objects, but that has such properties eternally rather than at a specific time. In the light of AC, the endurantist analysis (ITTa) invites us to a similarly endurantist analysis for sentences of form (ITS), which views them as saying that one single fictional character is F with respect to a given story and G with respect to another, thereby “enduring” from one story to another just as an ordinary object endures from a time to another: (ITSa) ∃1x (according-to-story-s (x exemplifies F)) & ∃1y (accordingto-story-s'(x exemplifies F)) & x = y).

are mutually independent has been defended (cf. Loux 1998, pp. 207-209, and Varzi 2001, pp. 122-123), but has not gained much favour. 7 In Prior’s presentism, times ultimately are propositions (cf. Prior and Fine 1977), but other presentists reject this reduction (Craig 2000, p. 213). 8 Following current usage in discussion of these topics, I use “eternally” as synonymous with “atemporally”, although, more precisely, “eternally” should mean “at all times”. 9 In the terminology of Cocchiarella 1982.

Francesco Orilia

197

Accordingly, (2) receives the following analysis: (2a) ∃1x ((according-to-the-Odyssey (x exemplifies Achaean who devised the trick with the Trojan horse)) & ∃1y (according-tothe-Inferno (y exemplifies Achaean who guided the expedition past the Pillars of Hercules)) & x = y). 4. The perdurantist analysis There are however considerations that militate against the correctness of endurantism and the idea that our conceptual scheme is committed to it. It has been argued in fact that relativity theory suggests a rather different ontological picture, perdurantism, a doctrine that is however also defended on more a priori grounds, independently of scientific considerations10. According to this view, reality is four-dimensional in that it is extended in time as well as in space. Consequently, objects themselves are four-dimensional and thus perdure through time, inasmuch as they are constituted by “temporal parts” or “slices”, ordered by temporal relations such as after or before, and objectively linked to each other by what has been called genidentity11. For example, the Eiffel Tower (qua four-dimensional ordinary object) is constituted by a sequence of (three-dimensional) temporal parts or slices that starts when the Eiffel Tower was built, goes on to encompass various slices that come after the Eiffel Tower at birth, e.g., the Eiffel Tower at the present moment, and ends with a slice that comes after all other members of the series, i.e., the Eiffel Tower at the last moment of its existence. The idea is that all these parts or slices, though numerically different, are (objectively) genidentical. On this view, a temporal part that is not in the present, say the original manuscript of the Phaedo at a past time t, exists (eternally) just as much as those that are in the present, for being present is a relative notion, which depends on the choice of an observer. Similarly, New York cannot be declared nonexistent, just because we focus on an observer situated in Rome. Perdurantism thus implies that there are entities, time slices, that exemplify existence-entailing properties of ordinary objects but not at a time, let alone the present time, but 10 See, e.g. Quine 1960, § 36, and Loux 1998, ch. 6. 11 See for example Chisholm 1971, note 9, p. 17.

198

Identity across Time and Stories

eternally. Clearly, then, according to perdurantism, presentism is false and its opposite, eternalism, is true (Loux 1988, pp. 208-209). Given AC, Perdurantism suggests a different, perdurantist, analysis for both kinds of reidentification sentences: (ITTb) ∃1x (x-at-time-t exemplifies F) & ∃1y (y-at-time-t'exemplifies G) & and x = y)12. (ITSb) ∃1x (x-according-to-story-s exemplifies F) & ∃1y (y-accordingto-story-s exemplifies G) & x = y)). In particular, (2) comes out as follows: (2b) ∃1x ((x-according-to-the-Odyssey exemplifies Achaean who devised the trick with the Trojan horse) & ∃1y (y-accordingto-the-Inferno exemplifies Achaean who guided the expedition past the Pillars of Hercules) & x = y). From this perspective, we should say that (2b) is true in virtue of the fact that there are two “character slices”, x-according-to-the-Odyssey and y-according-to-the-Inferno, which are fictionally genidentical, i.e., are connected by a relation that links different “parts” of fictional characters that appear in different stories. Thus, the fictional Ulysses would be made up of several parts, the Ulysses of the Iliad, the one of the Odissey, the one of the Inferno, and so on. Castañeda (1974, 1979) is rather close to viewing fictional characters in this way, when he postulates a relation of “transconsociation” that can link “guises” occurring in different stories. 5. The Generalised Russellian Robust Sense of Reality According to the well-known “Russellian Robust Sense of Reality” (RRSR), (merely) fictional characters such as Ulysses and Pinocchio (and more generally “nonexistent objects”) do not really exist and thus they must be “paraphrased away” somehow. RRSR seems embedded 12 “x-at-time-t” (and similar expressions) can be understood as singular terms that

result from two terms x and t, as combined by a special operator “-at-time-”, and denote a certain time slice of x occurring at t, provided x is an object and t a time. Mutatis mutandis, “x-according-to-story-s”, as used in (ITSb) below, must be understood analogously.

Francesco Orilia

199

in our conceptual scheme, for we all agree, in ordinary non-philosophical discourse, on claims such as (3) (the fictional character called) Ulysses does not exist. Yet, both the endurantist and the perdurantist analyses generate a conflict with RRSR. For example, it seems appropriate to identify the fictional Ulysses with the Achaean who devised the trick with the Trojan horse, according to the Odyssey, as well as with the Achaean who guided the expedition past the Pillars of Hercules, according to Dante’s Inferno. Hence, once we accept (ITSa) and thus (2a), we should also accept: (ITS2a) ∃1z (z is fictional and called Ulysses & (∃1x (according-tothe-Odyssey (x exemplifies Achaean who devised the trick with the Trojan) & ∃1y (according-to-the-Inferno (y exemplifies Achaean who guided the expedition past the Pillars of Hercules) & x = y & z = x & z = y). From which it follows that, for some property F, ∃1z(z is fictional and called Ulysses & Fx), which can naturally be taken to contradict (3) and thus RRSR. In sum, we end up accepting that Ulysses exists, in contrast with (3). Mutatis mutandis, we get to the same conclusion from (ITSb). As a matter of fact, Castañeda (1974, 1979) has appealed to reidentification sentences precisely as one of the arguments that can be used to question RRSR and promote a Meinongian ontology wherein, by a being/existence distinction, there are fictional characters that do not exist. In addition to RRSR, there is also what we may call the “Generalised Russellian Robust Sense of Reality” (GRRSR),13 according to which past and future objects do not exist. GRRSR seems part of our conceptual scheme just like RRSR, for we make claims such as (4) Socrates does not exist (any longer). Yet, by an argument analogous to the one we just saw for fictional characters and endurantism, it can be shown that the perdurantist analysis is at odds with GRRSR. It is doubtful that the endurantist and perdurantist analyses can be reconciled with RRSR and GRRSR, respectively, which suggests that 13 Not endorsed by Russell, who was rather consistently and explicitly a perdurantist.

200

Identity across Time and Stories

we should look for an alternative to both. This search must not forget that GRRSR is in tension with what we may call the Intentionality Datum (ID), especially appealed to by Castañeda and other contemporary philosophers hostile to RRSR14. ID tells us that we seem to talk and think about such items as fictional characters and past or future objects, as shown, e.g., by our acceptance of (5) Van Inwagen admires Ulysses; (6) Van Inwagen admires Socrates. One should try to resolve this conflict, because, by a principle of charity, we should not assume, as far as possible, that our conceptual scheme is incoherent. Perhaps, an alternative account of reidentification sentences may indicate a way to do this without sacrificing GRRSR. But in order to see how to shape this alternative, we must consider other reasons against the endurantist and perdurantist analyses. 6. Search for neutrality Once we consider the perdurantist option, it is problematic to claim that sentences of identity through time must be interpreted according to the endurantist analysis, for this seems to imply that our conceptual scheme is committed to endurantism (and thus to presentism). But this is not the case, for we are somehow able coherently to entertain perdurantism (and eternalism with it) or even to accept it (say, once we become convinced by relativity theory). It is of no avail to say that we had a conceptual scheme committed to endurantism and to some sort of pre-relativistic physics and that we changed it in favour of a different scheme committed to relativity theory and perdurantism. For we seem to be able to reason coherently from a vantage point from within which we can entertain and evaluate the two options in play and decide one way or another. It is this vantage point of view which deserves to be called our conceptual scheme. And given the possibility of switching from one option to the other, that is, from one theory 14 They may be neo-Meinongians, such as Parsons and Zalta, or philosophers who

reject RRSR on behalf of ficta, without viewing them as Meinongian nonexistent objects (Van Inwagen 1977, Thomasson 1999; Voltolini forthcoming seems to take a middle road).

Francesco Orilia

201

to the other, this scheme appears to be committed to neither. This suggests that neither the endurantist nor the perdurantist ontological analyses are the right ones15, if an ontological analysis is meant to tell us something about our conceptual scheme. Perhaps we need a more neutral analysis. 7. Theseus’ ship The need for an alternative analysis can be supported by further considerations. In the situation described above, (1) is obviously true. But there are problematic cases where it is not clear whether and, if so, in what sense it is legitimate to assert sentences of this kind. The famous story of Theseus’ ship probably provides the most well-known example of such a problematic case. According to Plutarch16, the ship, preserved by the Athenians, starts undergoing at a certain time, t, a step by step replacement of its planks, which culminates at a later time, t', in a ship with none of the original parts. To make things more complicated, Hobbes asks us to imagine that some man17, call him Reassemblarius, gathers the old planks and puts them together at t', so that we have at that point two ships, the ship preserved by the Athenians at time t' and the ship owned by Reassemblarius at time t'. Accordingly, the following classical problem arises: is Theseus’ ship at time t identical to the ship preserved by the Athenians at time t', or to Reassemblarius’ ship at time t'? As is well known, there are reasons that militate in favour of both options18 and thus we should accept: (7) The ship owned by Theseus at time t is the ship preserved by the Athenians at time t'; (8)

The ship owned by Theseus at time t is the ship owned by Reassemblarius at time t'. But clearly

15 Quine’s (1969) doctrine of ontological relativity leads to a similar conclusion,

but in the light of dubious behavioristic assumptions. 16 Plutarch’s Lives, Theseus, 23, I. 17 De Corpore, XI, 7. 18 See for example Coburn 1971, p. 93, and Swartz 1991, ch. 11.

202

Identity across Time and Stories

(9)

The ship preserved by the Athenians at time t' is not the ship owned by Reassemblarius at time t', despite the fact that (7) and (8) seem to imply the opposite of (9). Prima facie, then, our conceptual scheme is incoherent. Now, in contrast with the above-mentioned principle of charity, the endurantist analysis must subscribe this conclusion. For such an analysis leads us to accept these counterparts of (7)-(9): (7a) ∃1x (at-t(x exemplifies ship owned by Theseus) & ∃1y (at-t'(y exemplifies ship preserved by the Athenians) & x = y); (8a)

∃1x (at-t(x exemplifies ship owned by Theseus) & ∃1y (at-t'(y exemplifies ship owned by Reassemblarius) & x = y).

∃1x (at-t'(x exemplifies ship preserved by the Athenians) & ∃1y (at-t'(y exemplifies ship owned by Reassemblarius) & x ≠ y). Yet, it follows from (7a) and (8a), by the laws of first-order logic with identity, that (9a') ∃1x (at-t'(x exemplifies ship preserved by the Athenians) & ∃1y (at-t'(y exemplifies ship owned by Reassemblarius) & x = y). But (9a) and (9a') are in direct conflict and thus, given the endurantist analysis, we are trapped in an inconsistency. The perdurantist analysis, however, does not fare much better. It could perhaps question the fact that (7)-(9) are jointly true, but only under the unreasonable assumption that an ordinary object can undergo fission by having one temporal part genidentical with two distinct temporal parts that occur at the same time. To the extent that genidentity is taken to be natural and objective, perhaps something like that can be accepted, e.g., for an amoeba that splits by a biological process. In this case, there are exactly the same (natural, objective) reasons to claim, for each new amoeba, that it is genidentical with the original amoeba. But in the ship case, the reasons for (7) and those for (8) are (as we shall see) of very different kinds, which suggests that they cannot support the subsistence of one and the same natural and objective genidentity. Once (7)-(9) are taken for granted, as they should be, the perdurantist should analyse them as: (9a)

Francesco Orilia

203

(7b)

∃1x (x-at-t exemplifies owned by Theseus) & ∃1y (y-at-t' exemplifies preserved by the Athenians) & x = y);

(8b)

∃1x (x-at-t exemplifies owned by Theseus) & ∃1 (y-at-t' exemplifies owned by Reassemblarius) & x = y);

∃1x (x-at-t' exemplifies preserved by the Athenians) & ∃1y (yat-t' exemplifies owned by Reassemblarius) & x ≠ y). And thus again we get an inconsistent triad, for from (7b) and (8b) we can formally derive: (9b') ∃1x (x-at-t' exemplifies preserved by the Athenians) & ∃1y (yat-t' exemplifies owned by Reassemblarius) & x = y). By the principle of charity mentioned above, we must look for an alternative analysis of reidentification sentences, one that does not generate an inconsistency such as that inherent in the triads (7a)-(9a) and (7b)(9b). In searching for this alternative, we may note that the joint acceptability of (7)-(9) suggests the existence of a hidden parameter, a reidentification criterion, somehow linked to such sentences. The idea is that there are various such criteria and that (i) we implicitly appeal to them in asserting or presupposing sentences of identity across time to describe what goes on in situations spanning a certain amount of time; (ii) they are coherent with each other when these situations are typical, as when we assert (1); (iii) they pull in opposite directions in peculiar situations, such as that recounted by Plutarch and Hobbes. We are thus in the dark as to the truth value to be assigned to reidentification sentences relating to these situations, if we try to assign one, independently of the hidden parameter. (9b)

8. Sequentialism Chisholm 1970 comes to a similar conclusion. In his view, two ordinary objects x and y (other than persons), which exist at two different times, can be considered as identical only in a “loose and popular sense”, in the terminology of Bishop Butler, which Chisholm reproposes. As I understand him, he presupposes a four-dimensional view of reality, within which two temporal slices x-at-t and y-at-t' can be viewed as if they were genidentical (and thus as if they contributed to constitute an object extended in time as well as in space) in virtue of a

204

Identity across Time and Stories

criterion of reidentification of a conventional nature. Accordingly, it is only because we have conventionally chosen some such reidentification criterion that sentences such as (1), (7) or (8) are true, understood as asserting that two temporal parts belong in one object extended in time. And Chisholm (1970, p. 28) suggests that even a property like having such and such sailing schedule can be used to provide a reidentification criterion. That is, Chisholm 1970 avoids the inconsistency I have noted by appealing to reidentification criteria, just as I have suggested. However, he does more than that. He proposes an ontological picture, which (following Varzi 2001) we may call sequentialism, distinct from both endurantism and perdurantism. He presents it in a fourdimensionalist version that makes it closer to perdurantism, but we can also have a three-dimensionalist version of it, closer to endurantism, which in fact is preferred in Chisholm 1971. According to the former version, there are temporal parts ordered by temporal relations like before and after, but there is no objective relation of genidentity that relates some of these parts in such a way that they come to constitute one ordinary object that perdures in time. We can however speak of some sequences of these temporal parts as if they were objects perduring in time, given the choice of a reidentification criterion. According to the latter version, at each time new ordinary objects supersede those of the previous time (which have ceased to exist) in a continuous process of birth and death (due, say, to the different ways in which the indivisible particles that constitute ordinary objects are assembled). There are then no objects that (objectively) persist in time (except perhaps in some cases and/or for brief intervals of time). Nevertheless, once we presuppose a reidentification criterion, we can speak as if some objects endure in time even though there are in fact no such persisting objects. Although four-dimensional sequentialism does not recognize the existence of an objective relation of genidentity, it is still committed to a world consisting of a four-dimensional manifold in line with relativity theory. But we have already noted that it would be wrong to conceive of our conceptual scheme as committed to this. Accordingly, it would be wrong to embed four-dimensional sequentialism in a semantic analysis of temporal reidentification sentences that is meant to reveal something about our conceptual scheme. On the other hand, threedimensional sequentialism follows from mereological essentialism

Francesco Orilia

205

(which in fact Chisholm accepts; see his 1989), according to which an ordinary object x that loses whatever part cannot be strictly identical to the object y that results from this mereological change. It may be added that this form of sequentialism also appears to follow from the bundle theory, according to which an ordinary object at a given time t is a bundle of properties, [P1, ..., Pn]. For in this view, if an object [P1, ..., Pn] changes a property at time t, say P1, it is superseded at a successive moment t' by a different object, say [F, P2, ..., Pn]. Moreover, it seems to me that three-dimensional sequentialism can hardly be motivated apart from one of these views. At the same time, it is far from clear that our conceptual scheme is committed to them. For arguably it offers a vantage point from within which different philosophers can rationally dispute in favour of or against such doctrines. In sum, I think that the idea of reidentification criteria should be exploited to provide an analysis of reidentification sentences which is neutral not only with respect to endurantism and perdurantism, but with respect to sequentialism (in either version) as well. Accordingly, I shall first provide an account of reidentification criteria compatible with this goal and then proceed to propose one such analysis. 9. Conventional essences Call sequentialist proposition for a time t the proposition, P, that holds at time t and provides the most complete information about which properties and relations of ordinary objects are exemplified at t as well as about the numbers of individuals that exemplify such properties at t (and similarly about relations)19. The intuitive idea is that the sequentialist propositions for the various times provide all the information about what happens at the various times, provided this information is compatible with sequentialism, whether in its three-dimensionalist or four-dimensionalist version. For example, the truth at t of sequentialist proposition P for t may imply that a certain property Q is univocally exemplified at t and the truth at t' of the sequentialist proposition P' for t' that Q is also univocally exemplified at t', without implying that there is one enduring or perduring object that exemplifies Q both at t and at t'. More generally, the sequentialist propositions for the various times, 19 There may be various logically equivalent propositions of this kind. In which

case, assume we pick one somehow.

206

Identity across Time and Stories

taken jointly, imply nothing about whether or not time is a fourth dimension or whether there are objects that objectively persist in time. A reidentification criterion, C, can be viewed as an ideal procedure, which our conceptual scheme somehow presupposes in our use of natural language and in our reasoning practices. Given as input a time t and a property F, exemplified by n distinct objects in t, C yields as output, on the basis of the sequentialist proposition for t and for all (relevant) subsequent or following times, properties E1,...., En, corresponding to the n objects exemplifying F in t20. We assume that at t each Ei in the output is univocally exemplified and is coexemplified with F (i.e., it holds at t that ∃1x(Ei x & Fx). Intuitively, each Ei should be considered the essence of one of the objects exemplifying F at t. In fact, the idea is that each property in the output should allow us to view reality as if one object persisted in time as long as this property is exemplified (whether or not such a persistence actually takes place in an objective sense). Let us write “C(t, F, E)” to mean that given as inputs time t and property F, C provides E in its output. Moreover, let us say that a property is an essence according to criterion C, if, for some F and t, C(t, F, E). It might be the case that C(t, F, E) is false for any choice of t and F, although it might have been true for some t and F, had the sequentialist information about t and the other relevant times been different. In such a circumstance, let us call E a merely possible essence, according to criterion C. In contrast, a property E such that, for some t and F, C(t, F, E) actually holds may be called a real essence, according to criterion C. This distinction should prove useful in dealing with counterfactuals, but I shall not pursue that topic here. More pressing for present purposes is the distinction between real and fictional essences. We shall see that a criterion C can also take as input a property and a story (rather than a time). The properties provided as outputs in this case are what I call fictional essences, according to criterion C. For reasons we hinted at and that I shall further clarify, I assume that there are different but equally legitimate reidentification criteria. In connection with this point, we may want to call a property which is an 20 For reasons having to do with relations and to which we shall return, C may

also take in input a sequence of n times and an n-adic relation and provide as output sequences of n properties.

Francesco Orilia

207

essence according to a certain criterion a conventional essence. The fact that I am postulating conventional essences makes my view rather different from Plantinga’s (1974) essentialism, according to which each object has with metaphysical objective necessity, so to speak, one essence distinguishing the object from anything else. It would be very moot to claim that this doctrine is embedded in our conceptual scheme (it is widely disputed by many respectable philosophers) and thus as long as we are doing descriptive ontology, we should be neutral with respect to it, just as with respect to endurantism, perdurantism and sequentialism. Conventional essentialism, as we could call my view, does indeed have this neutrality. Here are some examples of temporal reidentification criteria and corresponding essences. We can envisage, to begin with, a criterion, CSP, which privileges spatio-temporal continuity, as this notion is understood along the lines, e.g., of Hirsh 1971 or Coburn 1971. In its output, CSP typically provides properties such as being spatio-temporally continuous with place p at time t, where a place is here conceived as an abstract volume of space, and, intuitively at t in p there occurs the birth of an object. Roughly, such a property is exemplified by an object at a time t', if at t' this object is in a place p' such that there is a continuous spatio-temporal path connecting the time-place pairs and , and, for any two contiguous time-place pairs in the path, there is an appropriate “qualitative similarity” of the two pairs21. Another criterion, CP, privileges what we may call parts continuity (see, e.g., Swartz 1991, p. 348). This is based on the intuition that (usually) objects preserve in time most of their parts22. In its output, CP typically provides properties like having most of the parts such that at time t' 21 A pair is qualitatively similar to another pair , if at t there is in

place p an object exemplifying exactly the properties in the set of properties S; at t' there is in place p' an object exemplifying exactly the properties in the set of properties S' and S and S' are very similar in that if P is a member of S, then P or a property qualitatively similar to P is in S' (e.g., weighing 10 pounds is in S and the same property or the qualitatively similar weighing 10+δ pounds is in S'). This criterion is not circular, as is often claimed, if places are viewed as abstract volumes of space existing apart from ordinary objects, rather than as themselves ordinary objects in need of being reidentified. 22 Or at least most of their parts result from a gradual change dictated by microphysical and/or biological laws.

208

Identity across Time and Stories

and place p there was an object constituted by such parts (or microphysical/biological “ancestors” thereof). Again, intuitively, at t' in p there occurs the birth of a certain object. Yet another criterion, CFC, privileges functional continuity. It is in play, it would seem, when we assert something like “the lizard’s tail fell off, but then it grew back”, which appears to presuppose a reidentification sentence such as “the lizard’s tail at time t (before the fall) is the lizard’s tail at time t' (after the fall and the re-growth)”. As I see it, reidentification criteria differ in that they tend to privilege one of these notions (spatiotemporal continuity, parts continuity, etc.) over the others, when a conflict among them arises, but otherwise they try to combine these various aspects. Each of these notions provide, we may say, a partial reidentification criterion and the various reidentification criteria that I am postulating differ in how they combine and rank these partial criteria (which in themselves are inherently defeasible; Swartz 1991, p. 349). 10. Diachronic identity and real essences Once we grant reidentification criteria and corresponding conventional essences, we can provide an ontological analysis of reidentification sentences of form (ITT), which is ontologically neutral in the sense explained above: (ITT_M) ∃1E (C(t, F, E) & ∃1E' (C '(t', G, E') & E = E'), where C and C ' are typically (though not always) the same reidentification criterion. According to this analysis, (ITT) says that, upon presupposing reidentification criteria C and C', there is just one essence which is assigned to F at t by C (which implies that F is univocally exemplified at t) and to G at t' by C' (which implies that G is univocally exemplified at t'). Intuitively (when C and C' are the same), this essence corresponds to one object that persists as we “move” from t to t'. Now, by presupposing the criterion based on spatio-temporal continuity, CSP, (7) can be paraphrased in the light of this analysis as follows: (7c) ∃1P (CSP(t, ship owned by Theseus, P)) & ∃1Q (CSP(t', ship preserved by the Athenians, Q)) & P = Q).

Francesco Orilia

209

Similarly, by presupposing the criterion based on part continuity CP, (8) can be paraphrased as follows: (8c) ∃1P (CP(t, ship owned by Theseus, P)) & ∃1Q (CP(t', ship owned by Reassemblarius, Q)) & P = Q). Clearly, (7c) and (8c) can be simultaneously true, in line with our tendency to see good arguments in favour of both (7) and (8). Moreover, (7c) and (8c) are compatible with whatever analysis of (9) is appropriate in the present perspective, e.g., (9c) ∃1P (CSP(t', ship preserved by the Athenians, P)) & ∃1Q (CP(t', ship owned by Reassemblarius, Q)) & P ≠ Q). This way of proceeding brings with it some controversial assumptions at the semantic level. It proposes in fact that a predicate for ordinary objects can be used in natural language in a “metaessential” way (relative to a time) to say that the property expressed by the predicate is linked, so to speak, to a certain essence, on the basis of a presupposed reidentification criterion. At the same time, in this approach it is not denied that the very same predicate can also be used in a “standard” way to say that at a given time the property in question is exemplified. It may then be worth underlining that these complications could be seen as worth paying, since they prevent us from conceiving of natural language, and the conceptual scheme behind it, as necessarily linked to ontological presuppositions such as those we have seen above, which can hardly be attributed to the typical speaker. On the other hand, the attribution to a typical speaker of implicit reidentification criteria (which may vary with the context) is justified by the examination of the way in which we use sentences of diachronic reidentification and in particular by the puzzles raised by controversial examples, such as that of Theseus’ ship. And once we admit these reidentification criteria, the acceptance of conventional essences and the double use of predicates that we just noted naturally goes with them. 11. Intertextual identity and fictional essences Given the Analogy Constraint AC, we should at this juncture provide a similar treatment of sentences of form (ITS). This can be done by assuming that a reidentification criterion C may take in input a property

210

Identity across Time and Stories

and a story in order to yield as output one or more fictional essences. If s is a story, we should assume the following, by analogy with what we said about times: when F is exemplified by n distinct objects in the story, then there should be n distinct fictional essences E1,...,En such that C(s, F, E1), ...,C(s, F, En). For, intuitively, if F is exemplified by n distinct objects in the story, each of them should have its own distinguishing essence. However, for reasons that we shall see below, we should not assume that whenever C(s, F, E) is the case, E is exemplified in s. But how should we understand the locution “F is exemplified by n objects in a certain story”? To answer this, let us first note that it is appropriate to take stories to be (usually very complicated) propositions23 that result from an in-depth, ideal interpretation of a concretely existing text generated by a certain author, an interpretation which may take into account the time and cultural environment in which it was written, the author’s intentions and whatever factors are deemed important in literary criticism. In this perspective, to say that a proposition P holds with respect to (in, according to) a certain story s means that s entails P, where the entailment in question is arguably of a nonclassical, indeed paraconsistent, nature (for we must allow for the fact that a story can contain contradictions and informational gaps; see, e.g., Deutsch 1985). Note also that a story is expressed in natural language in a way that typically involves, whether explicitly or implicitly, many sentences of identity across time. Thus, if my approach is correct, a story involves properties that are predicated in a metaessential way with respect to a time. Accordingly, that a property F is exemplified by n objects in a story s should be understood (at least by default) as the assertion that s entails the proposition that there are n real essences linked to property F through metaessential predication with respect to a time. Coexemplification in a story should be understood in the same vein. With this in mind, sentences of form (ITS) should be analyzed as follows: (ITSc)∃1E (C(s, F, E) & (∃1E' (C'(s', F, E') & E = E')),

23 See, e.g., Plantinga’s (1974, p. 159) notion of story line.

Francesco Orilia

211

where C and C' are (typically) the same criterion24. Thus, for example, (2) can be paraphrased in this way: (2c) ∃1E (C(Odyssey, Achaean who devised the trick with the Trojan horse, E) & (∃1E' (C(Inferno, Achaean who guided the expedition past the Pillars of Hercules, E') & E = E')). At least in typical cases, a fictional essence could be characterized as a salient character property, i.e., intuitively, something that allows one to keep track of the same character within a story. In relation to a story s, a property F of this kind is such that: (i) s entails (paraconsistently) 24 I have provided an analysis analogous to (ITSc) in Orilia 2002. Let me reply

here to some criticisms that have been put forward against it, since this might be useful for a better appreciation of my proposal. Frigerio 2004 argues that in many cases we are quite certain of the truth of intertextual reidentification sentences, whereas, if we should rely on reidentification criteria in the way I suggest, we should typically be uncertain (undecided between a judgement of similarity and one of identity). I don’t see how this can be the case, for the reidentification criteria I postulate are meant to be precisely the criteria we use when we claim that certain reidentification sentences are true. Frigerio also points out that, when E is an essence from the point of view of story s and thus, roughly put, corresponds to a character c of s, we could imagine a story s' where E is exemplified by a different character c'. He is right in this, but errs, when he thinks that this is at odds with my approach. His intuition seems to be that, if E is essential to c and c' exemplifies E, then c should be equal to c', contrary to the assumption that c and c' are different. But in fact my approach can very well account for what Frigerio is pointing at, for it allows for a story s' where a fictional essence E is exemplified (and thus, so to speak, there is a character in s' with that property), although E is not an essence from the point of view of s'. Let us turn now to Voltolini (2004), who argues against my use of identity in the counterpart of (ITSc) in Orilia 2002. For, he claims, most of the so-called judgements of identity across stories in fact assert a fusion of two characters and only seldom are two characters really claimed to be identical. To illustrate, he mentions that both the Berget and the Vington of the 1912 version of the Recherche are said to be the Vinteuil of the final version of Proust’s masterpiece. It seems to me questionable that such fusion claims are many and the identity claims rare. But even if the latter were so, we should account for them any way and thus I think that the use of identity in (ITSc) is as it should be. Needless to say, however, fusion claims such as the ones about Vinteuil must also be accounted for. This can be done, I think, in terms of a conjunction of essences or something like that: ∃1E(C(s, F, E) & (∃1E'(C(s', G, E') & (∃1E'(C(s", F, E") & E" = [E & E'])).

212

Identity across Time and Stories

that there are various times at which F is exemplified, but it is false that s entails that there is a time at which F is exemplified by more than one object; (i) s entails (paraconsistently) that there is a real essence with which F is coexemplified at each time at which F is exemplified. For example, in Sherlock Holmes stories, a salient character property (intuitively corresponding to the character Sherlock Holmes) is something like: being an entity exemplifying most of these properties: being called Sherlock Holmes, being very clever, being a detective, living in London, dressing in such and such a way, etc. In describing how reidentification criteria behave with respect to stories, we can distinguish, I think, two main kinds of them. Criteria of the first kind tend to privilege a partial (defeasible) reidentification criterion that yields as essences salient character properties that are exemplified in the story in input and possibly in other stories. A criterion of this kind can be considered at play in our judgement that one Sherlock Homes character appears in A.C. Doyle’s stories as well as in the wellknown movie The 7% Solution, where Holmes meets Freud. Criteria of the second kind, on the other hand, tend to yield essences, by favouring a partial (defeasible) reidentification criterion that relies on the fact that, roughly put, an author A takes another author A' as a source. In such a situation, when A views a property P as exemplified from the point of view of the story she is putting together, A is appropriately connected (by her beliefs and intentions) to a story s' previously conceived by A' (via one of its concrete linguistic realisations) in a way that can be described as “imagining that P is exemplified by (some sort of literary reincarnation) of such and such a character in s'”. In this case, the essence could be a fictional essence of the previous story s' (e.g., a salient character property exemplified in s), which may very well be a property that is not a salient character property in s or is not even exemplified in s. By relying on the existence of partial reidentification criteria of this kind, we can account for the fact that, e.g., we may judge that Joyce’s Ulysses is the Ulysses of the Odyssey25 (in spite of the fact that, roughly put, the salient character property linked to 25 Voltolini 2004 relies on this judgment to criticise the previous version of the

present approach in Orilia 2003. Although I do not agree with some details of his criticism (inspired, he says, by Thomasson 1999), I was at least in part motivated by it in putting forward this second criterion.

Francesco Orilia

213

Ulysses in the Odyssey is arguably different from that linked to Joyce’s Ulysses). In this way, we can account for the fact, emphasised by Thomasson 1999, that literary critics take into account, in judging whether a character c in story s is the same as a character c' in story s', whether s' is a source (possibly indirect) for the author of s. Thomasson argues that there should always be such a link between s and s' for c to be judged identical to c' (p. 67), for this is the way literary critics see the matter (p. 6). Literary critics would say, according to Thomasson, that, if there is no such link, c and c' can at most be “analogs”. But I doubt that this is correct. Even assuming that literary critics consistently judge in this way, which is moot, this might be explained by their (typically Western) interest in giving a prominent role to the creative capabilities of individual authors and thus to “copyright issues”. By following this interest, we may be tempted to give identity conditions for stories like those Thomasson suggests for “literary works” (p. 64), according to which there could be two stories which differ simply because they were invented independently by two different authors, but are identical as far as their content goes. From this point of view, it might seem sensible to consider as conceptually impossible (rather than simply very unlikely) that two authors may independently write the very same story with the very same characters. But if we take this line, we forget that language is an instrument that allows different people to express the same meaning with different tokens (perhaps even tokens of different languages). If we recall this, it is preferable to say that two authors may in principle express the same story (a proposition) just as two people express the same proposition by uttering two tokens of “snow is white”. It is worth noting that in my approach we can account for the intuition that a real object can be a character in a story (Castañeda 1977), for, as I see it, nothing prevents a real essence from being a fictional essences as well26. It is always an option in fact that a criterion C takes a story s to be “talking about a real object”, or perhaps more precisely, in the present perspective, as “talking about a real essence”. In this case, for some property F, we would have that C(s, F, E), where E is a real essence. For example, in Sherlock Holmes stories, A. C. Doyle can

26 See Orilia 2000b for more details on this.

214

Identity across Time and Stories

be assumed to talk about the real London. Thus, if we assume that being spatially continuous with is a real essence, according to a criterion C, an essence intuitively corresponding to London, we may say that this property is not only a real essence but also a fictional essence for it is the case that, say, C(The Sign of Four, called London, being spatially continuous with ). 12. GRRSR and ID reconciled Let us see where the approach I am proposing drives us, once we reconsider GRRSR and its apparent tension with ID. In the present perspective, there are fictional characters in the sense that singular terms that we prima facie take to be referring to them, typically refer to something. They refer to fictional essences. For example, “the Achaean who devised the trick with the Trojan Horse” expresses the denoting concept [THE Achaean who devised the trick with the Trojan Horse], which we may take to determine a certain fictional essence E, which the singular term in question refers to. But fictional essences are properties, i.e., entities that an actualist like Russell, or even a presentist, would accept in his ontological inventory. More generally, in being properties, I would say that fictional entities are, to adapt an expression from Armstrong 1997, “no real addition to being”, in that properties must be acknowledged in our ontological inventory anyway, quite independently of any analysis of fiction, and intertextual reidentification. This is in line with RRSR. Similarly, there are in my approach, so to speak, past and future objects that do not presently exist, but simply in the sense that singular terms that we prima facie take to be referring to them refer to real essences. For example, “the original manuscript of Plato’s Phaedo” can be taken to express the denoting concept [THE original manuscript of Plato’s Phaedo], which determines not so much an object, but a real essence E, a property which, for all we know, is not presently exemplified. Since real essences are properties, to admit past and future objects in this sense is again no real addition to being, in line with GRRSR. Compatibility with GRRSR is not gained however in a way that ignores ID. For, since singular terms for fictional, past and future objects typically have a referent in this approach, we can grant that sentences (5) (“Van Inwagen admires Ulysses”) and (6) (“Van Inwagen admires Socrates”) are true, by way of corresponding ultimately to a

Francesco Orilia

215

relational fact linking an individual and an essence, rather than to a mysterious relational fact with a nonexistent relatum. Let me further clarify this point. First note that for a predicate scheme like “x admires the F” to be true of someone, y, it must be the case that y has appropriate “admiration-beliefs”, that is, as I see it, beliefs involving, in this case, the denoting concept [THE F] and which could be expressed by sentences such as “the F was wise and wisdom is a virtue”. Now, from my point of view, a proposition expressible with “the F was wise” typically “purports to speak about an essence” and is true just in case there is exactly one essence E such that wisdom is truly predicated of it in the metaessential sense. Now, the admiration could be purely de dicto, when (unbeknownst to y) the denoting concept [THE F] does not determine anything (as is the case, for example, with the denoting concept expressed by “the coach of the team who won the soccer world championship in 1942”; there was no world championship in 1942). But the admiration is, in a sense, de re, when the denoting concept determines an essence, as with (5) and (6), in which cases there are relational facts involving a person (Van Inwagen) and an essence. We also have the intuition, I think, that in (5) the admiration is, so to speak, more de re, for Ulysses is fictional, whereas Socrates concretely existed. Now, it is appropriate in my approach to grant a secondary sense in which a singular term “the P” may singularly refer to an entity x. This sense applies when the denoting concept [THE P] expressed by the term refers (in the primary sense) to an essence E, which in turn is univocally exemplified by x. In this case, we may add, [THE P] indirectly determines x. With this further notion, we do justice to the intuition that we just considered: (5) differs from (6) in that only so far as the former is concerned Van Inwagen admires by way of employing a denoting concept that, in the past, indirectly determined an entity. 13. Relations and metaessential predication We have seen how “intentional” relations like admiring (and similarly being afraid of, thinking of, etc.) should be dealt with in the present approach. But it is worth considering also non-intentional relations such as touching or being taller than. For, once we assume that a property such as being a ship can be predicated in a metaessential way, we

216

Identity across Time and Stories

should assume the same for these non-intentional relations. Let us see how this can be understood. For simplicity’s sake we consider only dyadic relations (we can easily generalise from them). Let us assume that a criterion C, when taking in input one such relation R, also takes in input a pair of times . Correspondingly, we have in output pairs , where E1 and E2 are conventional essences. Thus, we should represent the metaessential predication of R with respect to E1 and E2, rooted, so to speak at times t1, and t2 as follows: C(, R, ). For an expression like this to be true, we require first of all that E1 and E2 be conventional essences in the sense explained above. Moreover, there are additional conditions, depending on the nature of R. We can distinguish in fact two main kinds of (non-intentional) relations. The intratemporal ones, like touching, kissing, being over, etc., require two individuals existing at the same time. When R is of this kind, the further condition is simply that there is a time t equal to t1, and t2 and such that, at t, there are individuals x and y, univocally exemplifying E1 and E2, respectively, and which are in the relation R. Further, there are intertemporal (non-intentional) relations, like taller than, wiser than, being an ancestor of, etc., which, so to speak, require two individuals possibly located at different times. For them, we must add this further condition. There are properties P1 and P2, such that at each ti, Ei is univocally exemplified by an object that also exemplifies Pi (for i = 1 or i = 2). Moreover, there is a relation R* (appropriately connected to R in a sense to be explained with a forthcoming example) such that is exemplified by R*. And here is the example. Consider the proposition that Clinton, as he is at the present moment, t, is taller than Napoleon, as he was at a certain moment t' in the past. We can take this to be true on the ground that, roughly, the following proposition is true: ∃1E (C(t, called Clinton, E) & ∃1E (C(t', called Napoleon, E') & E and being 6 feet tall are coexemplified at t & E' and 5 feet tall are coexemplified at t'', and six feet tall is an height greater than 5 feet tall. In this case, R is the relation of being taller, P1 and P2 are the properties 6 feet tall and 5 feet tall and R* is the relation of being a greater height. By analogy with what we did for times, we can extend this treatment of relations to stories, just as we did for properties. In view of this, “Polyphemus is taller than Clinton” receives, roughly, the following

Francesco Orilia

217

analysis: ∃1E (C(Odyssey, called Polyphemus, E) & ∃1E' (C(now, called Clinton, E') & ∃P1(E and P1 are coexemplified now & ∃P2 (E' and P2 are coexemplified in the Odyssey & P1 is an height greater than P2). Intuitively, P1 is the property six feet tall (suppose this is how tall Clinton is) and P2 is being as tall as a giant (a property clearly attributed to Polyphemus in the Odyssey). 14. Some tentative conclusions Since writing fiction is an essentially linguistic activity by means of which (by and large) we invent rather than describe, in explaining what it is about, we need not go beyond a description of our conceptual structure. Hence, if my analysis is right, there is no reason to postulate fictional objects (understood as particulars) over and above the properties (conventional fictional essences) appealed to in my account (in line with actualism). But in using sentences of identity through time, we try to describe the world. Accordingly, in relation to them, we cannot rest content with an attempt to describe our conceptual structure. My own ontological analysis in this respect ended up with the idea that there are conventional essences relative to a reidentification criterion. But, as we saw, this is compatible with endurantism, perdurantism, sequentialism, Plantinga's essentialism, etc. As I see them, these are different theories about the real world. They are grist to the mill of the prescriptive ontologist, who must decide about them not merely by trying to decipher our conceptual structure (it is neutral, if I am right, with respect to them), but by bringing in additional considerations, which may very well be of an empirical a posteriori nature (e.g., by appealing to relativity theory in order to argue for perdurantism). The main issue can be roughly put as follows. If a real essence is univocally exemplified by an object, is this object, so to speak, an enduring, a perduring or a merely “sequential” object, an ens successivum, in Chisholm’s (1989) terminology? Moreover, we could ask, is there a reidentification criterion that could be preferred to the others in that it (typically) yields essences that coincide with objective essences in Plantinga’s sense and/or correspond to persisting objects? We should not take for granted, I think, that there are universal answers. For example, Chisholm may be right in insisting that ships are entia successiva, whereas persons are persisting entities. And, there may be a reidentification criterion that functions better in capturing Plantinga

218

Identity across Time and Stories

essences of living beings and another to be preferred in an effort at capturing Plantinga essences of rocks and stones (if we assume essences of this kind)27. References Armstrong, D.A, (1997), A World of States of Affairs, Cambridge, Cambridge University Press. Castañeda, H.-N., (1974), ‘Thinking and the Structure of the World’, Philosophia, 4, pp. 3-40. –– (1979), ‘Fiction and Reality: Their Fundamental Connections’, Poetics, 8, pp. 31-62. –– (1980), On Philosophical Method, Bloomington, IN, Noûs Publications. Chisholm, R.M., (1970), ‘Identity Through Time’, in Language, Belief and Metaphysics, (eds.) H. E. Keifer and M. K. Munitz, Albany NY, State University of New York Press. Reprinted in Chisholm 1989, pp. 19-41 (the page reference in the text is to this edition). –– (1971), ‘Problems of Identity’, in (ed.) Munitz 1971, pp. 3-30. –– (1989), On Metaphysics, Minneapolis, University of Minnesota Press. Cocchiarella, N.B., (1982), ‘Meinong Reconstructed versus Early Russell Reconstructed’, Journal of Philosophical Logic, 11, pp. 183-214. Coburn, R.C., (1971), Identity and Spatio-temporal Continuity, in Munitz 1971, pp. 51-101. Craig, W.L., (2000), The Tensed Theory of Time, Dordrecht, Kluwer. Deutsch, H., (1985), ‘Fiction and Fabrication’, Philosophical Studies, 47, pp. 201-211.

27 This paper is my last try (so far) at implementing an appeal to conventional

essences in order to account for reidentification sentences and provide a general ontological picture that takes them as departure point. Since 1986 (cfr. Orilia 1989), I have pursued this line in various forms. Originally, in the context of Castañeda’s guise theory. In Orilia 1995, roughly, as part of a conception of fictional characters as denoting concepts. Since Orilia 2000b, by identifying fictional characters with fictional essences. In addition to colleagues, friends and mentors whose help has already been acknowledged in my previous publications on this topic, I wish to thank for their interest and suggestions all the participants in the 2005 Bergamo workshop on ontology, On What (Perhaps) There is, where this work was presented. Among them, Richard Davies deserves a special mention for his careful editing of a previous version of this paper and for stylistic suggestions.

Francesco Orilia

219

Hirsh, E., (1971), ‘Essence and Identity’, in (ed.) Munitz 1971, pp. 31-49. Frigerio, A., (2004), Review of F. Orilia, Ulisse, il quadrato rotondo e l’attuale re di Francia, Rivista di Filosofia Neo-Scolastica, 96, 1, pp. 217-224. Loux, M., (1998), Metaphysics, A Contemporary Introduction, London, Routledge. Montague, R., (1974), Formal Philosophy, (ed.) R. Thomason, New Haven, CN, Yale University Press. Munitz, M.K., (ed.), (1971) Identity and Individuation, New York, New York University Press. Orilia, F., (1989), ‘Identity across Frames’, Topoi, Supplement 4, pp. 85-97 (originally presented at the Indiana Philosophical Association, Kokomo, November 1, 1986 and at the conference The Object and its Identity, Trento, May 27-30, 1987). –– (1995), A Property-theoretical Account of Reidentification, unpublished ms., presented al the 10th International Congress of Logic, Methodology and Philosophy of Science, Florence, August 1995 and at the Indiana University Logic Group, Bloomington, IN, September, 6, 1995. –– (2000), ‘Property Theory and the Revision Theory of Definitions’, Journal of Symbolic Logic, 65, pp. 212-246. –– (2000a), ‘The Property Theoretical Performative-Nominalistic Theory of Proper Names’, Dialectica, 54, pp. 155-176. –– (2000b), ‘Essenze reali ed essenze fittizie’, in (ed.) G. Usberti, Modi dell'oggettività, Bompiani, Milan, 2000, pp. 221-239. –– (2003), ‘A Description Theory of Singular Reference’, Dialectica, 57, pp. 740. –– (2003a), ‘Logical Rules, Principles of Reasoning and Russell’s Paradox’, in (eds) T. Childers and O. Majer, The Logica Yearbook 2002, Prague, Filosofia – Institute of Philosophy, Academy of Sciences of the Czech Republic, pp. 179-192. Partee, B., A. ter Meulen, and R.E. Wall, (1990), Mathematical Methods in Linguistics, Dordrecht, Kluwer. Plantinga, A., (1974), The Nature of Necessity, Oxford, Oxford University Press. Prior, A.N. and K. Fine, (1977), Worlds, Times and Selves, London, Duckworth. Quine, W.V.O., (1960), Word and Object, Cambridge, Mass., MIT Press,. –– (1969), Ontological Relativity and Other Essays, New York, Columbia University Press. Russell, B., (1905), ‘On Denoting’, Mind, 14, pp. 479-493. Strawson, P.F., 1959, Individuals, London, Methuen. Swartz, N., (1991), Beyond Experience, Toronto, University of Toronto Press. Thomasson, A.L., (1999), Fiction and Metaphysics, Cambridge, Cambridge University Press. Van Inwagen, P., (1977), ‘Creatures of Fiction’, American Philosophical Quarterly, 14, pp. 299-308.

220

Identity across Time and Stories

Varzi, A.C., (2001), Parole, oggetti, eventi, Rome, Carocci. Voltolini, A., (2004), ‘Presentazione di: F. Orilia, Ulisse, il quadrato rotondo e l’attuale re di Francia, unpublished ms. presented at the workshop Ontologia del discorso narrativo tra fenomenologia e filosofia analitica, University of Macerata, February, 18, 2004. Voltolini, A., (forthcoming), How Ficta Follow Fiction, Berlin, Springer.

A Problem about Reference in Fiction* Giuseppe Spolaore In many works of fiction, at least prima facie, reference is made to real individuals. In such cases, it is plausible to think that those real individuals are characters of the works. Indeed, according to a very natural position in the theory of fiction, these two theses hold: (IF) A genuine proper name used in a work of fiction keeps referring to its customary, non-fictional, bearer. (RIF) If in a work of fiction reference is made to a real individual, that individual is a character of the work of fiction. (IF) is a formulation of a principle that may be labeled “Semantic innocence of proper names in fiction” (“Innocence of Fiction” for short). Innocence of Fiction states that the fact that a name occurs in a work of fiction is, by itself, referentially irrelevant. By “a genuine proper name” I mean a name which has been introduced to refer to a real individual. Obviously enough, Innocence of Fiction presupposes that genuine proper names can be used in a work of fiction. The arguments in favor of Innocence of Fiction parallel those in favor of the more general thesis of Semantic Innocence. (SIPN) is a particular instance of this latter principle (where the name is used, not mentioned): (SIPN) A proper name refers to the same individual in every linguistic context. As is well known, (SIPN) is typically held by direct reference theories of names. As long as one does not make and defend any explicit claim about the referential peculiarity of fiction, (IF) may be taken to follow from (SIPN). On the other hand, (RIF) is a natural principle if one thinks that a character of a story is any individual the story speaks of, i.e. an object of the narration.

* Thanks to Massimiliano Carrara, Pierdaniele Giaretta, Frederick Kroon and

Vittorio Morato for their very helpful comments.

222

Reference in Fiction

Here I shall introduce and discuss a problem for any realism of fictional characters1 willing to include (IF) and (RIF). Suppose that some time ago a man named “Caius” wrote a work of fiction entitled “Two Planets”. In Two Planets there occur two co-referential names, “Hesperus” and “Phosphorus”, and it is clear that they are employed in their customary use, namely to refer to Venus. Caius is a competent user of both names but, unfortunately, he believes that Hesperus and Phosphorus are two different celestial bodies2. Consequently, Two Planets clearly portrays Hesperus and Phosphorus as distinct planets, claiming, for example, that they have distinct orbits, different influences on the Earth, that one, but not the other, has been hit by a huge asteroid and so on and so forth3. Further, in the story both Hesperus and Phosphorus are portrayed as relevantly Venus-like. Now, it is natural to assume that such a story has, among the others, two distinct characters, Hesperus and Phosphorus, so that (1) Hesperus and Phosphorus are distinct characters of Two Planets. On the other hand, it is a matter of fact that (2) Hesperus is identical with Phosphorus. By (IF), (RIF) and (2) one can obviously infer that (3) Hesperus and Phosphorus are the same character of Two Planets. But (1) and (3) are inconsistent4. 1 By “realism of fictional character” I mean – roughly speaking – any view

according to which there are such things as fictional characters. For a similar use of the label see e.g. Walton 1990: 390 and Brock 2002: 1-5. 2 Here I am assuming that one may competently use two names even if one is not aware that they are co-referential. For a defense of this thesis, see Kripke 1979. 3 Frederick Kroon has proposed a conceptually analogous imaginary situation while dealing with the problem of informativeness of identity. See Kroon 2001. 4 Notice that he problem I am raising here does not depend on any specific feature of “Hesperus” and “Phosphorus”. It may be restated (mutatis mutandis, of course) by using any other pair of co-referential names, e.g. “Mark Twain” and “Samuel Clemens”, “Bill Clinton” and “Bill Blythe”, “Everest” and “Sagarmatha”, “Cicero” and “Tully”, “London” and “Londres” and so on and so forth.

Giuseppe Spolaore

223

Thus, either at least one among (IF), (RIF) and (1) is false, or some of the expressions involved in this piece of reasoning are ambiguous. Even if this problem is pretty general, I am going to aim at a precise polemical target; namely a view whose main supporters are Amie Thomasson and Nathan Salmon5. Following Amie Thomasson, I label such a view as “artifactual”. Artifactual theories are committed to three main principles: Creationism: Fictional characters are, in Salmon’s words, “manmade artifacts created by fiction writers”. This thesis is presented for completeness. My point is wholly independent of Creationism, depending, among others, on a more general thesis, which is obviously implied by Creationism, namely realism of fictional character. Direct Reference: A certain class of expressions – proper names, indexicals, etc. – directly refers to individuals. The semantic value of such expressions is simply their referent. Further, arguably, artifactualism supports (SIPN). Objectualism of Characters: The characters of a story are the objects a story speaks of; in particular, they are the denotations of the proper names and denoting terms occurring in the text and the objects the story ascribes properties to. This paper comes into three parts. In the first (§§ 1-2) I shall specify and motivate (IF) and (RIF). In the second (§§ 3-4) I shall try and show that the problem of Two Planets can hardly be solved in an artifactual framework. In the third (§5) I shall briefly make some general remarks about this problem and sketch a possible diagnosis. Before starting, I want to emphasize that the question here is metaphysical rather than epistemological. That is to say, I assume that Caius has referential intentions when using “Hesperus” and “Phosphorus”, that these referential intentions are really directed toward a single object, that is Venus, and that the context makes it clear which intentions Caius has. The point is not what someone who is not aware of Caius’ intentions would say about the names’ reference or about the 5 See e.g. Thomasson 1999 and Salmon 1998. See also e.g. Predelli 1997.

224

Reference in Fiction

number of characters of Two Planets. The point is what we, given that Caius’ intentions are pretty clear, would say about the reference of “Hesperus” and “Phosphorus”, and why we, Caius’ intentions notwithstanding, would say that Hesperus and Phosphorus are distinct characters of Two Planets 6. Now, let us start by examining Innocence of Fiction, (RIF) and some evidence militating in their defence. §1. Innocence of Fiction As it stands, (IF) can be held only by supporters of Direct Reference, and in particular of (SIPN). For example, a fiction may very well introduce that-clauses and other contexts in which, according to some theorists, and in particular Fregeans, names do not keep their customary referent. But, intuitively, this problem has nothing to do with what Innocence of Fiction is taken to claim; namely that genuine proper names do not acquire new referents when occurring in a fictional context. Since I am primarily dealing with artifactualism, which supports 6 One may be tempted to solve our problem by claiming that there really are two

possible interpretations involved here. This move may be supported by appealing to the phenomenon of unreliable narrations. Take as an example the following short story: (†) In French, “canard” means water. France is nice but Frenchmen are strange: when you ask them for some water they bring you a duck. This story has of course two possible interpretations. According to one, “Canard” does mean water and Frenchmen are a mass of buffoons. In the other, the story itself is unreliable, and “canard” means, as it actually does, duck. In the same vein, one can argue that Two Planets is open to two interpretations. According to the first one, call it “True Planets”, Hesperus and Phosphorus are distinct individuals. In the second interpretation, call it “Untrue Planets”, Hesperus and Phosphorus are one and the same planet, and the story itself is unreliable. In this way, our problem would vanish, since in this case (1) and (3) should be understood with reference to different interpretations, and so they would be true of different characters. This alleged solution is misguided. In the first place, we don’t have different intuitions about the number of characters of Two Planets. We may very well assume that, after reading the story, everybody would without a doubt subscribe to (1). But even if there were two interpretations, Innocence of Fiction and (RIF) would nonetheless imply that Hesperus and Phosphorus are exactly one character of Two Planets in both the interpretations.

Giuseppe Spolaore

225

Direct Reference, I shall ignore this problem and stick with formulation (IF); but I want to stress that Innocence of Fiction may be understood as not presupposing the truth of (SIPN), and more in general of the Theory of Direct Reference. To test our intuitions about (IF) it is useful to assume that Caius, being an amateur astronomer, wrote a second – non-fictional – work, entitled “Serious Planets”. It is not necessary to assume that Serious Planets and Two Planets are word by word identical. We may just suppose that they are relevantly similar. Finally, suppose that both works include an inscription of this sentence: (4) Hesperus and Phosphorus are distinct planets. Now consider these two questions: (5) Do “Hesperus” and “Phosphorus” as they occur in the Serious Planets’ inscription of (4) refer to Venus? and (6) Do “Hesperus” and “Phosphorus” as they occur in the Two Planets’ inscription of (4) refer to Venus? Now, notoriously, it is very likely that the right answer to (5) is “Yes”; at any rate, we may happily assume that “Yes” is the artifactualist option7. So, arguably, if artifactualists want to answer “No” to (6) they must deny Innocence of Fiction. To understand this point, it may be useful to consider a perfectly symmetrical case concerning beliefascriptions. Suppose that Caius seriously states (4), meaning to refer to the planet we actually know as Venus. Suppose that later, using “Hesperus” and “Phosphorus” with the same referential intentions, he says: (7) I believe that Hesperus and Phosphorus are distinct planets. Now, it would be double thinking, if not open incoherence, to subscribe to Semantic Innocence – that is, for what matters here (SIPN) –, and at the same time claim that the semantic value of “Hesperus” and “Phosphorus” is not the same in (4) and (7). Analogously, one can

7 See Kripke 1979. Kripke’s point is that to answer “No” to such a question one

should impose implausibly strong constraints on the competent use of a name.

226

Reference in Fiction

hardly accept Innocence of Fiction and at the same time give different answers to (5) and (6). Since I take this to be an extremely important point, I want to stress it by considering an intuitive reaction to our problem. One may be tempted to stick with Innocence of Fiction and solve the puzzle by postulating a referential shift. In Amie Thomasson’s words: It is hard to specify the point at which a name-use practice shifts from merely spreading false information about a certain individual, real or fictional, to the point at which we would say that it is no longer about that individual but about some mythical or imaginary entity. […] [T]he maintenance of an historical chain of name use leading back to baptism is only a necessary, and not a sufficient, condition for a particular use of a name to refer back to the appropriate individual8.

No one can deny that referential shifts may happen. But appeals to referential shifts have their limits. Suppose that, one evening, Caius said, pointing to Venus: (8) Two Planets speaks of that planet, which we refer to by the name “Phosphorus”. The question is, what we, knowing what we know, would now say about Caius’ claim? If we think it is false, then we have to postulate a referential shift between the Caius’ uses of “Phosphorus” in normal, non-fictional, speech and in fictional narrations. But again, it is hard to see how this may be done without appealing to some specific semantic feature of fictional texts, in other terms without denying Innocence of Fiction. Now let us consider some evidence one may offer to support Innocence of Fiction. In the first place, one can make some general appeals to intuition. Here are some examples. Appreciation of fiction (in Amie Thomasson’s words9): Most works of historical fiction would lose much of their poignancy if they were not set amid real historical events and individuals […]. Tom Stoppard’s Travesties, for example, would lose much of its humor if it did not involve the real Lenin, Tristan Tzara, and James Joyce coming together in Vienna, but only some similar fictional individuals (in a similar fictional city). 8 Thomasson 1999: 51. 9 Thomasson 1999: 104.

Giuseppe Spolaore

227

Lies (in Terence Parsons’ words10): […] I see no difference in the [following] referential situations (i) Telling a lie about Carter (ii) Telling a lie about Carter which is very long (e.g. book length). (iii) Making up a story about Carter which is not intended to deceive anyone, and which contains falsehoods.

Intuitively, there seem to be few – if any – referential differences among these situations. But it would be absurd to claim that a name occurring in a lie does not refer to its customary referent. So, it is quite plausible to assume that a name used in a work of fiction keeps referring to its customary referent. Name interpretation Notoriously, there is no syntactic mark distinguishing fiction from nonfiction. Further, there may be no available evidence at all that a text is fictional. Now, imagine the following situation. You know that both “Hesperus” and “Phosphorus” refer to Venus. You read a sequence of sentences S in which those names appear. You are told that S occurs in a book written by Caius, who used them to refer to Venus. You know that Caius wrote exactly two books, a scientific treatise and a work of fiction. S is such that it leaves the question open whether it occurs in one work or in the other. Do you need to be sure that S does not occur in the work of fiction to justifiably claim that both names refer to Venus? “No” seems by far the more plausible answer. But this strongly suggests that knowing whether a text is or is not a piece of fiction is actually irrelevant for correctly interpreting a name. Further, Innocence of Fiction follows from some widely-held – though not unchallenged – tenets regarding semantics, pragmatics and the nature of fiction. For example, from (9) and (10)11:

10 Cfr. Parsons 1980: 58, where it is intended as evidence for a slightly different

thesis. 11 (9) is a widely-held tenet regarding the relationships between the pragmatic properties of a sentence and the referential properties of a name occurring in that sentence. (10) is a formulation of the pragmatic view of fiction. Even if it has not gone unchallenged – see e.g. Deutsch 2000 – at the present time such a view is probably the standard approach.

228

Reference in Fiction

(8) The illocutionary force of a sentence (or the speech act it is used to perform), its communicative function or other similar pragmatic features are irrelevant as far as the reference of a genuine proper name occurring in that sentence is concerned. (9) What makes a text/discourse T an instance of fiction are exclusively its pragmatic features such as the (perceived or designed) illocutionary force or communicative function of the sentences occurring in T12. Even if some semantic or stylistic features of T may be regarded as evidence that T is an instance of fiction, still, they are not what make T an instance of fiction. Thus, two texts may be one an example of fiction and the other an example of nonfiction even if they share the same stylistic and semantic features. Now let us consider the thesis (RIF) (“If in a work of fiction reference is made to a real individual, that individual is a character of the work of fiction”). §2. The objectual view of characters If one agrees with principle (RIF), one is very likely to adopt Objectualism of Characters. To make the objectual conception clearer, it may be useful to introduce also its opponent, which I arbitrarily label the “non-objectual view of characters”. According to a non-objectual view, characters are not the objects a story speaks of; rather, to be a character of a story is to be an entity that is determined by – or supervenes on – that story. If a story speaks of an object, say Mont Blanc, Mont Blanc with its snowfields is not a character of the story. Neither is the character Sherlock Holmes the referent of “Sherlock Holmes” as 12 For instance, Richard Gale claims that an author of fiction “does not perform

the illocutionary acts which are typically performed by the use of the sentences making up his story” (Gale 1971: 325); Richard Ohmann argues that sentences occurring in fiction lack the illocutionary force they typically have (Ohmann 1971); John Searle maintains that authors pretend to perform speech acts while writing a work of fiction (Searle 1974). See also Martinich 2001. Gregory Currie identifies a specific, “fictive”, type of speech act which is characteristic of storytelling (Currie 1986). That fictions have a distinctive communicative function is recognized by many authors; see e.g. Currie 1990, Walton 1990.

Giuseppe Spolaore

229

it occurs in the Conan Doyle’s stories. Nevertheless, the character Sherlock Holmes is the referent of the name “Sherlock Holmes” as it occurs in statements like (10) Sherlock Holmes is a fictional character. Notice that the distinction between objectualism and non-objectualism concerns the notion of a character. In other terms, the disagreement is about the condition a given entity must satisfy to be a character of a story. Indeed, the distinction may be stated in a way that is relatively neutral with respect to what sort of objects characters are. Further, and typically, antirealist theories of fictional characters do not accept fictional characters in the ontology precisely because they support objectualism of characters. Fictional characters do not exist precisely because there is nothing that names like “Holmes” refer to as they occur in the relevant works of fiction. At any rate, any objectual theory of characters agrees with a principle that may be formulated as (O) (where P is – more or less – any property): (O) A character of a work of fiction W is any object x such that, according to W, x is P. This principle should capture the idea that characters are what stories speak of. On the other hand, realist objectual theories of characters – that is objectual theories that admit fictional characters in the ontology – hold that attempts at reference occurring in a work of fiction are bound to be successful, so that any name (even if formerly empty) when used in a work of fiction actually refers to an entity that is a character of that work13. In favor of objectualism one may offer an argument that mutatis mutandis parallels a famous passage in Quine. In “On What There Is”14 the imaginary philosopher McX claims that Pegasus exists, but as an idea, not as a concrete individual. Quine replies that McX’s thesis is a desperate move: we never confuse a real object – say the Parthenon – with idea(s) of it, say the Parthenon-idea(s); so why suppose that we 13 See e.g. Thomasson 1999: 12-13, Salmon 1998: 293-304. 14 Quine 1948.

230

Reference in Fiction

make such a confusion where Pegasus is concerned? Analogously, an objectualist may observe that we never confuse the Parthenon with other entities that are determined by, or supervene on, a description of the Parthenon. So why suppose that we regularly make such a confusion when dealing with Holmes or Pegasus? Realist objectual theories of characters grant an intuitively appealing and uniform semantic treatment of a large number of sentences about fiction. For example, one may consider all these sentences as straightforward claims about one and the same object, Sherlock Holmes: (11) Sherlock Holmes is a fictional character. (12) “Sherlock Holmes seemed delighted at the idea of sharing his rooms with me” (A Study in Scarlet, Chapter 1). (13) According to the story, Sherlock Holmes is a detective. (14) I admire Holmes. The thesis (RIF) does not imply objectualism. Ideally one could be an objectualist with respect to real (or historical) characters and a nonobjectualist with respect to fictional ones. I ignore such a possibility, which would implausibly jeopardize the very notion of a character. On the other hand, a realist objectualism can imply (RIF) in a merely trivial manner. In fact, one may grant that characters are the objects of the story, but maintain that stories can never speak of real individuals, so that the antecedent of (RIF) is false and (RIF) itself trivially true. I shall not consider such a position, which is obviously incompatible with Innocence of Fiction. §3. Weakening Innocence of Fiction Let us return to our puzzle, and introduce a natural way out for artifactualists, namely by weakening Innocence of Fiction. What a supporter of an artifactual theory should argue – in the very end – is that: (WIF) Typically a genuine proper name used in a work of fiction keeps referring to its customary, non-fictional, bearer; but there are some exceptions. For example, if two genuine co-referential names “a” and “b” appear in a work of fiction W, and according to W a is distinct from b, then neither “a” nor “b” refers to its customary referent.

Giuseppe Spolaore

231

Obviously, (WIF) is as ad hoc as it could be. To see in what sense, it is sufficient to consider, once again, the debate about semantic innocence. Imagine a former supporter of (ISPN)* (“A genuine proper name retains its semantic value – i.e. its referent – in every linguistic context”) who, faced with some recalcitrant data – for example prima facie failure of substitutivity salva veritate of co-referential names in belief ascriptions – replies with the following formulation: (WSI) Typically, a proper name occurring in a sentence embedded in a belief ascription retains its semantic value. But there are some exceptions. For example, if two genuine co-referential names “a” and “b” appear in the embedded sentence, and if the embedded sentence implies that a is distinct from b, then neither “a” nor “b” refers to its customary referent. One should justify (WSI) on general and independent grounds. I think that it should be clear in advance that this enterprise is hopeless, at least as long as one is willing to endorse Direct Reference. The same, arguably, goes for (WIF). Anyway, to see exactly where the problem is, it may be useful to make an attempt. A prima facie good candidate, as an intermediate step toward a justification of (WIF), is something like: (15) A genuine proper name N occurring in a work of fiction W stops referring to its customary referent if W puts on the referent of N conditions that the customary referent of N cannot conceivably (or logically, or metaphysically) satisfy. In turn, one can try to justify (15) by an appeal to some linguistic data. Consider this battery of examples, and imagine that they appear in works of fiction written by authors who meant to refer to Gottlob Frege: (16) Gottlob Frege is Professor of Cultural Studies at the University of Hawaii… (17) Gottlob Frege is an Australian aboriginal… (18) Gottlob Frege is a man and is not a man… (19) Gottlob Frege is a giant frankfurter sausage… Suppose, further, that the stories in which these sentences occur keep portraying the referent of “Frege” in the same vein. As an example,

232

Reference in Fiction

suppose that the story including (19) proceeds by saying that Frege is better with mustard than with ketchup, that boiled he is more digestible than roasted and so on and so forth. Even if there may be some vagueness, let us assume that we cannot conceive that Gottlob Frege, as we know him, is a sausage. And let us assume that, for this reason, we do not interpret the “Gottlob Frege” inscription in (19) as referring to the famous logicist. One may try and explain this alleged datum by claiming, with Kendall Walton and others, that fictional statements are implicit prescriptions to imagination. Now one may argue as follows. We are simply unable to interpret (19) except as a prescription to imagine that there is a sausage whose name is “Gottlob Frege”. It is plainly impossible to imagine that Gottlob Frege, as we know him, is a giant sausage. Thus, one may be tempted to say, Innocence of Fiction holds as long as it does not interfere with the function that works of fiction typically – or properly – have in communication, that is to prescribe imagining that what they state is true. If a prescription to imagine cannot be accomplished, then we immediately switch to the most similar accomplishable prescription. And if this latter prescription is not about, say, Frege, then the name “Frege”, as it occurs in the fiction, does not refer to its customary referent. In other words, a fiction, by virtue of its very communicative function, may require us to renegotiate the semantic interpretation of some names. This thesis should be considered as a justification of (15). In my mind, this justification is a good one, but not for (15). Setting aside the obvious problem of specifying notions such as being conceivable or being impossible to imagine, I think that sentences like (19) prove nothing at all. Indeed, the very same reasoning we have applied to fiction may be restated for lies as well. One may claim that, by uttering (19), one is able neither to tell a lie about Gottlob Frege. Other things being equal, I am still unable to see any referential difference between (19) as a fiction and (19) as a lie. And lies are typically regarded as nonfiction. So, the point has, in my mind, nothing to do with fiction after all. The point is that if the name “Frege” as it occurs in (19) is not interpreted as referring to Frege – no matter whether (19) is fiction or lie – it is because we are not able to recognize the relevant intentions of the speaker. If we take the term “semantics” in a somewhat restricted sense, this is a pre-semantic question, regarding how an

Giuseppe Spolaore

233

utterance of a name is to be interpreted15. On the other side, Innocence of Fiction is a semantic thesis, regarding how a name semantically behaves in certain contexts given that it has the semantic value it has. To see what I mean, assume that you are not able to interpret “Frege” – as it occurs in (19) – as referring to the German philosopher. In this case you would coherently deny that the story speaks of Gottlob Frege. Now, let us assume that “Hesperus” and “Phosphorus”, as they occur in Two Planets, do not refer to Venus. Now, suppose that Andrea Bottani, pointing at Venus, says: (20) Two Planets speaks of that planet. If neither “Hesperus” nor “Phosphorus” referred to Venus, why would it be so natural to agree with Andrea? Notice that this is a serious problem for artifactualism. Indeed, it is true that the intuition that Two Planets speaks of Venus may be regarded, at least in part, as theory-laden. But artifactualists are committed to a theory (i.e. Direct Reference) that grounds such intuitions and takes them very seriously16. At the end of the day, qualifying Innocence of Fiction can hardly be considered as a viable solution for artifactualists. And rejecting Innocence of Fiction would lead to a radical change of theory. On the other hand, the radical rejection of (RIF) (“If in a work of fiction reference is made to a real individual, that individual is a character of the work of fiction”) as long as one subscribes Innocence of Fiction, implies a rejection of the whole objectualist conception of characters. So, I am not taking this possibility into account. Instead, I

15 See e.g. Kaplan 1989: 558-563, Perry 2001: 102-105. 16 There is a last point that is worth noticing. Suppose that Two Planets includes

(‡) Hesperus and Phosphorus are distinct individuals. If “Hesperus” and “Phosphorus” receive a novel semantic interpretation when used in fiction, then (‡) is straightforwardly true. This is quite surprising. Of course, one wants to say, if a so-called principle of poetic licence is true, then the author rules as long as what is true-according-to-the-story is concerned – whatever this should mean. But this is not tantamount to saying that an author rules on what is actually true. And claiming that (‡) has no truth value (for whatever reason) is hardly compatible with the general semantic framework of artifactualism.

234

Reference in Fiction

shall consider the thesis that two distinct sorts of characters are involved in the problem of Two Planets. §4. The ambiguity of “character” Artifactualists might stick with Innocence of Fiction and buy from Terence Parsons (who is not an artifactualist) a distinction between two sorts of characters of a work of fiction, “immigrant” and “surrogate”. So, they might say that Two Planets has one immigrant character, the real planet Venus, and two surrogate characters, Hesperus and Phosphorus. Surrogate characters are abstract artifacts. There is of course a problem here, which may be put in the form of a dilemma. Either Two Planets refers simultaneously to both one and two objects by the names “Hesperus” and “Phosphorus”, or surrogate characters are not characters at all, since they do not fit the objectualist characterisation of the term “character” (that is more or less “any object the story speaks of”). The first horn is unbelievable; the second implies that (1) (“Hesperus and Phosphorus are distinct characters of Two Planets”) is false after all. This difficulty shows that it is not enough to postulate two sorts of objects. To solve our problem an artifactualist has to postulate an ambiguity between two notions of character as well. So let us assume that “character” is ambiguous. Let us assume that an objectualist has to grant that at least in one sense, the word “character” denotes an object a story speaks of. Since it would be absurd to claim that the same story, by the very same individual name, speaks contemporarily of two different objects, one has to postulate a nonobjectual notion of a character as well. Thus, at least in one sense “character” must denote an object somehow determined, but not referred to, by a story. I shall use the term “O-character” for the objectual sense and “N-character” for the non-objectual sense of “character”. Now notice that, if there is exactly one individual that is the referent of “Hesperus” and “Phosphorus”, then there is exactly one individual the story speaks of by those very names. So, (21) Hesperus and Phosphorus are the same O-character of Two Planets. Armed with the thesis that “character” is ambiguous, one may now claim that sentences about the number of the characters of a story such

Giuseppe Spolaore

235

as Two Planets should be intended as claims about the number of the N-characters of the story. But now one may cast some doubts on the entire objectualist proposal. Indeed, in this novel framework it is true that (22) When two actually co-referring names as “Hesperus” and “Phosphorus” are used in a story, we have to admit – no matter what the story says – that “Hesperus” and “Phosphorus” stand for exactly one O-character of the story. But, if so, there seems to be no more ground to deny that: (23) When an actually non-referring name such as “Sherlock Holmes” occurs in a story – no matter what the story says – “Holmes” stands for no O-character of the story. Why, indeed, should considerations about the actual reference of names be relevant only if such names are co-referential, and not if they are non-referential? At least to my mind, there is no intelligible answer to such a question in an artifactual framework. Notice that now the fact that we do say, for example (24) There are fictional characters, is no longer an evidence of the existence of O-characters, since the existence of N-characters is sufficient to explain the intuition expressed by (24). So, one may conclude, as long as one is willing to adopt objectualism and (IF), one has – at least partially – to subscribe non-objectualism; but once one has in part adopted non-objectualism, there seems to be very little evidence left for a realism adopting objectualism. Now there is just one way out left for artifactualism, i.e. denying (1) and claiming that Hesperus and Phosphorus are one and the same character of Two Planets. But this is quite an implausible move for any realism. Further, in this case, one may argue for antirealism along the lines we have seen so far [see (22), (23)]. As always in philosophy, one cannot say that the problem of Two Planets proves that artifactualism, as it stands, is false and is not going to be resurrected. For example, one may theorize a peculiar form of referential shift, or some peculiar form of ambiguity. But I think that it is a severe threat, and artifactualists should take it seriously into account. More in general, since I have made no use of creationism, I

236

Reference in Fiction

think that our puzzle is a serious problem for every realism including (IF) and (RIF). Now let us briefly sketch some more general remarks about this problem and a diagnosis. §5. Concluding remarks and a diagnosis The problem of Two Planets is rather peculiar, since, while resembling other very well-known puzzles about identity and opaque contexts, it does not involve any assumption regarding substitutionality or any explicit appeal to intentional contexts. I think that this peculiarity depends entirely on the notion of a character; but this is a point I have not addressed here. It is quite natural to think that the puzzle arises because artifactualists let “Phosphorus” and “Hesperus” be co-referential in the story too, and that the solution should be a denial of Innocence of Fiction. My diagnosis is different: artifactualists confuse fact and fiction. In this particular case, they confuse semantic facts with semantic fictions. On the one hand, there are the semantic properties a story really has, i.e. the semantic values of the terms and sentences occurring in it. From the view-point of direct reference, this means, for instance, that “Hesperus” and “Phosphorus”, as they occur in Two Planets, are coreferential; that (4) (“Hesperus and Phosphorus are distinct planets”) is false both in Serious Planets and in Two Planets; that “Sherlock Holmes”, as it occurs in Doyle’s novels, is empty; and so on and so forth. On the other hand, there are the semantic properties one has to ascribe to the story to assume – or pretend, or make believe – that it is true. One should never confuse what the story asks us to do with what the story really says. And, as a matter of fact, we never do it when truth is concerned. So why should we do it when co-reference – or, if you like, identity – is? This I take to be a general point, regarding whatever theory of fiction, even if, obviously, it can lead to different conclusions depending on what semantic value your favorite semantic theory assigns to the relevant linguistic items occurring in a work of fiction. References Brock, S., (2002), ‘Fictionalism about Fictional Characters’, Nous, 36, pp. 1-21.

Giuseppe Spolaore

237

Deutsch, H., (2000), ‘Making Up Stories,’ in (eds.) Everett and Hofweber, Empty Names, Fiction, and the Puzzles of Non-Existence, Stanford, CSLI, pp. 149181. Gale, R., (1971), ‘The Fictive Use of Language’, Philosophy, 46 pp. 324-340. Currie, G. (1986), ‘Works of Fiction and Illocutionary Acts’, Philosophy and Literature, 10, pp. 304-308. –– (1990), The Nature of Fiction, Cambridge, Cambridge University Press. Kaplan, D., (1989), ‘Demonstratives’, in (eds.) J. Almog, J. Perry, H. Wettstein, Themes from Kaplan, Oxford, Oxford University Press, pp. 481-563. Kroon, F., (2001), ‘Fictionalism and the Informativeness of Identity’, Philosophical Studies, 106, pp. 197-225. Martinich, A., (2001), ‘A Theory of Fiction’, Philosophy and Literature, 25, 1, pp. 96-112. Ohmann, R., (1971), ‘Speech Acts and the Definition of Literature’, Philosophy and Rhetoric, 4, pp. 1-19. Parsons, T., (1980), Nonexistent Objects, New Haven, Yale University Press. Perry, J., (2001), Reference and Reflexivity, Stanford, CSLI. Predelli, S. (1997), ‘Talk about Fiction’, Erkenntnis, 46, pp. 69-77. Quine W., (1948), ‘On What There Is’, Review of Metaphysics, 2, pp. 21- 38. Salmon, N. (1998), ‘Nonexistence’, Nous, 32, pp. 277-319. Searle, J., (1974), ‘The Logical Status of Fictional Discourse’, New Literary History, 6, pp. 319-332. Thomasson, A., (1999), Fiction and Metaphysics, Cambridge, Cambridge University Press. Walton, K., (1990), Mimesis as Make-Believe, Cambridge, Mass., Harvard University Press.

Philosophische Forschung Philosophical Research Volume one

Georg Meggle (Ed.) Social Facts & Collective Intentionality ISBN 3-937202-08-0, Hardcover, 478 pp., EUR 138,00

Volume three

(Ed.) Georg Meggle Ethics of Terrorism & Counter Terrorism ISBN 3-937202-68-4, Hardcover, 410 pp., EUR 98,00

Social Facts & Collective Intentionality: the combination of these two terms refers to a new field of basic research. Working mainly in the mood and by means of Analytical Philosophy, at the very heart of this new approach are conceptual explications of all the various versions of Social Facts & Collective Intentionality and the ramifications thereof. This approach tackles the topics of traditional social philosophy using new conceptual methods, including techniques of formal logics, computer simulations and artificial intelligence.

We are supposed to wage war against Terrorism – but exactly what we are fighting against in this war, there is nearly no consensus about. This volume is an analytical attempt to end this disaster. What is Terrorism? Are terrorist acts to be defined exclusively on the basis of the characteristics of the respective actions? Or should we restrict such actions to acts performed by non-state organisations? And, most important, is terrorism already by its very nature to be morally condemned?

Der zweite Band der Reihe Philosophische Forschung spannt zwei Kerngebiete der Analytischen Philosophie zusammen: die Semantik und die Ontologie. Was sind die Grundbausteine unserer Ontologie? Wie beziehen wir uns sprachlich bzw. geistig auf sie? Diese und weitere Fragen werden von international renommierten Philosophen aus historischer und systematischer Perspektive diskutiert.

Over the last two decades foundationalism has been severely criticized. In response to this various alternatives to it have been advanced, notably coherentism. At the same time new versions of foundationalism were crafted, that were claimed to be immune to the earlier criticisms. This volume contains 12 papers in which various aspects of this dialectic are covered.

ISBN 3-937202-43-9, Hardcover, 445 pp., EUR 93,00 ISBN 3-937202-70-6, Hardcover, 293 pp., EUR 89,00

Semantik und Ontologie Beiträge zur Forschung Mark Textor (Hrsg.) Mark Siebel

Volume two

Basic Belief and Basic Knowledge Papers in Epistemology (Eds.) René van Woudenberg Sabine Roeser, Ron Rood

Series edited by Johannes Brandl • Andreas Kemmerling Wolfgang Künne • Mark Textor ontos verlag International Publisher Frankfurt • Paris • Ebikon • Lancaster • New Brunswick www.ontosverlag.com

Volume four