Grazer Philosophische Studien: Internationale Zeitschrift Für Analytische Philosophie

Grazer Philosophische Studien INTERNATIONALE ZEITSCHRIFT FÜR ANALYTISCHE PHILOSOPHIE

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

VOL 76 - 2008

Amsterdam - New York, NY 2008

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

The paper on which this book is printed meets the requirements of “ISO 9706:1994, Information and documentation - Paper for documents Requirements for permanence”. Lay out: Thomas Binder, Graz ISBN: 978-90-420-2420-5 ISSN: 0165-9227 © Editions Rodopi B.V., Amsterdam - New York, NY 2008 Printed in The Netherlands

INHALTSVERZEICHNIS

TABLE OF CONTENTS

Abhandlungen

Articles

Mark STEEN: Chisholm’s Changing Conception of Ordinary Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

Javier KALHAT: Structural Universals and the Principle of Uniqueness of Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

57

Miloš ARSENIJEVIĆ & Miodrag KAPETANOVIĆ: The “Great Struggle” between Cantorians and Neo-Aristotelians: Much Ado about Nothing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

79

Benedikt LÖWE & Thomas MÜLLER: Mathematical Knowledge is Context Dependent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

91

Kristoffer AHLSTROM: Epistemology and Empirical Investigation . . .

109

Daniel WHITING: The Use of “Use” . . . . . . . . . . . . . . . . . . . . . . . . .

135

Achim LOHMAR: The Failure of Pure Cognitivism . . . . . . . . . . . . . . .

149

Gerhard ERNST: Der Sinn für Schönheit . . . . . . . . . . . . . . . . . . . . . .

167

Diskussionen

Discussions

Darrell P. ROWBOTTOM: An Alternative Account of Epistemic Reasons for Action: In Response to Booth . . . . . . . . . . . . . . . . . . . . . . . .

191

Benjamin SCHNIEDER: Further Remarks on Property Designators and Rigidity (Reply to López de Sa’s Criticisms) . . . . . . . . . . . . . .

199

Essay-Wettbewerb

Essay Competition

1. Preis: Lars DÄNZER: A Neglected Argument for Compatibilism . . .

211

2. Preis: Anselm SPINDLER: Über moralische Verantwortung und alternative Möglichkeiten . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

219

3. Preis: Andreas MAIER: Weeding in the Garden of Forking Paths—Yet Another Look at Alternate Possibilities . . . . . . . . . .

228

Besprechung

Review Essay

Christian KANZIAN: Substanzen – Neue Perspektiven auf ein altes Thema . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Buchnotizen

237

Critical Notes

Wayne M. MARTIN: Theories of Judgment: Psychology, Logic, Phenomenology. Cambridge: Cambridge University Press, 2006. (Robin ROLINGER) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

247

Anna SIERSZULSKA: Meinong on Meaning and Truth. Frankfurt et al.: Ontos Verlag, 2005. (Venanzio RASPA) . . . . . . . . . . . . . . . .

250

Graham PRIEST: Towards Non-Being. The Logic and Metaphysics of Intentionality. Oxford: Clarendon Press, 2005. (Maria REICHER)

255

Marie McGINN: Elucidating the Tractatus: Wittgenstein’s Early Philosophy of Language and Logic. Oxford: Oxford University Press, 2006. (Denis McMANUS) . . . . . . . . . . . . . . . . . . . . . .

259

John GIBSON, Wolfgang HUEMER (Hg.): Wittgenstein und die Literatur. Frankfurt am Main: Suhrkamp Verlag, 2006. (Klaus PUHL) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

263

Michael STÖLTZNER, Thomas UEBEL (Hg.): Wiener Kreis. Texte zur wissenschaftlichen Weltauffassung. Hamburg: Meiner Verlag, 2006. (Thomas MORMANN) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

268

Daniel COHNITZ, Marcus ROSSBERG: Nelson Goodman. London: Acumen, 2006. (Georg PETER) . . . . . . . . . . . . . . . . . . . . . . . .

275

Dean ZIMMERMAN (ed.): Oxford Studies in Metaphysics, Volume 2, Oxford: Oxford University Press, 2006. (Dale JACQUETTE) . . . .

279

Christian BEYER: Subjektivität, Intersubjektivität, Personalität. Ein Beitrag zur Philosophie der Person. Berlin: De Gruyter, 2006. (Uwe MEYER)

286

Peter KÜGLER: Übernatürlich und unbegreifbar. Religiöse Transzendenz aus philosophischer Sicht. Wien: Lit-Verlag, 2006. (Daniel VON WACHTER) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

290

Eingelangte Bücher / Books Received . . . . . . . . . . . . . . . . . . . . . . . . .

295

Grazer Philosophische Studien 76 (2008), 1–56.

CHISHOLM’S CHANGING CONCEPTION OF ORDINARY OBJECTS Mark STEEN St. Louis University Summary Roderick Chisholm changed his mind about ordinary objects. Circa 1973–1976, his analysis of them required the positing of two kinds of entities—part-changing ens successiva and non-part-changing, non-scatterable primary objects. This view has been well noted and frequently discussed (e.g., recently in Gallois 1998 and Sider 2001). Less often treated is his later view of ordinary objects (1986–1989), where the two kinds of posited entities change, from ens successiva to modes, and, while retaining primary objects, he now allows them to survive spatial scatter. Also (to my knowledge) not discussed is why he changed his mind. This paper is mostly intended to fill in these gaps, but I also give some additional reasons to prefer Chisholm’s later view. Also, I discuss how mereological essentialism can be further defended by how it informs a theory of propertyinherence which steers between the excesses of the bare particularists and bundle theorists.

1. Review of Chisholm’s earlier position 1.1 Introduction Chisholm’s views on the status of ordinary objects can be divided into two major stages. In the first stage, from around 1973 to 1976, he held a somewhat restricted mereology, and believed that most problems of material constitution and persistence are brought about by not realizing the following two things. First, in addition to the strict and philosophical sense of ‘identity’, there is a loose and popular sense of ‘identity’, and the two often get conflated. Second, real objects have all of their parts essentially. According to this view, ordinary part-changing objects are logical constructions or ‘fictions’, constructed out of (or reduced to) a succession

of objects which do not change their parts.1 In the second stage, from 1986–1989, Chisholm maintained mereological essentialism as well as the dichotomy of loose versus strict identity. However, he believed in a far less restricted mereology, and instead of constructions or fictions, he identified ordinary objects with reified modes of a succession of mereologically stable bare objects. These modes are not fictions constructed out of mereologically stable objects; they are rather genuine entities which ‘pass through’ a succession of mereologically stable masses of matter which are the substrates of the modes.2 The second view is superior, but underdeveloped. But first I will examine the original view in detail. The paper has three central sections. In section (1), I will show how Chisholm’s entia successiva account of ordinary objects is a sensible first-pass at solving some problems of material constitution. In section (2), I will show how Mereological Essentialism (‘ME’) has some additional support making it even stronger than Chisholm initially supposed, namely, that ME offers a strong alternative to the bundle and bare particular views of property inherence. Yet, his earlier account as it is cannot solve the paradox of coincidence. Also, unmodified, it is too close to four-dimensionalism, a position he is against. Also, in section (2) I will show how Chisholm’s earlier conception of objects is not as well-equipped as his later conception to solve as many problems. I will give some circumstantial support that Chisholm might have changed his mind about ordinary objects due to some of the concerns I raise (some of which were raised by Wiggins in 1979), but this is somewhat speculative. Lastly, in section (3), I show how his later account gets around some of the problems of the earlier view, and hence is an improvement upon it. 1.1 Identity strict and loose Chisholm’s earlier view is encapsulated in “Parts as Essential to Their Wholes,” “Mereological Essentialism: Some Further Considerations,” and Person and Object.3 I will mostly focus on Person and Object. 1. It is questionable whether a ‘logical constructions’ view is compatible with a ‘fictionalist’ view, as well as which kind of view Chisholm preferred. I believe he was not really interested in deciding this question at the time of writing Person and Object. 2. For a similar view, see Karmo 1977. For a discussion of Karmo’s 1977 and Chisholm’s 1986, see Zimmerman 1995. 3. Chisholm 1973, 1975, and 1976 (especially Chapter III and Appendix B), respectively.

2

Chisholm begins discussing identity over time in Chapter III of Person and Object by pointing out the puzzle of the Ship of Theseus.4 Suppose that a man named Theseus had a ship made entirely out of wood, with planks labeled 1–1,000. Let us call this ship, before any part changes, S1. Theseus’ ship comes into port twice a year, and each time has twenty planks replaced with aluminum ones. After twenty-five years, we have a ship S2, which we most likely would say is the same ship, made of completely different parts. But, the Hobbesian twist is that,5 let us suppose, the shipwright who replaced the boards took the original planks and arranged them so that they are identically placed as they were with S1, so that after twenty five years the shipwright has made a ship S3 which is qualitatively identical to S1 and has all the same parts. Both S2 and S3 are good candidates for being identical with S1, but if they both are, then we have the absurd consequence that each of two ships is identical to one ship. This much is obvious: were it not for S3, we would say that S2 is identical with S1. Similarly, if S2 did not exist, we would find no fault with the claim that S3 is identical with S1. But why should what occurs with another ship determine whether this ship is identical with S1? Chisholm attempts to solve this puzzle and others by pointing out the distinction between strict and loose identity, and defending mereological essentialism. For Chisholm, these notions go hand in hand towards constructing a theory of ordinary objects and their persistence conditions. We can clarify the notion of loose versus strict identity by looking at several ways Chisholm says that we play ‘fast and loose’ with identity talk. Sometimes we play fast and loose with identity when we say things like “Route 6 is Point Street in Providence and is Fall River Avenue in Seekonk” (1976, 93). Since Fall River Avenue is not Point Street, they cannot both be identical to Route 6. In these kinds of cases we use an apparent ‘is’ of identity as short for more clear but unwieldy locutions. Similarly, we play fast and loose with identity talk when we employ façon de parler fusion and fission talk, such as when we say “This train will be two trains after Encinitas”, or, “Those two trains will be one train after Leucadia.” Sometimes we also fudge things a bit when we conflate a description, whose There is considerable overlap in the three. The content of 1976 contains most of that in 1973 and 1975. When I refer to 1973 I will use the page numbers from Chisholm 1989, where 1973 is re-printed. 4. The original appearance of this puzzle is in Plato’s Phaedo, 58a. My description of the puzzle is not exactly the same as either Plato’s or Chisholm’s. 5. Concerning Body Chap IX Sxn 7.

3

referent can change, with a rigid designator. For instance, one who hears “The President of the United States was Eisenhower in 1955, and Ford in 1975,” who is not aware of certain conventions can be forgiven for thinking that there is this entity, the President, who was identical with Eisenhower in 1955 and Ford in 1975 (Ibid.). Lastly, a common error is confusing numerical with qualitative identity, or types with tokens, such as when I find someone with a mandolin and I say “you play the same instrument I play.” If my hearer insists that, no, it is his instrument, and I have never played it, he’s made an obvious error. So much is easily agreed upon. The folk sometimes use ‘as-if ’ identity talk that is not strictly speaking talk of identity (and the same goes for the ‘is not’ of difference). Chisholm courts controversy, however, when he compares these ‘fast and loose’ uses of identity with a very common usage, when we attribute identity to things across time that have changed their parts. Bishop Butler once “suggested that it is only in ‘a loose and popular sense’ that we may speak of the persistence of such familiar things as ships, plants and houses. And he contrasted this ‘loose and popular sense’ with ‘the strict and philosophical sense’ in which we may speak of the persistence of persons” (Ibid.). 1.2 Mereological essentialism: philosophical motivations and historical support Thomas Reid, like Bishop Butler, also held that the persistence of persons is a paradigm instance of the persistence of a substance or thing, whereas that of tables or chairs, which change their parts, is not genuine persistence: The identity of a person is a perfect identity; wherever it is real, it admits of no degrees … for this cause, I have first considered personal identity, as that which is perfect in its kind, and the natural measure of that which is imperfect. (Reid 1854, 345. Quoted in Chisholm 1976, 89) All bodies, as they consist of innumerable parts that may be disjoined from them by a great variety of causes, are subject to continual changes of their substance, increasing, diminishing, changing insensibly. When such alterations are gradual, because language could not afford a different name for every different state of such a changeable being, it retains the same name, and is considered as the same thing. Thus we say of an old regiment that it did such a thing a century ago, though there now is not a man alive who then belonged to it. We say a tree is the same in the seed-bed and in the forest.

4

A ship of war, which has successively changed her anchors, her tackle, her sails, her masts, her planks, and her timbers, while keeps the same name is the same. (Reid 1854, 346. Quoted in Chisholm 1976, 96)

Bodies are to be identified with all of their parts, and it is only in a loose and popular way that we say that a person has the same body when they have changed their parts. Many have agreed. David Hume, for instance, said “all objects, to which we ascribe identity, without observing their invariableness and uninterruptedness, are such as consist of a succession of related objects” (Hume’s Treatise, I.iv.6, Selby-Bigge edition, 255. Quoted in Chisholm 1976, 211). Bodies have all of their parts essentially. When we talk about a thing that has changed its parts yet remained the same, we are actually talking about a succession of objects. As Hume says: … suppose any mass of matter, of which the parts are contiguous and connected, to be plac’d before us; ‘tis plain we must attribute a perfect identity to this mass, provided all the parts continue uninterruptedly and invariably the same, whatever motion or change of place we may observe either in the whole or in any of the parts. But supposing some very small or inconsiderable part be added to the mass, or substracted [sic] from it; tho’ this absolutely destroys the identity of the whole, strictly speaking; yet as we seldom think so accurately, we scruple not to pronounce a mass of matter the same, where we find so trivial an alteration. The passage of the thought from the object before the change to the object after it, is so smooth and easy, that we scarce perceive the transition, and are apt to imagine, that ‘tis nothing but a continu’d survey of the same object. (Hume’s Treatise, I.iv.6, Norton and Norton edition, 167)

As Chisholm points out, “Abelard held that ‘no thing has more or less parts at one time than at another’(Quoted from Henry 1972, 120). Leibniz said “we cannot say, speaking according to the great truth of things, that the same whole is preserved when a part is lost” (see Chisholm 1976, 145). A great many philosophers have embraced this doctrine, frequently called mereological essentialism. Mereological essentialism has some intuitive appeal, but has an immediate counterintuitive consequence. Particular bodies, or masses of matter, cannot change parts; commonsense objects, such as tables and chairs, can change parts; so, commonsense objects are not particular bodies or masses of matter. But what are commonsense objects then? Chisholm answers, along with Hume, that commonsense objects are successions of distinct objects that it is convenient for us to ‘feign identity’ about. A chair at t1

5

which has lost a part by t2, is only identical to the later chair in a ‘loose and popular sense’, but is, strictly speaking, distinct. Why would anyone accept mereological essentialism (‘ME’), and why in particular does Chisholm? What exactly does the doctrine amount to, and what is the precise relation between the commonsense objects and the ‘mereologically inflexible’6 ones? Unfortunately, Chisholm is much clearer about how we can hold ME and make it workable, rather than why we should regard it as true in the first place.7 Chisholm, in general, never spends much time defending why we should identify an object with its parts, except to point out the problems with mereological inessentialism (which we will get to), and what we can see in one short passage.8 I will try to get clear on why Chisholm thinks it is true, which we can only glean by perusing a wide variety of Chisholm’s material. But first let us examine how Chisholm develops ME to clarify the relation between commonsense part-changing objects and the more philosophical, mereologically inflexible ones. 1.3 Entia successiva and entia per se Chisholm says that the principle of mereological essentialism is the following: … for any whole x, if x has y as one of its parts then y is part of x in every possible world in which x exists. The principle may also be put by saying that every whole has the parts that it has necessarily, or by saying that if y is part of x then the property of having y as one of its parts is essential to x. If the principle is true, then if y is ever a part of x, y will be part of x as long as x exists. (1976, 145)

The principle does not, of course, entail that a part y is necessarily a part of x, or needs to be a part of x in order to exist. In order to explicate the difference between the parts of mereologically inflexible and commonsense or ‘mereologically incontinent’9 objects, we need the distinction between a ‘strict part’ versus a ‘loose part’.10 Chisholm 6. Chisholm introduces this phrase in 1976 for objects which cannot change parts. 7. Although by answering the how he lessens the negative pull of the why question. 8. Here is the passage: “There is in its favour: a certain intuitive plausibility; the support of an impressive philosophical tradition; and the fact that it enables us to deal with what otherwise seems to be insoluble philosophical puzzles.” 1976, 151. 9. Zimmerman coins this phrase in 1995. 10. Note that Chisholm does not say anything about what a loose-part is except that they are not strict parts, i.e., they are parts the folk think an object can lose or gain yet retain its identity.

6

uses the term ‘S-part’ for strict part. With the phrase ‘S-part’, we mean by ‘part’ what is often meant by ‘proper part’, or, a part of something that is distinct from that (entire) thing. Chisholm presents three initial axioms that give us transitivity and asymmetry for S-part, and ME. (A1) If x is an S-part of y and y is an S-part of z, then x is an S-part of z. (A2) If x is an S-part of y, then y is not an S-part of x. (A3) If x is an S-part of y, then y is such that in every possible world in which y exists x is an S-part of y. The fourth axiom Chisholm presents is controversial even to many people who identify themselves as mereological essentialists: (A4) For every x and y, if x is other than y, then it is possible that x exists and y exists and that there is no z such that x is an S-part of z and y is an S-part of z.11 This principle, which we could call ‘The Contingency of Wholes’ is an explicit rejection of unrestricted mereology.12 The principle of unrestricted mereology (‘UM’) or collectivism is that, for any distinct x and y, there exists a z such that x and y compose, or ‘make up’ z. UM is itself controversial, since it states that, for any two individuals, regardless of their temporal or spatial spread, they ‘fuse’ together to compose a ‘third’13 individual. If UM is correct, then there is an individual made up of the electron in my nose furthermost from my center of gravity and the last dinosaur. 1.4 Chisholm’s restricted mereology and an answer to the Special Composition Question First, let me introduce some rough-and-ready terminology. By ‘bare objects’ let us mean what have been variously referred to as ‘masses,’14 or ‘fusions’ 11. All four principles are in 1976, 151. By ‘other’ in ‘x is other than y’ Chisholm seems to mean either completely disjoint or partially overlapping. 12. See Wiggins 1979, 302. 13. I put the scare quotes because it is an open question whether composition is identity. If it is, the ‘third’ thing is not really anything in addition that we are ontically committed to beyond the two composing items. See Lewis 1991, 72–87 and Wallace ms for details on this debate. 14. The term ‘masses’ comes from Zimmerman 1995. These entities are also called ‘aggregates-’ (Burge 1977), ‘parcels-’ (Locke’s Essay), ‘collections-’(Wiggins *), ‘consignments-’ (Karmo 1977), and ‘quantities-’ (Cartwright 1974 and 1979)—of matter.

7

of matter such that they persists just so long as all of their constituent parts do, regardless of whether they fall under a commonsense sortal (e.g., ‘car’, ‘cat’). By ‘partially nude objects’ let us mean fusions which persists just so long as all of their parts are stuck together to form objects of commonsense (where we include contiguous ‘pieces’ or ‘bits’ under the extension of ‘commonsense object’). Both bare and partially nude objects are metaphorically naked in the sense that their persistence conditions are unencumbered by the clothing of natural- and artifactual-kind Aristotelian substance sortals, and hence are somewhat immodest in their persistence conditions. The main difference between the two kinds of objects is that bare, but not partially nude, objects can survive scatter. UM is one answer to the question posed by Peter Van Inwagen called the ‘Special Composition Question’ (‘SCQ’), namely, when is it true that there exists something such that some distinct things compose it?15 Some rivals of UM answer: ‘never’16 (nihilism), ‘when the things compose an organism’17 (organicism), ‘whenever we intuitively think some things do’18, ‘it is a brute, unexplained fact when some things compose something’19 (brutalism). Chisholm, while not explicitly addressing this question (he wrote Person and Object eighteen years before Van Inwagen’s Material Beings), has an answer as well. Two things compose a thing when they are strictly joined: (D.B.3)

x is strictly joined with y =Df There is a w such that w is strictly made up of x and y.

But, when does the condition on the right obtain? We might say that two things are strictly joined if no third individual falls between them; then we could say that two things are joined if part of the one is strictly joined with part of the other. This would allow us to say that scattered subatomic particles may be parts of an individual thing. But we would not need to say that a suite of furniture separated by various objects is itself an individual thing. Axiom (A4) and these criteria allow us to say that some things that are not in direct or indirect physical contact may be parts of the same individual thing, but they do not require us to say that any two separated things are parts of one individual thing. (Chisholm 1976, 153) 15. 16. 17. 18. 19.

Van Inwagen 1990, 21–33 Dorr 2002. Van Inwagen 1990. Tacitly assumed by many such as Wiggins 1980. Markosian 1998.

8

This would be extremely informative, if we knew what counts as an individual for Chisholm. Regions of space? If so, then everything would be joined to everything else (but, since strictly joined and not joined is what is ontologically important and clarified, it is not clear whether this would be bad). If not, then any two things, no matter how distant, would compose a further thing if no third thing comes between them, which is a kind of thesis that Chisholm seems against at this period. He never tells us directly, so, it is a bit difficult to tease out exactly what Chisholm had in mind, and it wasn’t his intention to answer the then unasked SCQ. But, from this and other passages it seems that Chisholm held (at this period) that two or more things compose something just in case they are relatively compact and fall under some commonsense sortal, where we allow piece or piece of ____ to count as commonsense sortals as well (where piece here implies contiguity and ‘stuck-togetherness’ of its constituent parts. This sense of ‘piece’ is commonsense, and not the same as the aforementioned ‘parcel’, ‘aggregate’, or ‘fusion’ of the philosophers). As we’ll see, Chisholm held that if some bits of matter compose a hunk of matter that constitutes an object, that that hunk of matter goes out of existence when a tiny bit gets unattached and scattered from the (distinct) hunk of matter it used to be attached to. He does not believe that the former hunk of matter is still around, just with a piece no longer as contiguous with the rest as before (See Chisholm 1975, 481 and fn 4). To use our (not Chisholm’s) terminology, Chisholm does not believe in bare objects at this point. That is, he does not believe that there are aggregates which persist just so long as their parts do. He does, however, believe in what I call partially nude objects, that is, aggregates that ‘fuse’ or aggregate (verb) just so long as all of their parts are either joined or contiguous with each other in such a way that they fall under an Aristotelian substance sortal, including piece or hunk of stuff. Partially nude objects won’t survive the macroscopic discontiguity of their parts, unlike bare objects. As we’ll see, partially nude objects are an inferior and somewhat arbitrary postulate as compared to bare objects, and Chisholm’s acceptance of the partially nude, but not completely naked objects, gets him into trouble. Chisholm did, however, in the second stage that we will discuss, embrace bare objects. 1.5 Relation of primary to ‘vulgar’ objects Switching to Chisholm’s more tasteful usage, what we have called partially nude objects are what Chisholm calls primary objects (Chisholm 1989, 78).

9

Primary objects are either simple particles, gunk,20 or fusions of either which fuse according to the criteria laid out above. Primary objects are also mereologically inflexible; they have all of their parts essentially. For Chisholm, primary objects are the genuine objects, and commonsense or ‘vulgar’ putatively part-changing objects are logical constructions out of the former. Given ME, is there no mereological change? There is: If what I have said is correct, at least four types of mereological change are possible. The first two are coming into being and passing away; for wholes do come into being and pass away … And the second two types of mereological change are joining and disjoining. Objects may be joined together to form a whole that hadn’t previously existed. And objects may be disjoined from each other and, unlike the whole that they had formed, survive the change. (1976, 153)

When we except creation ex nihilo and annihilation into nothing of simples or any composite whole’s part(s), there are actually only two kinds of mereological change, since joining is equivalent to coming into being, and passing away is equivalent to disjoining. How do primary objects relate to vulgar ones? Let’s use Chisholm’s example of a very simple table, with only two (salient) parts—a stump and a board, one of which changes every day. Suppose on Monday that the table is made up of parts A and B, Tuesday, of B and C, and Wednesday, of C and D. Let us suppose that the changes the table goes through lead the folk to say that it is the same table throughout the time period, and let us further suppose that at no time through the part-changes would we suppose that there is not a table occupying the region occupied by at least two of the aforementioned parts. Is the table on Monday the same table as the table on Wednesday? ‘Yes’, say the folk (and most philosophers, let us presume). Is the table on Monday the same primary object as the table on Wednesday? ‘No’, says Chisholm (and other ME adherents). If, as Chisholm contends, there really are only primary objects, what is it that the folk are talking about? Chisholm calls the folk table an ens succesivum—“the ‘successive table’ that is made up of different things at different times” (1976, 98). The successive table and all other entia successiva, the chairs, cats, staplers and trees of common sense, are all logical ‘fictions’ or ‘constructions’ out of primary objects. But what is the relation, between, say, the table made up out of AB on 20. ‘Gunk’ refers to stuff which has proper parts, each of which also contain proper parts, ad infinitum.

10

Monday and BC on Tuesday? Table BC is a direct table successor of AB. y is a direct table successor of x just in case x is a table at t1, y is a table at t2, and, there is a z such that part of z is part of x at t1, and part of z is part of y at t2, and at every moment between t1 and t2 inclusive, z is itself a table.21 What about the relation between AB and CD? CD is not a direct table successor of AB, but it is a table successor. Roughly, y is a (non-direct) table successor of x just in case x is a direct table successor of some z which is either a direct table successor of y, or is a direct table successor of a direct table successor of y, or is a direct table successor of a direct table successor of y or … [repeated a finite number of times] … of y (Chisholm 1976, 99). With some tinkering, we could generalize what it is to be a _____-successor for any count sortal, and define any part-changing object as a series or succession of mereologically inflexible ones. Entia successiva, which are logical constructions, have their properties by proxy. For example, a successive table is blue (at t), iff some non-successive entity that stands in for it or ‘does duty for it’ is blue (at t). A successive table can prop up a plate and a glass only if a series of mereologically changeless, almost instantaneous tables, hold up a plate and a glass (or, rather, hold up a series of instantaneous plates and glasses).22 With our situation of the table from Monday to Wednesday, the question arises—how many tables total are there? Three? One? Four? Chisholm replies: In saying that there are exactly three tables in the situation described one is speaking in the strict and philosophical sense and not in the loose and popular sense. In saying that there is exactly one table one is speaking in the loose and popular sense and not in the strict and philosophical sense. But the statement that there are four tables—AB, BC, CD and the successive 21. Chisholm’s official definition is as follows: D.III.1 x is at a t a direct table successor of y at t’ =Df (i) t does not begin before t’: (ii) x is a table at t and y is a table at t’; and (iii) there is a z, such that z is a part of x at t and a part of y at t’, and every moment between t’ and t, inclusive, z is itself a table. (1976, p. 99) I changed condition (iii), because it won’t work as it stands. If A is a stump, and B is a board, if we turn the successive table into BC by placing stump C on top of B, turn it over and remove stump A, the ‘table’ z (ABC) that links x and y is not a part of x or y. Rather, part of z (AB) is part of x (AB) (an improper part), and part of z (BC) is an (improper) part of y (BC). Z itself is a table in this case, though a rather odd one. 22. “(D.III.8) The successive table that is at place P at time t is F at t =Df There is exactly one thing at place P at t that constitutes a successive table at t and that thing is F at t.” (1976, 101)

11

table—is simply the result of confusion. One is trying to speak both ways at once (1976, 103).

1.6 Dealing with the puzzle cases How does Chisholm’s ME account deal with the puzzle cases? Some it deals with quite readily, while with others it does not fare so well. We’ll also see that ME can deal reasonably well with the standard types of objections to it. With the Ship of Theseus, the question ‘which ship is identical to S1; S2 or S3?’ loses its bite. S2 is loosely identical to S1, and related by shipsuccession. S3 is strictly identical to S1, and neither loosely identical to S1 nor strictly or loosely identical with S2. The confusion comes about when we believe that our commonsense loose concept of sameness over time is in contention with our strict ME intuitions. They are not. We are using distinct standards, and assume that there is only one standard. As long as we keep the standards clear in our minds we can avoid confusion. Chisholm can allow for conflicting intuitions and accept that both S2 and S3 are loosely identical to S1, but deny that a contradiction is entailed since loose identity is not transitive. Chisholm’s account also deals well with problems of ‘fusion’.23 Here is how Theodore Sider presents the puzzle: We begin with a cat, Tibbles, and a certain proper part of Tibbles, Tib, which consists of all of Tibbles except for the tail. Tibbles and Tib are obviously numerically distinct. But suppose now that Tibbles loses her tail; it seems that both Tibbles and Tib survive: Tib because nothing has happened to it beyond having something external to it detached, and Tibbles because cats, like trees, can survive the loss of certain parts … Tibbles and Tib are distinct; but they coincide after detachment. (Sider 2001, 142)

Chisholm’s account lets us resolve the difficulty and avoid saying either that two things became one, or that Tibbles (at least loosely speaking) went out of existence. Tibbles before the tail-severance is loosely identical to Tib after the severance, but is strictly speaking distinct. Tibbles’ tail is only an L-part of a successive cat, but an S-part of the hunk of matter that composes Tibbles. So, when Tibbles loses its tail, the strict and philosophical object goes out of existence. But now, by ‘Tibbles’ we refer to Tib, who is still in existence. Since Chisholm’s account can solve these and other 23. See Wiggins 1980, 209, and Rea 1997, xviii.

12

puzzles (e.g., the paradox of increase24), this gives us strong inductive support for Chisholm’s ME account by inference to the best explanation. 2. Further motivations for, and objections to, the Entia Successiva Account 2.1 Entia Successiva do not solve the Paradox of Coincidence What problems are there with Chisholm’s earlier account? One main problem is that, unmodified, it cannot solve the paradox of coincidence. Just in case you have been hiding under a rock, the paradox of coincidence is often elucidated as follows. Suppose you have a piece of clay. Call it “Piece”. Suppose you shape it into a statue, which you dub “Statue”. Then you squash the statue down, and call the resultant thing “Clay”. Now look at the following argument: (1) Piece = Statue (2) Piece = Clay (3) Statue ≠ Clay So, given the transitivity of identity, Statue ≠ Piece and Piece ≠ Clay. Furthermore, Piece ≠ Piece! Something’s gone wrong. The reasoning goes as follows. The Piece is identical to the Statue, since fashioning a statue out of a piece does not make the piece go out of existence. The Piece is identical to the Clay, since they are both the same piece of clay, and yet, Statue is not identical to the Clay, since you make the Statue go out of existence when you squash it. But then, absurd results follow. Many have argued that the culprit is premise (1). Piece is actually distinct from Statue. Supposing this runs us into all kinds of problems, as we now have two distinct objects in the same place at the same time (these problems are well documented elsewhere,25 so I won’t cover them here). Chisholm, while not explicitly discussing this problem in Person and Object, would likely have proposed a solution as follows. The foregoing argument only gets its force by not recognizing that ‘identity’ refers to (at least) two relations, ‘strict identity’ and ‘loose identity’. One is trying to talk both ways at once, in the strict and philosophical sense, and the loose 24. Chisholm 1976, 157. 25. E.g., in Zimmerman 1995.

13

and popular sense. Once we clear up the equivocation we can see that we can only treat (1)–(3) above as either of the following (where we use ‘S=’ to mean ‘strict identity’, which is just the plain old logician’s ‘identity’ and ‘L=’ to refer to ‘loose identity’, which is the commonsense ‘same as’ or ‘same F as’ which does not entail strict identity): (I) Piece S= Statue (II) Piece S= Clay (III) Statue S= Clay26 (1′) Piece L= Statue (2′) Piece L= Clay (3′) Statue L≠ Clay From (I)–(III) we obviously cannot conclude that the Statue is strictly distinct from Piece. Given Chisholm’s methods, since the Piece/Clay/Statue never changed parts, he would most likely regard them as all strictly speaking identical. From (1′)–(3′) we can only conclude that Piece is loosely distinct from Clay just in case ‘loose identity’ is transitive. And, not only should we not think, from all that Chisholm has said, that loose identity is transitive, but, since this is consistent with the Statue being strictly identical to Clay and Piece, it is not clear what the problem would be even if they were loosely distinct. Chisholm would most likely say that we have one primary object which constitutes different entia successiva at different times. Still, I don’t think Chisholm has solved the paradox of coincidence here, and this is somewhat indicated by the change in his metaphysics of ordinary objects by 1986 which is custom-tailored to deal with coincidence. At this point, I think we can see what the problem with his solution is at this point in time via the following argument.27 By the term ‘lump*’ let us mean a piece of clay which can survive part changes. The folk would certainly support the existence of such objects, and not object that we have the same lump* around after I tear off a tiny piece and annihilate it. Are lumps* reducible? Yes. Chisholm would say that part-changing lumps* reduce to, or can be explained in terms of, successions of primary objects which themselves do not change parts. 26. Of course, (III) is not a translation of (3) above. But, if we decide to disambiguate, this is the closest statement to number three that Chisholm would assent to in regards to how Statue and Clay stand to each other vis-à-vis strict identity. 27. Thanks go to a referee for Grazer for some very helpful comments on this section.

14

A key question here is what Chisholm thinks about reduction and the ontological status of the relata in the reduction relation. Let us say that the F’s reduce to the G’s. That is, we can say everything we want to about the F’s in terms of the G’s, without quantifying over F’s. Now, the key question is—do F’s exist? It seems we can go either way in how we answer this question. We can agree, for instance, that biological entities reduce to chemical entities, and chemical entities to elementary particles and their properties. But, it does not seem forced upon us to deny that there really are cells, mitochondria, or H2O molecules. Rather, fundamental particles in the right arrangements are just what cells, mitochondria, and H2O molecules are. But, we could be pushed to say that, since mitochondria are nothing ‘over and above’ their fundamental particles (or simples) in the right arrangements, that mitochondria do not, strictly speaking exist. There are just simples arranged ‘mitochondriacally’. Call the reductivist who thinks that, strictly speaking, the reducible relata or supervenient phenomena do not exist ‘pessimistic reductionists.’ Call the reductivist who believes that the reducible relata or supervenient phenomena do exist (and are to be identified with the subvenient base) ‘optimistic reductionists’. Now, here is the question. Is Chisholm an optimistic or pessimistic reductionist? I think the answer is that it is actually underdetermined by Chisholm’s writings whether he is an optimistic or pessimistic reductionist, although the evidence slightly favors holding him to be an optimist. Certainly Chisholm seems to often connote that part-changing objects do not really exist, which is what the ‘fiction’ label is supposed to indicate. But, Chisholm seems to contradict this when he says quite clearly that primary objects are (predicatively) ordinary objects which actually exist when they constitute them. “To say that there are four tables, in the strict and philosophical sense, is to say that there are four different things, each of them a table.”28 And: D.III.1 x is at t a direct table successor of y at t′ =Df (i) t does not begin before t′: (ii) x is a table at t and y is a table at t′; and (iii) there is a z, such that z is a part of x at t and a part of y at t ′, and at every moment between t′ and t, inclusive, z is itself a table.29

28. Chisholm 1976, 102. 29. Chisholm 1976, 99.

15

So, since primary objects ‘are’ tables etc. when they constitute them, it seems somewhat doubtful that Chisholm is a pessimistic reductionist. Hence it seems that he favors a more optimistic ‘logical construction’ view of ordinary objects, where ordinary objects must exist in some sense. I think it is difficult to make a strong case either way, but I think I can show that, whether or not Chisholm is an optimistic or pessimistic reductionist, in neither case does he satisfactorily solve the paradox of coincidence in the earlier period. Take our aforementioned lumps*. They reduce to non-part-changing partially nude objects. Do lumps* exist, at least in ‘some sense’? If so, then lumps* are definitely entia successiva, which, according to this interpretation, have some kind of being. If not, then either there are no referents for “lump*” or “lumps*,” and so all talk about them is strictly false (i.e., we embrace an error theory), or, while talk about them is strictly speaking false, talk of lumps*, like tables, is regulated by useful normative rules, so that some talk about ordinary objects is ‘true’ (or, at least true in the fiction of ordinary part-changing object discourse), other talk of ordinary objects—false (or, false in the fiction). So, if we accept this, we embrace some kind of ‘fictionalist’ theory.30 Let’s see first what the problems are for Chisholm if he is an optimist, and, secondly, a pessimist. If he is an optimist, then ‘Piece’ in the above argument for coincidence, either names an ens successivum (possibly-partchanging) lump* or a mereologically inflexible primary object. If Piece (and, by extension, Clay) is an ens successivum, then there can be two entia successiva in the same place at the same time, since the lump* (by hypothesis) is an ens successivum and is constituted by, throughout, the same primary object. Since the Statue is loosely distinct from the Clay (and possibly the Piece), since they have different persistence conditions and histories, then there are two entia successiva in the same place at the same time. The other option available for him is to treat instead Piece and Clay as primary objects. But, if Piece is a primary object, or a mereologically inflexible one, then either there can be no lumps*, or, there is also a lump, in which case we still have the same problem as mentioned above. And, there’s reason to think that Chisholm does believe in things like lumps*, so we do have the above problem. So, coincidence has not been satisfactorily avoided. 30. For an explanation of fictionalism in philosophy, see Eklund 2007.

16

Of course one obvious response is that coincidence of entia successiva is no problem. But, if entia successiva are logical constructions, and these constructions exist, then tables and lumps* exist. Merely calling them ‘logical constructions’ in no way eliminates the problems of coincidence (e.g., how could two objects be in the same place at the same time, while made up of the same matter in exactly the same internal relations?). So, the option of treating Piece and Clay as lumps*, and lumps* as real does not solve the problem of coincidence. He could try to further ameliorate the problem by treating these distinct entia successiva as sharing temporal parts, but then his position would seemingly collapse into four-dimensionalism, a position he decries (more on this later). But, Chisholm could take the pessimistic line. Tables and lumps* do not exist. But, does he seem more sympathetic to an error theory or a fictionalist theory? An error theory has the advantage in that all part-changing ordinary-object discourse gets the same universal treatment. There just are no part-changing objects, or entia successiva, and since non-existent things cannot coincide, there is no problem. The disadvantage of an error theory is that one must suppose that the folk are radically wrong about their object discourse, and that, strictly speaking, all of our talk about part-changing objects is false. And, Chisholm himself does not seem to favor a version of eliminativism combined with an error theory. In Person and Object, in too many places he seems to imply that there are norms and rules governing part-changing ordinary-object talk, and even talk of loose identity. It’s not anything goes, even though entia successiva do not exist in the same way that primary objects do. If I change my tire, I can’t say that my old car is (loosely) identical to my new house. It is neither loosely nor strictly identical, and is definitely both strictly and loosely distinct. So, it does seem that Chisholm holds that there are rules governing our fictional discourse (if he does indeed believe that part-changing ordinary object talk is part of a realm of fictional discourse), and, if so, then the problems of coincidence have not been solved, but merely pushed around like a bump in a carpet. This is because, if talk of lumps* and tables and so on is part of a fictional rule-governed area of discourse, that now, answering the problems of coincidence would amount to giving a story about how we can render coherent folk talk about lumps*, masses of matter, tables and so on. (If it can’t be rendered coherent, it would be best to go back to the error-theoretic option). But, Chisholm has given us no idea about how to make the folk talk which governs the use of ‘loose identity’ and its interactions with strict identity coherent. As I said above, one apparent way to dissolve the

17

coincidence argument is to disambiguate between the equivocal uses of ‘identity’, and show how when we do not equivocate in no case can the absurd conclusions follow. But, if we look back at the disambiguation above, where (1)–(3) get treated as follows: (1′) Piece L= Statue (2′) Piece L= Clay (3′) Statue L≠ Clay I argued that from these premises the absurd conclusions of (1)–(3) do not follow. But, what does follow? Is Statue loosely identical to Clay? Is Statue loosely distinct from Piece? I have no idea. We are given no way of understanding what loose identity and distinctness amounts to. But, if Chisholm embraces a fictionalist treatment of part-changing ordinary objects, then a solution to the problem of coincidence should be satisfied within this framework, or, shown how it does not arise in the fictionalist framework. But, Chisholm does not do this, and so the problem of coincidence is still outstanding in Chisholm’s treatment of ordinary objects around the time of Person and Object. The above problems of Chisholm’s in answering the paradox of coincidence are rooted in two features of his ontology: (1) He allows partially nude but not bare objects, and (2) he allows commonsense object sortals to give us criteria for identity of objects of any ontological kind, whether they be ‘loose objects’, ‘pieces,’ ‘fictions’ or ‘constructions’. We will see that something along the lines of his account fares better when we reject both (1) and (2) by asserting that there are bare objects. Chisholm’s assertion that there are masses of matter is good, but when he combines this notion with the persistence conditions of standard sortal essentialism, he ‘takes back’ the problem-solving features that (bare) masses of matter are particularly placed to bestow. In addition, as it stands, we will see how Chisholm’s account is too much like a stage-theoretic four-dimensionalism, and falls prey to the same kind of triviality objections he brings against the four-dimensionalist. 2.2 Problems Chisholm’s Mereological Essentialism can avoid But, before this, let us see what is not wrong with Chisholm’s account. As Chisholm notes, most objections to ME come in one of the following forms:

18

(A) My car had parts last week that it does not have this week. But this can’t be true if ME is, so ME is false. (B) My car could have had different tires than it in fact has. But, this cannot be true if ME is, so ME is false.31 Chisholm notes how we can dissolve these objections by careful disambiguation. With regards to (A), the ME’er can admit that a person can own a strict object which can get disjoined from one strict object, and joined to another. But, the ME’er will deny that when the car changes parts that we have one and the same (strict) object before us. Rather, the car now is a mereologically inflexible object, as was the previous car, and the later car is a car successor of the earlier car. The parts that changed were L-parts of the ens successivum car, but S-parts of each mereological whole which they constitute. Similarly, with (B), the ME’er can admit that the mereological whole car one now possesses can have a part, namely, all of the car except the tire, which could have been joined to a different tire. (In other words, ME does not apply to entia successiva). But that is not to admit that the very same (strict) object, a mereological whole, could have had different parts than it in fact does. These objections all arise by an equivocation over ‘part’. So, these objections ought not to move the mereological essentialist. But, on to a further main objection. What is the problem with objects which persist just so long as their parts are contiguous (namely, ‘partially nude objects’), as opposed to bare objects which persist just so long as all their parts do, regardless of (spatial) contiguity? For starters, supposing so entails either the contradictions noted above, or that we should give up on thinking that ‘loose’ identity has any coherent logical relations at all. Mainly, however, denying that there are bare objects undermines one main, implicit motivation for ME to begin with. 2.3 Motivating Mereological Essentialism as an aid in delineating a meaty substrata view of property inherence It is curious that Chisholm spends quite a bit of time laying out a mereological essentialist system, defending it from objections, but little time motivating ME intuitions in the first place. His main, explicit motiva31. Chisholm 1976, 154–155. These are responses to objections brought up by Plantinga 1975.

19

tions for ME are that it has “a certain intuitive plausibility; the support of an impressive philosophical tradition; and the fact that it enables us to deal with what otherwise seems to be insoluble philosophical puzzles” (Chisholm 1976, 151). This is kind of odd since most people think that ME is wildly counterintuitive, and the fact that many philosophers have thought something is not usually given as grounds for believing something, and, ME, at least in Chisholm’s form, seems to bring up as many problems as it solves. In this section I will try to show that Mereological Essentialism can be further defended by how it naturally guides us to a certain well-supported view of substrata and property-inherence that steers between the excesses of the ‘bare particular’ and bundle theorists.32 Furthermore, I will also show that there is some evidence in Chisholm’s writings that supports that these kinds of concerns were implicitly motivating him. But, whether or not the view of property-inherence I raise is one Chisholm would endorse, I think we can see how at least this gives us some (non-conclusive) reasons for favoring ME. Property realists owe us an account of how properties stand to the particulars which have them. Providing such an account would help explain the metaphysical underpinnings of predication. Such an account would help shed light on the all-important difference between objects and the properties they have. Lastly, providing such an account would also solve the problem of ‘individuation,’ or, how to explain (or, explain away) numerical difference in the face of exact qualitative similarity.33 Let us refer to theories that give an account of these various issues under the umbrella phrase ‘theories of property-inherence.’ Classically, in exegeses on the now tired dialectic, there seem to be two main kinds of theories of property-inherence— bundle theories, and substrata views.34 The substrata views are themselves divided into two main kinds, the ‘bare particular’ theories, and more meaty substratum accounts. Bundle theorists include in the fundamental level of reality only one kind of entity—properties. An apparent object, such as a red, round bowling-ball, is actually just the features red, round, being-three-holed, and so on, that are ‘compresent’ with one another. There is no non-quali32. I am especially grateful to a referee for Grazer who provided much helpful and needed prodding on this section, which lead to its being much improved. 33. There is not really one or the problem of individuation, but rather, several. This is one of them. 34. See Loux 1970, 1998, and 2002.

20

tative thing in addition to the properties, there are just the compresent properties. Substrata theorists, both bare particular theorists and the more meaty substratum theoreticians (let us refer to the latter as ‘substratum theorists’ in distinction to the bare particularists, from now on, even though bare particular theorists are, strictly-speaking, substratum theorists as well), hold that, in addition to properties, there must be a particular in which the properties inhere. A bowling ball has the properties of being red, round, and three-holed, but is not itself a mere exhaustive collection of its compresent properties. What bare particular theorists believe, but both meaty substratum and bundle theorists deny, is that, when one ‘subtracts away’ all the properties of a thing, that something is left over—a bare particular. Substratum theorists and bundle theorists both believe that bare particulars, i.e., things without properties that are the havers of properties, are incoherent. I agree. Furthermore, if bare particulars ‘in themselves’ or, essentially, have no properties of their own, then they do indeed seem extravagant.35 But, what the substratum theorists still insist on, but the bundle theorists deny—is that we need more than just properties to account for the nature of concrete particulars and to understand property-inherence. The bundle theorist has an advantage over the bare particularist in that they postulate no mysterious bare particulars, and respect empiricist principles, and have a more sparse ontology. The problem with the bundle theory, however, is that, to my knowledge, it has never been shown how the bundle theory can respond to this classic objection: the bundle theory entails that the Identity of Indiscernibles is true, but it is false. The Identity of Indiscernibles states that if some x and a y have all the same properties, then x is identical with y.36 This principle comes to grief when we employ a thought experiment thought up by Max Black.37 Imagine that there are two qualitatively identical but numerically distinct spheres in a radially symmetrical universe. For each property one sphere has, the other one has it as well. If objects are nothing more than collections of properties tied together by compresence, then this world is impossible, since these ‘two’ spheres have all the same 35. In Sider 2006 Sider gives a decent defense of bare particulars, but, what he says about them there seems to indicate that he actually believes more of a view like the ‘meaty substrata views’ I speak of below. 36. Another way of stating this objection is that the bundle theory conflates qualitative sameness and difference with numerical sameness and difference. 37. Black 1952.

21

properties. What is possible is for there to be one multi-located sphere, but, strictly-speaking, there cannot be two qualitatively identical objects. But, we know that there can be two qualitatively identical objects, so, the bundle theory is false. Some have tried to ameliorate this worry, but, none, as far as I can see, have done so satisfactorily.38 So, if the bare particular and bundle views of property-inherence are no good, then that leaves us with the more meaty substratum views. How does the selection of substratum views on offer fare? Not well. Either the views on offer are too flat-footed, or, while more rich, fail to satisfactorily explain what the non-qualitative component of particulars is. Peter van Inwagen, often in passing, denigrates the bare particular view. For instance, A chair cannot, for example, be a collection or aggregate of the properties ordinary folk say are the properties of a thing that is not a property, for a chair is not a collection or aggregate of all those things one could truly say of it … (I hope no one is going to say that if I take this position I must believe in ‘bare particulars’. A bare particular would be a thing of which nothing could be said truly, an obviously incoherent notion).39 Another perfectly meaningless term … would be ‘bare particular.’ A bare particular would either be what you get when you subtract the tropes from an ordinary concrete object (and thus the term would be meaningless), or else a thing of which nothing is true; and of course, the idea of a thing of which nothing is true makes no sense at all.40

This is fine, as far as it goes, since I don’t think one ought to embrace a bare particular view even if one rejects the bundle view, but, since van Inwagen does not lay out anywhere (to my knowledge) a more meaty substratum view, then we are left in the dark as to the non-qualitative component of concrete particulars. We are left embracing the need for a non-qualitative component of objects for them to inhere in, but are left in the dark as to what it could be. Call this the ‘flat-footed’ response. It doesn’t say, with 38. One can, of course, try to individuate the spheres because they occupy different regions. But this has just pushed the problem back. What individuates the regions? Each region has exactly the same properties as the other. If the regions are held to be brutely individuated, then regions are in fact playing the mysterious substratum role that the bundle theorists decry. (Cf. Martin, 1980, 8). For some relatively recent discussion of the problem, see Hawthorne 1995 and Zimmerman 1997, 39. van Inwagen, 2004, 135. 40. van Inwagen, 2006, 202.

22

Locke, that a substratum is a “I-know-not-what”, but rather, that it is a “I-know-what-but-I’m-not-going-to-tell-you.”41 One is left with the suspicion, however, that the latter is really the same as the former. In any case, rejecters of both the bundle and bare particular views cannot be satisfied with the flat-footed response except as a temporary stop-gap to be filled in by a theory at a later time. There have been more meaty, fleshed-out substratum accounts. While this is not the place to give a thorough examination of all of the accounts of substrata on offer, I will say at least a little. The most promising accounts I have seen can be found in the more recent work of Michael Loux, D.C. Long, and the somewhat Lockean-inspired account of C.B. Martin (who in turn influenced E.J. Lowe ).42 (We will see later how Chisholm also fits into this picture). They have all, in turn, influenced me.43 What I present below is a composite of these various views.44 What is common to these substrata views is (i) a rejection of the bundle theory, (ii) a rejection of the terms of the old dialectic between bundle theorists and bare particularists, (iii) endorsing some kind of (sortal) essentialism, and, (iv) accepting that, in some sense, objects are self-individuating, or, are their own substrata. (How the ‘in some sense’ is filled in makes for the differentia amongst the theories). I think these views are broadly on the right track, but incomplete. As I hope to show later, holding that the stuff of an object is its substrata is a fruitful addition to, and I hope completion of—this approach. I will also show an important precedent for this in the work of Michael Jubien.45 These ‘meaty’ substrata theorists all reject the bundle theory. That is, they reject the notion that an object is nothing more than the sum of its properties. But, none of them accept that the non-qualitative component which somehow completes the object and in which all the properties inhere is somehow ‘bare’ as the bare particularists argue. Rather, the particular is a characterized particular of some kind.46 D. C. Long calls 41. See Locke’s Essay, II xxiii 1–4, and, for a secondary source, Jolley 1999. 42. See Loux 1998 and 2002, Long 1970, Martin 1980, and Lowe 2000 and 2003. 43. Along with, as you will see, Michael Jubien. 44. In a way this will be somewhat unfair to each of the writers, but, I hope, convenient for the reader. It is not my intention to present an exegesis of the works of the aforementioned philosophers, but, rather, to present a general strategy they all, in broad terms, represent, while of course disagreeing about much of the particulars. 45. See Jubien 1993 and 2001. 46. See Loux 1970, 193–6.

23

his theory ‘Qualified Particular Theory’, or QPT.47 Michael Loux in a somewhat Aristotelian vein notes that particulars are chiefly characterized by being of a certain kind, and, the having or instantiating of this kind guarantees that a distinct individual of the kind exists, and is the thing which has the properties.48 He argues (along with E. J. Lowe in 2000) that concrete particulars are not the kind of entities to be reduced to more basic ones (such as properties and bare particulars), but rather we can help ourselves to concrete particulars of certain kinds from the outset. In many ways these substrata theorists are rejecting the old terms of the debate. The debate, at least in modern times, started with the ‘Humean mistake’ (see Chisholm 1969). The Humean mistake, committed by both bundle theorists and bare particularists, was (in part) to suppose that, when we observe a particular, we do nothing more than observe its qualities.49 As our idea of any body, a peach, for instance, is only that of a particular taste, color, figure, size, consistency, etc., so our idea of any mind is only that of particular perceptions without the notion of anything we call substance, either simple or compound.50

The Humean mistake can be summarized as follows. One assumes, as a good empiricist seems she should, that what one observes when they observe a thing is actually just a property or collection of properties. Then, once we accept this, we wonder why we need in addition to the properties postulate a weird, non-qualitative thing which has them. And so one often becomes a bundle theorists. But then, folk often believe that something like the ‘compresence’ relation of the bundle theory is too loose to tie these qualities together. And so, the bare particularist accepts the terms of the debate, but rejects the bundle theory, and in turn grudgingly accepts the goofy bare particular.51 The meaty substrata theorist thinks that the right thing to do is to reject one of the major premises that led to bundle theory or bare particular theory in the first place, namely, the idea that when we experience a thing what we directly experience is only its properties, the implicit commitment 47. 48. 49. 50. 51.

See Long 1970. Loux 1998 and 2002. See also Long 1970, 278. Hume’s Enquiry, 194 Hendel version. For some bare particularists, see Allaire 1963 and1965, and Bergmann 1953.

24

of this view being that objects are something which we infer rather than directly experience.52 Chisholm: One is tempted to say instead that our idea of a peach is an idea of something that has a particular taste, color, figure, size, and consistency; and analogously for the other familiar physical things. But even this is not quite right. Our idea of a peach is not an idea of something that has the particular qualities, say, of sweetness, roundness, and fuzziness. It is an idea of something that is sweet and round and fuzzy. More pedantically, our idea of a peach is an idea of an individual x such that x is sweet and x is round and x is fuzzy.53

Long chimes in that supposing we observe properties but not necessarily the things that have them is absurd: Again, suppose that one is asked to locate in instance of a colour, say, the red colour of the ball in our example. We are to locate not the ball, however, but simply its colour … There is no way to give a location of the color per se. If we try to pin it down we merely find ourselves attempting to locate the pigmented surface of the ball … To give still another illustration of my point, the claim that redness and being four inches in length are moving through space can only be understood as meaning that some object which is red and four inches long is moving through space.54

Finally, Martin also shows his rejection of the Humean mistake implicitly by showing how committing that mistake led to asking the wrong questions: To ask, ‘What are the properties of that which has all the properties it has?’ or to ask, “What are the properties of the bearer of properties other than the properties it bears?’, is to ask the wrong questions, and very wrong ones at that.55

To one degree or another, views like this veer towards some kind of essentialism. Michael Loux, for instance, believes that an appeal to kinds can cut the Gordian knot of the bundle/bare particular dilemma. One universal, the kind a thing belongs to, has a sui generis individuating role: Every substance exhibits a universal whose exemplification is by itself constitutive of an individual substance and whose multiple exemplification is by itself constitutive of a plurality of different individual substances.56 52. 53. 54. 55. 56.

Cf. Long 1970, 278. 1969, 8, ital. his. 1970, 272. 1980, 6. 1998, 243.

25

Key to this view is rejecting the ‘constructivist’ view of both bundle and bare particular theorists, who both take it that concrete individuals are constructed out of, or reduced to, more basic ontological entities: … [T]o allow substance kinds to play the proposed role in our ontological characterization of substances is to reject the reductivist/constructivist framework that structures the debate between bundle theorist and substratum theorist. It is to hold that the concept of a familiar concrete particular is given us at the beginning of the ontological enterprise in the substance kinds under which ordinary objects fall … Substances are not wholes made up of constituents [i.e., universals alone or universals plus bare particulars]; in virtue of instantiating their proper kinds, substances are irreducibly basic entities … A framework of first-order properties like colours and shapes is simply too impoverished to provide the materials for reconstructing the framework of substance kinds … The idea of a particular substance is not something we need to construct.57

Loux is, if I may be allowed to stretch things a bit, saying that the thing which has the properties is the very thing itself, as characterized by the essential properties. So, the thing with the essential properties has the accidental properties, and the thing with the essential properties is one and the same as the thing which has the accidental properties. The object has the essential properties in the very same way that it has the accidental properties, but the essential properties help us characterize the substrata ‘core’ that we are attributing the properties to.58 This, it seems to me, broadly correct. E.J. Lowe continues with a proposal which is somewhat similar to this, and characterizes it in part at as follows: It may, I concede, strike some philosophers as being entirely at odds with any doctrine of substratum to say that the substratum of a property-possessing object is that very object itself, but I disagree. Substrata are invoked, primarily, to play the role of something which is not itself a property, and upon which the properties of an object can depend for their existence (and perhaps also their identity), on the assumption that properties themselves are not independent beings. But the object which possess the given properties is at least formally equipped to play this role, because it is not itself a property, nor, I should say, is it a collection or bundle of properties.59 57. Loux 1998, 243-4. For similar remarks, see Martin 1980, 6–7. 58. Cf. Long 1970, 277–84. 59. 2000, 508.

26

Lastly, we even have a diagnosis for why the flawed debate arose and continued unabated for so long. Martin, in appealing to Locke, gives a gloss that runs like this: our powers of abstraction can seemingly impute ontological divisions where none actually exist, and our ability to think of one feature in abstraction from its others, when taken to extremes, can lead us to abstract every feature away from an object except for its property-bearing feature. And it is indeed correct that objects having a property-bearing character—that is almost definitional. The problems arise when we believe either that we can drop the notion that there are objects, distinct from properties, which bear properties, or, when we think that the thing which bears properties is something distinct from the object itself: The substratum view can be put least misleadingly and most clearly by employing Locke’s device of ‘partial consideration.’ When we are thinking in the most general possible way of the attribution of properties (each and every one) to an object, we are thinking of, or partially considering, the object, perhaps a passionfruit, simply qua or simply in its role as, the bearer, not itself borne, of its properties without at the same time considering it in terms of the actual properties it undoubtedly bears. Partially considering a passionfruit, as what bears whatever properties it bears, is thinking of it under a partial, incomplete description—as a bearer of properties. This is not to think of the passionfruit as a passionfruit kind of object, nor, of course, is it to deny its being of this kind. It is, rather, to consider the passionfruit as a bearer of properties (without attending to what those properties are) which itself is not borne as a property, or set of properties, by anything else. The passionfruit under this partial consideration, and incomplete description, is indeed the substance or substratum. Where, then, is the harm?60

I believe the foregoing proposals allow us to say most of what we want to say, and represents a reasonable, non-mere-table-pounding solution to the problems of property inherence. It avoids the problems of the bundle theory. It avoids the problems of the bare particular theory. It countenances a thing distinct from its properties, yet denies that this thing lacks any properties. This is what we should want. That’s the good news. The bad news is that nothing about these accounts actually solves the metaphysical problem of individuation, i.e., accounting for numerical diversity in cases of exact qualitative similarity. Rather, these accounts presuppose an answer to the problem. Appealing to kinds does not help. Loux says that “multiple exemplification is by itself constitutive of a plurality of 60. Martin, 1980, 9–10.

27

different individual substances.”61 Of course, multiple exemplification is always sufficient to establish a plurality of objects, whether you appeal to kinds or not, if one means by ‘multiple exemplification’ the instantiation of more than one thing. If, however, by ‘multiple exemplification’ means the apparent multi-instantiation of kinds, where we leave open the possibility that the ‘two’ things might be one—then certainly nothing about multiple exemplification by itself ensures the exemplification of two distinct things. How does Loux’s account differentiate between the following two worlds—world 1, wherein there is only one multi-located sphere, and world 2—where there are two distinct but qualitatively identical spheres? It does not differentiate between them. I think at this point it would be helpful to step back and look at the desiderata of a substratum view. Then, I will show how it is fruitful to hold that the material stuff of an object is its substratum, by showing how it fulfills the desiderata. After this, I will show how Chisholm seems to have been somewhat similarly motivated. Why are we postulating substrata in the first place? What’s their job? What are they for? I think the following is a helpful first pass at a good list of the desiderata to be fulfilled by a substratum theory. Note that the items on this list overlap significantly. Also, I do not claim that this list is exhaustive: Desiderata for a Substratum View of Property-Inherence: 1. A substratum theory should account for metaphysical individuation, such as in Max Black’s case, but should not do so by ad hoc appeals or mysterious means. 2. A substratum theory should show how various properties are tied together as the properties of one thing. That is, it should show how the nature of substrata, or inherence base, can tie together its various qualities. 3. It should also explain how properties are dependent on the objects that have them, but, also, explain how objects are not dependent on their (non-essential) properties. 4. Explain both the sense in which the substrata has properties, and the sense in which a substrata is a thing apart from its properties, or, less misleadingly, is not merely a sum of its properties. (Another way of stating this is that a substratum account should characterize the nonqualitative component of concrete particulars). 61. 1998, 243.

28

5. A substratum theory should shed light on the distinction between objects (i.e., property-bearers) and properties. Now, I would like to show how a mereological essentialist view of property-inherence along Chisholmian lines, which holds that the stuff or matter of an object is its substratum, satisfactorily fulfills (1)–(5). But, first, we must take an illustrative detour through an important precedent for this view in the work of Michael Jubien. This will in part be clarified by examining an exchange between him and Ted Sider. We will also see how this ME view of property-inherence, when combined with a Chisholmian combinatorial theory of possibility (I will explain what this is), makes this whole picture more strong than the views of property-inherence and possibility considered separately. The notion of ME as motivated as an alternative to the bundle and bare particular views has also been at work in the work of Michael Jubien, but seemingly goes unnoticed by one interlocutor, Theodore Sider. Jubien argues in the following way for ME in Ontology, Modality, and the Fallacy of Reference: … think about things in the abstract, in isolation from everyday descriptions and associations. So first recall that an arbitrary thing is just the occupier of some arbitrary, full region of space-time. Let x be any such arbitrary thing, and let y be an arbitrary proper part of x … there also exists a third thing—the thing that is all of x except for the part y. Let’s call it z. If we agree to use ‘+’ and ‘–’ in the natural way for mereological sum and difference, we have z = x – y. Now imagine another situation, as much like this situation as possible, but in which the entire thing y simply does not exist. This certainly seems like a situation in which x doesn’t exist either, but z does. I think it very difficult to deny this intuition without somehow relying on prior convictions involving everyday descriptions and associations, like the belief that a certain house could have had (somewhat) different parts. (1993, 18–19)

Sider replies: The crucial claim in this argument is that in a possible situation in which y does not exist but z does, x does not exist either. What is the support for this claim? Jubien asks us to forget about nearly all features of the objects in question, but he does draw our attention to the fact that x fills a certain region of spacetime. In particular, Jubien draws our attention to the composition of x: it is the sum of y and z. But why is this particular feature of x the only feature one may consider in thinking about x’s modal properties? Let us sup-

29

pose that x is, in fact, a house. Being a house is then just as much a feature of x as is being composed of y and z; in thinking about x’s modal properties, why should we abstract away from the former, but not the latter? … think about x in abstraction from its material composition; just think about the fact that x is a house. Couldn’t that object have existed even though one of its small parts failed to exist? The answer now seems to be yes … the problem for Jubien here is that he is committed to an unjustified asymmetry between predicates like ‘is a house’ on one hand, and ‘is composed of y and z’ on the other … the latter expresses a property that is essential to its bearers, whereas the former does not. Why this difference? Here is an object with many properties. It is a house. It is made up of certain parts. The former is just as good a candidate to give the essential nature of the object as the latter … Jubien’s choice of the latter candidate appears arbitrary. (Sider 1999, 288. Two longer emphases mine)

But Jubien’s choice of the latter is anything but arbitrary. Sider seems to ignore crucial passages of Jubien (1993, 41–45) where the latter argues that “the property being made of this stuff … is a perfect candidate for being the haecceity of the thing in question” (Jubien 1993, 43). Haecceities, or primitive ‘thisnesses,’ are often introduced to deal with problems like Max Black’s example wherein there are only two qualitatively indiscernible globes in a radially symmetrical space (Black 1952). It seems obvious that a world just like this world could have existed where the two globes have reversed positions. But, this does not describe a world that is qualitatively distinct, and so, either this is not a distinct world, or, there exist brute and primitive individuating (non-qualitative) components such as bare particulars, or, there are merely formal individuating properties, such as being this globe. Some have settled for the latter, the classical name for these brute individuators being ‘haecceities’ or thisnesses. By accepting haecceities the problem is ‘solved’ in differentiating our worlds. In these worlds, the spheres have merely switched their haecceities. But, this often comes across as a desperate, ad hoc, and mysterious move. Also, strictly speaking, it does not solve Max Black’s case, since, haecceities, being properties, make it such that, if one globe has one haecceity, the other globe a different one, then they are not in fact qualitatively identical. (If haecceities are not in fact qualitative then they’re really quite like bare particulars, with the exception that properties cannot inhere in them. It is hard to see what the advantage of this over classic bare particulars could be.) Jubien asks, what counts as a qualitative difference? In particular, does being made of such and such stuff count as a quality? (Jubien 1993, 42). Since the globes are made of different stuff, and if the foregoing type of 30

properties are qualities, then the positing of qualitatively indistinguishable globes does not make sense. But, being made of this stuff does not seem like a general quality, since, if Jubien’s right that things are nothing more than precise parcels of stuff, the property is not repeatable or multi-instantiable (Ibid., 42–3). This, however, would play into his hands: We already saw that the property being made of such and such stuff, considered as a property of official things, could not be general. If this is enough to show it isn’t a quality, and if haecceities must be nonqualitative, then it is a very plausible candidate for being the haecceity of the thing that has it. (Ibid., 43)

But, whether or not we want to count properties like being made of such and such stuff as qualities that a thing has, Jubien notes that What is important is that the property is a perfect candidate to serve as a thing’s haecceity whether we say it is nonqualitative or not … This candidate does everything we could reasonably ask of a haecceity: It is an essential property of the thing that has it. It is necessary that if anything has the property, then that thing is the thing that actually has the property. To the extent that it makes sense to speak of other possible worlds, and to the extent that there is any need to ‘ground’ talk of ‘identity across possible worlds,’ this candidate fills the bill. So I nominate it for the office. (Ibid., 45)

I second the nomination, but with a qualification. Jubien is quite right that the property of being made of some stuff is a perfect candidate to play the role of the haecceity or individuator for an object. We get all the benefits of brute individuation, with none of the cost of weird particularizing but non-qualitative properties: Appeal to haecceities has often been regarded as a desperate measure, one primarily designed to save an uncertain intuition. Desperate, it is said, because it is an ad hoc appeal to what are fundamentally mysterious entities … I think some conceptions of haecceities have invited this criticism, but the present one does not. There is nothing mysterious about the idea that things are made up of stuff. And there is nothing mysterious about the idea that the stuff over here is different from the stuff over there. So there is nothing mysterious about the property being made of this stuff. (Ibid.)

While this solution does merely switch the problem of individuating from the individual to the stuff, it does, it seems rightly to me, as well as Jubien, 31

put the primitiveness where it belongs (Ibid., 46). My only quibble here is that the notion of ‘haecceities’ such as being made of this precise stuff are not doing any additional work than just supposing that there are parcels of stuff that are distinct from one another, where the stuff plays the substratum role. But, if properties such as being made of this precise stuff are to do the work required, then the stuff must be postulated to be distinct and play the substratum role anyway. So, I prefer an account like Jubien’s except that substrata are used rather than haecceities. (I’ll develop this further below). While Jubien does not explicitly lay out the idea of the stuff of a thing playing the haecceity-role as a species of ‘meaty substrata’ technique to maneuver between the horns of the bundle and bare particular dilemma, we can see how it would help. The stuff of a thing plays the role of the bare particular, but there is nothing bare or odd about it. The stuff is a bare object, and bare objects are like clothed bare particulars, in that they provide both the substratum for the thing and give us its sort (namely, ‘mass’, ‘fusion’, ‘parcel of matter’, ‘primary object’, etc.). Looking at things in this way has several advantages. Not only do we get a metaphysical interpretation of predication or property inherence for commonsense objects that steers between the goofy (bare particulars) and the goofier (bundle theory), and solves the problem of Black’s globes, but it also helps provide answers to many troubling questions about persistence and possibility, as well as giving us the tools to break a putative impasse. We can note here, and in other defenses of (e.g., in Jubien 2001) and attacks upon ME, a certain pattern and impasse. The ME’ist says, think of certain hunk of stuff—don’t think of anything else about ‘it’. How can it survive a loss of parts? Those non-sympathetic to ME reply—well, tell me more about it. What is the hunk of stuff? Is it a house? Then of course it can. Those who are against ME think that the features that we think of as giving an essence depend on the description of the item: “When viewed as a house, that object might have lacked y; when viewed as a sum, having y as a part is essential to it” (Sider 1999, 289, emphases mine). Whereas those in favor of ME believe that there is nothing to giving an essence of an item other than listing its ultimate parts (or, referring to some gunk). We can break this impasse in favor of ME when it is seen as motivated by being a third option that avoids the difficulties of the bare particular and bundle theorists’ views. If the primary bearers of properties are hunks of stuff and primary objects are their stuff (commonsense objects being ‘constructions’ or ‘fictions’) then Sider’s demand that we ‘think about x

32

in abstraction from its material composition’ is absurd (since we’re being asked to think of a thing in abstraction from itself ), as is his claim that properties like ‘being a house’ are just as much a candidate for being essential to an object as being composed of y and z. The ME’ist can defensibly claim that ‘bare hunks of stuff’ are all that we pick out when we pick out objects. For instance, look again at the Sider quote: “When viewed as a house, that object might have lacked y; when viewed as a sum, having y as a part is essential to it” (Ibid., emphases mine). What is ‘it’, or the object such that it can be viewed as a house, or as a mere hunk of stuff, and can be thought of multiple ways so that different features are apparently essential to it under different descriptions? Sider’s view (in this article) of physical objects is pretty mysterious, whereas Jubien’s is utterly clear. For Sider, there is some odd thing such that it can have different persistence conditions when thought of differently. For Jubien, there’s just some stuff, and all of it is essential to it. While it is true that we can ‘turn the argument around’ on Jubien and be asked to think of the house in abstraction from its material composition, this is question-begging at worst, and moot at best. In the context in question of Jubien’s book, he is explaining that by regarding the property of ‘being made of such-and-such stuff’ as the haecceity of the object in question we can solve certain problems. It should be no surprise that if we ignore the reason that a thesis was introduced (namely, that by identifying genuine objects with their stuff we can solve the problem of Max Black’s spheres) and instead focus on other problematic elements of it, that it will be easy to apparently defeat the proposal. Anyways, if the mysterious ‘it’ is nothing but the hunk of stuff, and such hunks are the primary bearers of properties, then, not only are Sider’s claims about ‘it’ (e.g., that it could have lacked a part when thought of one way) false, but, Jubien’s choice of ‘being composed of x and y’ versus ‘being a house’ as giving the essential features of an object is not ‘arbitrary’, pace Sider. Now I’ll try to develop this Mereological Essentialist version of property-inherence, and show how it fits the desiderata (1)–(5) listed above. But, first, let me give a general idea of this theory. Note also that this is more of a proto-theory than a fully developed one. I just want to make clear the general structure of this position and its advantages. The stuff or matter of a commonsense object is distinct from it. This is because the matter will both pre-exist the object and last past its destruction. Even if an object and its matter are both simultaneously created and then later destroyed (where we suppose that none of the matter which

33

constitutes the object is lost during the interim) the matter could have existed past the destruction of the object.62 The stuff of commonsense objects is identical with fusions of either simple, non-composite particles or non-atomic ‘gunk’. This stuff of an object—the mass of matter—all of its parts are essential to it. Different parts—different thing. In fact, in a certain sense, these masses of matter are the only objects there really are— they are the ‘primary objects.’ Commonsense objects, such as tables and chairs, are somehow derivative entities. When we say that a commonsense object, such as a particular table, has a property, such as being brown and square—how are we to understand this? Is there a bare particular, which has the property of being brown and square, which ‘in itself ’ has no properties? Do we assert that there is a compresence of browness and squareness? No, and no again. A table is not a mere compresence of properties, and no bare particular is needed. A table is what results when a mass of matter exemplifies the phase sortal ‘being a table,’ or, as I prefer, when a mass of matter undergoes the process or activity of ‘tabling’.63 What makes it the case that two qualitatively identical but numerically distinct things are distinct? The stuff. Distinct stuff—distinct things. If Max Black’s spheres are described as being made up of the same stuff—then there is actually only one multi-located mass of matter. If they are stipulated to be made up of different stuff, then they are distinct things. No problem arises. What’s the substratum of a commonsense object? Its matter. That’s why an ens successivum is, strictly speaking, a succession of distinct tables (when its parts change). The only genuine objects are masses of matter. Tables are masses of matter in which qualities such as being a table inhere (or, the property of being a table is a Platonic entity that the mass participates in), or, are masses of matter in which the property of being a table is immanent, or the mass of stuff possesses the trope being a table, and so on. But, you may ask, what makes one parcel of stuff distinct from another? A parcel of stuff x and a parcel of stuff y are identical just in case they have all of the same ultimate or fundamental parts. But, suppose we get down 62. All of this, actually, has to be highly qualified. I am against coincidence, so, strictly speaking, I would deny that there are two distinct objects in the same place at the same time, made up of out of all the same material parts—such as a mass of matter and a distinct statue. On my view, and the later Chisholm’s view, as we’ll see, the matter and the statue are of two different kinds. The matter is some stuff, and it is statueing. The statue is an activity, only the mass is an object. The standard problems of coincidence do not arise on this view. Others do, to be sure, but they are more tractable. 63. Cf. Jubien 1993, and section 3 below.

34

to ultimate parts, such as electrons.64 What makes two electrons distinct? Well, perhaps they are composed of two different ‘super-strings’ vibrating in different parts of space. What makes two strings distinct? I do not know. At some point we stipulate or suppose that there are different parts, and use this supposition to do further work. Regardless of one’s theory of property inherence, we will ultimately get down to some point where the individuation is just stipulated and presupposed. I just choose to locate this bruteness in the matter itself, rather than in some bare particular, or, in what seems to me derivative entities, commonsense things such as cabbages or airplanes. Sounds like materia prima, you think? It is not. Different particles of matter have different essential properties—let the scientists tell us what they are. There are fundamental substances (or, Aristotelian ‘primary substances’), but we need only suppose they are things like electrons, gluons, and fusions of them, not horses and galaxies. The basics of this view is that the ultimate havers of properties are just simple particles or collections of them (or collections of gunk)—full stop. Analysis stops there. To ask for a bare particular is to impute a more complex ontic structure to objects than they really have, or need. To not require something like masses of matter for the properties of objects to inhere in is to embrace the bundle theorists odd, possibly free-floating compresent properties. Also, unlike meaty substrata theorists like Loux, where objects are their own substrata, we need not suppose that concrete particulars like frogs and stereos are fundamental and basic entities—this seems wrong. We get to suppose that frogs, stereos, and so on are composed out of more basic entities, i.e., masses of matter in which the property of being a frog or stereo inheres. What is it exactly about mereological essentialism that is doing the work here? Its being true is a necessary precondition for presupposing that the stuff of an object is its substratum. If we suppose that an object can change parts, then we cannot identify the substratum of an object as the stuff which composes it. If an object can change parts, then the parts cannot be construed as what has the properties—rather, there is a substrata which has the property of having some parts at one time, and having another set of parts as parts at another. If this is allowed, then surely it is not the case that a precise collection of stuff is the haver of properties, rather, a distinct thing has the properties of having certain stuff as parts. 64. I could have stated this in a different way about gunk as well, but prefer not to.

35

This position recommends itself only if we believe that properties are parts of objects in the same way that material parts are parts of objects—which is surely a mistake. I will say more on this further below. Now, how does this view of property inherence satisfy the desiderata laid out above? Let’s go through them. (1) A substratum theory should account for metaphysical individuation, such as in Max Black’s case, but should not do so by ay ad hoc appeals or mysterious means. This has already been adequately explained above. (2) A substratum theory should show how various properties are tied together as the properties of one thing. That is, it should show how the nature of substrata, or the inherence base, can tie together its various qualities. We can see how the ME property-inherence view ties together various properties as properties of one thing. Distinct properties are properties of the same thing provided they inhere in either all of the same stuff, or inhere in different parts of the same collection of stuff. (3) It should also explain how properties are dependent on the objects that have them, but, also, explain how objects are not dependent on their (non-essential) properties. The ME property-inherence account can satisfy this desideratum as well. Properties are dependent on the objects which have them (whether bare or partially nude), since no property can float free and detach itself from the matter that has it. Bare objects are not dependent on their non-essential properties, since being a cup, being the Empire State Building, and, being red are quite obviously properties that a mass of matter can lose while continuing to exist. As far as the essential properties of masses of matter—this would be a function of all of the essential properties of the fundamental particles that make up that particular mass. What are the essential properties of fundamental particles? Once again, ask a scientist. Commonsense objects are not dependent on their non-essential properties, since, although they are strictly speaking fictions or constructions, the assertability conditions for the identity or not of entia successiva are given

36

by folk usage, and folk usage, for instance in the case of an automobile, assures us that we identify a car as the same one when it changes tires, even though, strictly speaking, we have a different object before us after the tire-change. (4) Explain both the sense in which the substrata has properties, and the sense in which a substrata is a thing apart from its properties, or, less misleadingly, is not merely a sum of its properties. (Another way of stating this is that a substratum account should characterize the nonqualitative component of concrete particulars). The satisfaction of this desideratum would depend in large part on one’s theory of the nature of properties—an issue I don’t want to try to settle here. But, I see no reason why the desideratum cannot be satisfied regardless of one’s theory of properties. Substrata have properties, because the properties are instantiated by a collection of stuff. We cannot say, in general, anything very informative about what it is for a mass of matter to have a property. We can say a bit more about more particular cases. A mass of matter has the property of being spherical, for instance, if it occupies a spherical-shaped portion of space. A mass of matter is solid if it generates a certain amount of impenetrability to another mass of matter. And so on. But, this ME account of property inherence does not identify an object with all of its properties. Subtract away all the non-essential properties of the mass of matter. What do you have? The mass of matter. This thing is distinct from all of its contingent or accidental properties, even though it always has some accidental properties or other. Also, we can employ ‘partial consideration’ and consider the mass of matter only in its property-bearing aspect. But, we would be fooling ourselves if we thought we were thinking about a non-propertied haver of properties. More roughly put, we can imagine a certain mass of matter distinct from its particular arrangement, and then we are thinking of the mass qua mass, even though the mass always has some arrangement or other. Another way ME satisfies this desideratum of avoiding identifying an object with all of its properties is that the ME’ist realizes that objects do not have properties as parts in the same way that objects have material parts as parts. When you subtract away all the properties of an object, what do you have? The object and all of its material parts. An object’s ultimate parts are not parts of it in the same way that its properties belong to it. While the predicate ‘having such and such a part’ is certainly true of a mass of

37

matter, it does not refer to a property in the same way that the predicates ‘being red’ or ‘being a frog’ do. Why is this? It is because of the following disanalogies of partial consideration. We can ‘abstract away’ from an object’s color, shape, and so on. If ME is true, we cannot ‘abstract away’ from an object its material composition, because, if ME is right, that’s what objects ultimately are—their material parts. (This is not to say that we cannot think of a color in the abstract). We can imagine the shape as apart from the color, and the color as apart from the shape, but we cannot imagine the shape or color as apart from the object’s material composition. Things with no material composition have no shape and color, and can have no shape and color (regions of space excepted as regards shape). The whole idea of construing properties as parts of objects is just wrongheaded. I can cut the top half off a cube and destroy it, and have left an independently existing entity.65 I cannot take off only the shape, leaving the color behind. Treating properties “as if they were objects in their own right to be bundled as so many sticks in a pile” would be a big mistake.66 This is because “an object is not just a group of properties, because properties are not themselves objects to be grouped.”67 Thinking of objects as bundles of properties, where properties are construed like objects is what gave rise to the problems for both the bundle and bare particularist theorists in the first place. We don’t need bundles, and don’t need bare particulars. We need masses of matter. (5) A substratum theory should shed light on the distinction between objects (i.e., property-bearers) and properties. Objects are masses of matter, properties are the ways that these masses are, or are the truth-makers for what can truly said about them. More could be said, but only if I present a theory of properties, which I will not do. But, suffice it to say, that I see no reason the major theories of properties can’t be made consistent with my proposals. Now, you may remember that this is an exegetical paper about Chisholm. My point in this section is not just to show how ME can find further support in this way, but to show that Chisholm was at least implicitly motivated by the notion that the stuff of an object is its substratum. 65. See Martin 1980, 7. 66. Martin 1980, 6. 67. Martin 1980, 7.

38

We can find some evidence for this interpretation in Chisholm’s “On the Observability of the Self.” Chisholm insists that we avoid the Humean mistake, and supports the notion that we do not perceive a thing’s properties when we perceive it, we perceive the thing which has the properties. Our notion of concreta is much more solid than our notion of a thing’s properties. While we can doubt that there are properties, like the nominalist does, we can not (reasonably) doubt that there are concreta. Chisholm discusses with approval Leibniz’s criticism of Locke’s empiricist epistemology and theory of perception: When we consider any person or thing, he said, what comes before the mind is always a concretum and not a set of abstract things or qualities; we may consider something as knowing, or something warm…but we do not thereby consider knowledge or warmth … The abstract things, he noted, are far more difficult to grasp than are the corresponding concreta. I cannot help but think that the point is a simple-minded one. ‘Our idea of a peach is not an idea of sweetness, roundness, and fuzziness …; it is an idea of something that is sweet and also round and also fuzzy …’ One would not have even thought of mentioning it, had not philosophers denied it and constructed fantastic systems on the basis of its negation. (1969, 9)

Chisholm is attempting to turn a certain kind of property-realists argument upside down. It is not the case that all we observe are an objects properties—objects themselves being something we infer. Rather, we observe objects, their properties (when we are realists) being something we infer. When we see the above in the context of the rest of Chisholm’s writings, we see a consistent picture emerge where the basic items of our acquaintance are primary objects, and primary objects (i.e., hunks of matter) are the primary bearers of properties but are not identical with sums of properties. This is another way of saying that hunks of matter are substrata. Further indirect evidence for my interpretation can be found by examining Chisholm’s only direct argument for ME. It is an argument by elimination that I call ‘the argument from modal vertigo’(Chisholm 1976, 147–49). In its bare outlines it goes like this:68 (A) One alternative to ME is what we could call ‘Extreme Mereological Inessentialism.’[‘EMI’] On this view, any whole could have been made up of any two or more things whatsoever. 68. I have not mentioned some of the other alternatives to EMI and ME he discusses.

39

(B) EMI is obviously false. [Chisholm’s comments on this is that EMI implies haecceitism, which is false.] (C) All the other (viable) alternatives between EMI and ME, or modest mereological inessentialisms, would allow that, if a table x in W1 is made up of simple parts 1–10.000, and a table y in W1 is made up of simple parts 10.001–20.000, then there is a W2 where x is made up of parts 2–10.001 and y is made up of 1 & 10.002–20.000. (D) But, if this is allowed, then it would be allowed that there is a world W3 where x is made up of parts 3–10.002, and y is made up of 1–2 & 10.003–20.000. (E) But, if this is allowed … then we have a world W10,000, where x is made up of parts 10.001–20.000, and y is made up of parts 1–10.000. (F) But then modest mereological inessentialisms imply EMI.69 (G) So, ME is true. One of the lessons here seems to be that we track composite macroscopic individuals across worlds, not, pace Lewis (see Lewis 1986), by tracking qualitative similarity (which gets us only loose and popular transworld identity, which is probably what Chisholm would regard counterpart relations of commonsense macroscopic objects as), but by tracking the possible arrangements of the non-composite simples or base-level primary objects. Chisholm says as much in his earlier “Parts as Essential to Their Wholes”: The theory of possibility does not require us to say, of any of these commonsense objects—the automobile, the table, the station … and the fish—that they exist in any other possible worlds. But it does require us to say, of the strict and philosophical wholes that constitute these common sense objects, that they exist in other possible worlds. The theory of possibility does not require us to say of any nonprimary object that it exists in any possible world other than this one. But it does require us to say that primary objects exist in possible worlds other than this one. What we can truly say about the unrealized possibilities of nonprimary things may be reformulated more precisely in terms of the unrealized possibilities of primary things. We do not need to suppose, therefore, that there are possible worlds which are indiscernible except for the fact that some nonprimary things are 69. Note that the move to (F) is invalid. [The argument does not, for instance, show that my foot could have been composed of the Empire State Building and Julius Caesar, which EMI, unqualified, seems to suggest]. I present Chisholm’s argument to point out certain features of it, not, however, to condone it.

40

constituted by one set of primary things in one of them and by another set in another. (1989, 79).

What parts a thing has is not merely one more property we can use to track a thing—it is the very basis of (material) thinghood itself, the minimal necessary conditions for a thing to have any properties whatsoever, be called a thing, or be reidentified. A (primary) thing just is its stuff. It is not some thing other than its parts, which can have the property of having some-parts-or-other, nor is a thing a cluster of properties, one of which happens to be ‘having some stuff x as a part’. This is not to say that things cannot be predicated of as having certain material parts, since of course they do. But, the ME’ist contends, having some material parts is not just one feature among many that a thing contingently has. The property of ‘having such-and-such precise parts’ has a halo of some kind. Chisholm himself renders modes and attributes as explicitly distinct from bodies or substances in the later Self-Profile. Modes, or the ways things are, are dependent upon their substrates. Substrates, or things themselves which have modes, are not modes of anything (e.g., regions). And a substrate for Chisholm is a substance, but this substance’s defining feature is having the material parts it does. His definition there of a thing or substance is, “x is a substance iff: for all y, if y is a part of x, then x is necessarily such that y is a part of it” (1986, 67). Things are not modes of anything else. And, things are identical with all of their parts. So, the property of having-as-parts the parts that they do are not modes or attributes of anything else, such as a bare particular. This view also naturally combines with explaining our modal intuitions about genuine substances as a kind of mereological combinatorialism from certain bits of stuff, and not as built up out of a Humean mosaic of property distributions. If the ME’ist views things this way, then it seems she has a leg up. The ME’ist can claim that most of our modal intuitions about putative commonsense substances are actually about properties and property entailment, whereas our modal intuitions about primary objects are about bare hunks of stuff in their own right. For instance, Jubien, in “Thinking About Things,” (2001) claims that certain intuitions, such as that a certain dog, Fido, is necessarily canine, is not at all about a substance—Fido. Rather, our intuition is that if some-stuff-or-other has the property of being Fido, then that same stuff would have the property of being canine. Our intuitions about the hunk of stuff, however, that composes Fido now, such that it can not lose any parts, is an intuition about a genuine substance. I won’t decide on the issue of whether this ME41

informed proto-theory of modality is correct. I merely want to point out what seems to be motivating the mereological essentialist. Genuine objects are nothing more than precise collections of stuff. Having x as a part is not merely one of many properties an item has (although of course it can be expressed by a predicate), rather, this predicate picks out part of x such that it could not have any properties (without that part) whatsoever. As we will see, however, although Chisholm seems to favor the view that the substrata of objects are their stuff, the position that there are bare objects gels better with this position than the earlier Chisholm’s view that masses of matter are partially nude. 2.4 Problems for partially nude objects Why, though, does Chisholm believe in partially nude objects, but not bare ones? What are the advantages of his position? One advantage is that he gets to severely delimit the number of objects. There is no object composed of me and the Eiffel Tower, although there is the primary object (right now) that constitutes me and the distinct one that constitutes it. I would also agree that we would want to rule out objects such as the fusion of me and the last dinosaur. But, why, if I have a hunk of clay in my hands, will the primary object that constitutes the hunk go out of existence merely by my tearing off a piece? Why isn’t the same primary object still around, just scattered? If I continually tear off pieces, combine them with bits and pieces from other bits of clay, I can casually be creating and destroying objects all afternoon while I play with my nephew.70 Also, since macroscopic objects, in the actual world, are actually (within their interior regions) mostly empty space, why does apparent macroscopic contiguity and ‘stuck-togetherness’ matter so much for a hunk of stuff to be an object? Another way of asking the question is, why is Chisholm’s mereology so restricted? Chisholm never quite explains why joining is so important to a hunk of matter’s existence, and staying joined so essential to its persistence. Rather than believing that separating an object’s parts and then sticking them back together again first destroys an object, and then brings it back into existence (or brings a distinct object into existence), why not just believe that the object existed throughout, and had its parts scattered, and its parts were then made more contiguous again? Certainly nothing about accepting 70. I believe van Inwagen 1990 has an example like this, but I cannot remember the exact location.

42

ME per se requires us to accept joining as a condition for composition. Also, we can ask the following Euthyphro-like question—‘Are partially nude objects countenanced by the folk as objects because they are, or, are partially nude objects objects because the folk countenance them?’ The former seems more true than the latter. If so, what is to prevent bare objects from being objects as well? I will argue though, that Chisholm should accept bare objects over partially nude objects in part because of Theodore Sider’s argument from vagueness, but also because only by accepting this can Chisholm get around certain objections given by André Gallois. We can object to partially nude objects by looking at a paraphrased version of a thought experiment of Sider’s (2001, 122). Imagine that a precise bunch of stuff, either ‘simples’ or ‘gunk’ composes a fusion; in this case an iron cube. (In the rest of this paper, for simplicity, I will write as if there are simples only, and not gunk. I believe nothing here hinges on this). Now, imagine something like the following operation taking place. At t1 we turn on a strong magnetic field which permeates the sealed room the cube is in, and gradually moves all the constituent parts of the cube away from its center of gravity at the rate of one ¼ Planck length [10–33/4 cm] per one millionth of a nano-second. Suppose that at some time t3, when the simples are, say, spread around the room evenly in an invisible cloud, they no longer compose an object. The problem with this is that if composition is definite, then there must be some exact time t2 where the simples compose a fusion, where the ‘next’ instant they do not, and this change is something like the ¼ Planck length distancing of its bits apart from the initial center of gravity, over the course of one millionth of a nano-second from t2.71 This is absurd. Any of (these) kind of changes which could supposedly destroy the fusing of the simples and hence the fusion, and hence the object, are arbitrary. And arbitrariness won’t do. Furthermore, since composition is non-arbitrary and definite, if you hold that the simples do not compose a fusion when they are spread around the room in a cloud, then see what occurs when you reverse the magnetic field and watch them coalesce. By the same kind of reasoning, if we reject arbitrariness, then the simples cannot ever come to compose a fusion—that is, the cube can never come into existence as an object! There will rather be simples ‘arranged cube-wise,’72 but not a cube. So, Chisholm should not accept that joining is a precondition of being a fusion. But, if he wants a 71. Please allow me the convenience of distinct, yet contiguous ‘instants’. 72. This the view Peter Van Inwagen espouses, at least for non-organic material objects, in 1990.

43

system of mereological essentialism similar to his own, he ought to accept that bare objects (fusions which persist just so long as all their parts do), not partially nude ones, are the primary objects. Furthermore, Gallois in Occasions of Identity gives the following objection to Chisholm’s program (1998, Part III, Chapter 8). Most properties that objects can gain and lose are what he calls ‘mereologically destabilizing properties’ (‘MDP’). That is, most salient properties or changes that an object goes through, in the actual world, result in a change of parts. For instance, painting Theseus’ ship, or his ship being such that Theseus walked across it. Having the latter obtain would obviously scuff off a few molecules. Gallois defines MDP as follows: Let us say that property φ is mereologically destabilizing just in case it is physically necessary that if x has φ then x changes at least one of its component parts. If Chisholm’s account was correct, then there could be no genuine or primary object which has been made to, say, glow; heating an iron cube to the point of glowing brings about the release of photons. Theseus could not walk across his (mereologically stable) ship. Supposing for simplicity that Thesesus does not lose any parts, Chisholm can however reply that Theseus can indeed walk across an ens successivum that changed parts—a ship series. This, however, is not quite a successful reply. Chisholm must admit that not only can objects not change their parts, but, strictly speaking, objects can hardly change at all. This problem just disappears if Chisholm accepts bare objects and loosens up his restricted mereology. If he does, all the changes we would like to predicate of objects can occur to mereologically stable objects. All such mereologically destabilizing properties Gallois discusses can be cashed out in terms of internal changes in larger fusions which have the before- and after-change fusions as proper parts. A cube can come to glow, since that is a change in re-arrangement in all the particles or mass-energy that compose the cube before it is heated and the photons which are emitted afterward. Theseus can walk across his ship, since what he walks across contains the fusion of all the particles that constitute the ship when he starts walking, and the same fusion after he finishes walking. However, if Chisholm did pursue this line, he would be giving up considerably on the spirit of his program. There would be no need to posit entia successiva in the way he proposed. Talk about the whole life

44

of Theseus’ ship can be cashed out in terms of a very large fusion which always exists, which has different proper parts that have something like the property of being Theseus’ ship at different times. If, as Chisholm is wont to do, we would like to ignore this large fusion and instead focus on just those bits that have the ship property, it seems we would have to identify Theseus’ ship with a succession of time-slices of the succession of proper parts of the large fusion which successively constitute the ship. But Chisholm is rabidly anti-four-dimensionalistic. A way out of this would be instead to accept bare objects, but deny four-dimensionalism, and rather identify the ‘ship’ with a property, mode, or process of a succession of distinct bare-objects. And this is exactly what he does later. Although I cannot find explicit mention of this anywhere in Chisholm’s material, it seems that he changed his position in part to avoid having a four-dimensionalistic kind of position. 2.5 Entia successiva: just like four dimensional objects? We can see that Chisholm’s proposal, which countenances partially nude objects, actually is too close to four-dimensionalism for (his) comfort. Indeed, the very objections Chisholm gives to four-dimensionalism apply to his own account. In many ways, Chisholm’s earlier account of ordinary objects is no different than a stage-thereoretic four-dimensionalism.73 Chisholm’s main objection to four-dimensionalism, as contained in Appendix A to Person and Object, is as follows. The main arguments for four-dimensionalism are that it can help us solve a variety of puzzles about identity through time. For instance, Heraclitus famously asks, how can we bathe in the same river twice, since rivers are water, and the waters keep flowing on and on? Quine answers that things have temporal as well as spatial parts, and “that the temporal parts of individual things are like the temporal parts of the careers, histories, or biographies of those things” (Chisholm’s gloss on Quine. Chisholm 1976, 143). Quine writes: a physical thing … is at any moment a sum of simultaneous momentary states of spatially scattered atoms … Now just as the thing at a moment is a sum of these spatially small parts, so we may think of the thing over a period as a sum of the temporally small parts which are its successive states. (Quine 1959, 210. Cited in Chisholm 1976, 143)

73. I just became aware that David Wiggins made the same point in his 1979, esp. 302.

45

So, the ‘solution’ is that “you can bathe in the same river twice, but not in the same river stage” (Quine, 1963, 65. See Chisholm 1976, 143). But, as Chisholm notes, if this solution is to genuinely solve Heraclitus’ puzzle, it must “presuppose the concept of the persistence of an individual thing through time—the concept of one and the same individual existing at different times. Even if all rivers are sums of river stages, not all sums of river stages are rivers” (Chisholm 1976, 143). How do we know that the different stages that Heraclitus bathed in are stages of the same river? Quine’s putative solution to this is in part to define a relationship called ‘river-kinship’ (Quine 1963, 66), what Chisholm calls ‘cofluvial’. We can say that “a, b and c are stages of the same river iff they are cofluvial with each other” (Chisholm 1976, 144). Chisholm’s complaint then is that there is bootstrapping circularity going on here (the following parrots an imagined exchange in Chisholm 1976, 144): Q1: A1: Q2: A2:

How do I step into the same river twice? By stepping at different times into things that are cofluvial. What is it for things to be cofluvial? Things are cofluvial provided they are temporal parts of the same river.

Indeed, we can make the same charge against Sider’s stage-theoretic fourdimensionalism, which holds that objects that we think persist are actually just successions of instantaneous stages related by a temporal counterpart relation: Q3: A3: Q4: A4:

How do I meet Ted twice? By meeting two stages that are Ted-temporal-counterpart-related. What is it for things to be Ted-temporal-counterpart-related? Things are Ted-temporal-counterpart-related provided they are person-stages and each stage is Ted. (Sider 2001, 193–208)

The four-dimensionalist can respond that they can give a more informative answer than A2 (or A4), by appealing to causation. Something like: Things x and y are cofluvial (short for, ‘river-temporal-counterpart related’) just in case x is a river, y is a river, and x’s being the way it is causes y to be the way it is, or x causes some z1,… which causes some zn, where zn causes y to be the way it is, and each of z1…zn is a river. Indeed this is

46

just what Sider does.74 Note, however, that this is very similar to the way that Chisholm himself defines a ‘direct table-successor’ and the non-direct ‘table-successor’: D.III.1 x is at t a direct table successor of y at t′ =Df (i) t does not begin before t′; (ii) x is a table at t and y is a table at t′; and (iii) there is a z, such that z is a part of x at t and a part of y at t′, and at every moment between t′ and t, inclusive, z is itself a table. D.III.2 x is at t a table successor of y at t′ =Df (i) t does not begin before t″ (ii) x is a table at t and y is a table at t′; and (iii) x has at t every property P such that (a) y has P at t′ and (b) all direct table successors of anything having P have P. (Chisholm 1976, 99) There is even a Chisholmian analogue for a spacetime ‘worm’, namely, an object-series: (D12) C is an object series =Df C is a class having as its members an object-pair x, all the object successors of x, everything of which x is an object successor, and nothing which is unrelated to x by object succession (Chisholm 1989, p77). And, of course Chisholm would have to say something like this, since, as we can see, the template he used above can be used against him75: Q5: How can I touch the same table twice? A5: By touching at different times distinct items that are related by tablesuccession. 74. Well, not exactly. But, Sider connotes things along this lines, and it seems likely that something along these lines would be his reply to the triviality objection of Chisholm. Cf. Sider 2001 103 “… I accept that causation is a prerequisite of personal identity,” and ibid. 94 “The temporal counterpart relation is the same relation used by the worm theorist to unite the stages of spacetime worms … it may be analyzed in some way (in the case of persons perhaps in terms of memory or bodily continuity) …”. Also, see ibid., 227–236 75. Wiggins in 1979, 311 also shows how Chisholm’s entia successiva account suffers a similar objection to one frequently leveled against four-dimensionalists, namely that if ordinary objects are entia successiva, then they are event-like and could not have had a different history than the one they in fact have. Whether the objection is any good I leave open, but I think it’s important to note that Chisholm’s suffering of the same objections as four-dimensionalistic kind of positions strengthens the thesis that his position is very similar to them.

47

Q6: What is it for things to be related by table-succession? A6: Things are related by table-succession provided they are both tables and are parts of the same table-series. His definition of table-succession and table-series gets him out of the charge of complete circularity, just like my Sider-imputed definition which employs causation. So, the four-dimensionalist is really no worse off than Chisholm here, since he can avail himself of some criteria for cofluviality or ‘river-kinship’ which is just as informative as Chisholm’s definition of ‘river-succession’. But, if the four-dimensionalist is no worse off than Chisholm is, it is only because Chisholm’s entia successiva are too much like time-slices. David Wiggins seemed to be getting at this when he wrote: Only events or processes can have temporal parts. For someone in my position indeed Chisholm’s “things that do duty on different days for the successive table”… have too great a likeness to temporal parts of the table, and it is a real question whether Chisholm’s ens successivum treatment of tables will quite escape from all the good objections which are urged in Chisholm’s Appendix A section 4 against such notions as that a, b, c are stages of the same river if and only if they are cofluvial with one another.76

One thing that is interesting to note here is that Chisholm’s notion of ‘loose identity’ and Sider’s notion of ‘temporal-counterpart-related’ play the exact same role, namely, what I call a fudge factor for the folk. In both Chisholm and Sider’s case, there is no genuine persistence (for ordinary objects). There is, rather, a succession of instantaneous objects, each distinct from the next, and they are sometimes tied together by relations like ‘river-kinship’ or ‘river-succession’. Sider says that the stages are instantaneous objects which are part of a four-dimensional worm F when the stages are temporal-counterpart-related and each are F. Chisholm says that an ens successivum which is F is a succession of short-lived objects (given the fact of rapid microscopic change) each of which are F and each of which are F-direct-successors of one of the others in the series. These both give us a “fudge factor for the folk” in that each system allows us to have an “as-if ” talk of persistence of commonsense items, while not really holding there to be any that persist across time. Granted, Chisholm’s system is far different, in that objects (can) endure, and if objects never changed parts, they would not be like stages at all. But, given how the world in fact is, and given that Chisholm accepts partially nude 76. Wiggins 1979, 302.

48

objects which exist only when joined, (composite macroscopic) objects would exist only for an instant, and Chisholm’s system would amount to the same thing as Sider’s. The problem with both systems is that neither really gives us a theory of persistence (of ordinary objects, not primary objects) at all—only a way to save appearances after admitting that there really is no persistence over any significant time-frame. Or, more charitably, if the world is as they say it is, given that there is no genuine persistence, their accounts make the best of a bad situation, and tell us what the best candidate for ‘persistence’ is; either temporal counterpart relations or object-succession. But both these accounts have too much of a ‘fictionalist’ flavor—indeed, Chisholm admits as much himself,77 as does Sider,78 and if we can account for commonsense objects as persisting in a way that does not treat them as strictly speaking not persisting, then we ought to do so. Chisholm’s account ought to be accepted only if we cannot find some other entities to play the persisting object role—entities that genuinely exist and persist over time. 3. The later account: objects as modes, and modes as activities And Chisholm did find some other entities to play the commonsense object role: pairs of bare objects and modes. Chisholm changed his mind. I am not sure if he accepted bare objects for the reasons or concerns I raised above, but later, in his Self-Profile volume (1986), and “Scattered Objects” (1987), he allows both bare objects and modes, and the latter are somewhat like processes.79 77. Chisholm 1976, 96 “The point could also be put by saying that such things as the Ship of Theseus and indeed most familiar physical things are really ‘fictions’ …” Chisholm excepts persons from this treatment, however (see 1976 chapter III, section 5). 78. Sider 2001, 96: “I must concede, however, that tenseless statements of ‘cross-time’ identity are false …,” and, ibid.: “But, assuming four-dimensionalism is true, counterpart-theoretic persistence is as good as it gets, and is thereby the best candidate, and is thereby true persistence.” The last phrase does not follow. The ‘best candidate theory’ of content does not really entail that the best candidate for the use of a term F is always the true candidate for the term F. If this was so, then ‘the ether’ would mean ‘the vacuum,’ which was obviously the best candidate. The best candidate theory states that the candidate which best meets certain minimal conditions is the true candidate for reference, and I can’t see how stage-theoretic persistence meets the minimum standards. But, this is another paper. For a description of the ‘best candidate’ theory of reference/content, see Lewis 1983 and 1984. 79. Chisholm does, however, officially naysay process ontology (see Chisholm 1989, 94–95). But, what he was arguing against was both four-dimensionalisms and radical process ontologies

49

Chisholm quite clearly accepts bare objects at this period. No longer is being joined a necessary condition for the persistence of a fusion: If Harry is that object that has parts A, B, and C and that occupies the place that Charlie occupies on Monday, doesn’t Harry exist with precisely the same parts on the next three days? He becomes somewhat scattered on Tuesday, more widely scattered on Wednesday, and still more widely scattered on Thursday when he becomes a mass of jetsam. (Chisholm 1989, 93)

However, he never gives up ME, but merely expands its scope to cover scattered objects. In “Scattered Objects” he defines a substance in this way: (D4) x is an individual substance =Df If x has parts, then for every y, if y is part of x, x is necessarily such that y is part of it (Ibid.). He also calls the objects that ME is true of aggregates or heaps (Chisholm 1986, 68) and, in allowing them to persist just so long as all of their parts do, has squarely entered the bare objects tradition, and given up on the partially nude. Part of the explanation for this could be that Chisholm realized that his entia successiva account did not solve the paradox of coincidence. There is some circumstantial evidence to think this. In the Self-Profile volume, to deal specifically with coincidence, he develops what I call his “mode account”. Chisholm notes that one way to solve the paradox of the statue and the clay is to hold that, while ‘the statue’ and ‘the clay’ both refer, that ‘the statue’ picks out, not a substance, but rather a mode of a substance. And the substance of which the statue is a mode is, of course, the piece of clay. In this instance, the piece is a substrate of the statue. But what is a mode, as Chisholm understands it? We know that modes are not (i) essential properties of the substrates that have them, or (ii) universal properties of things (i.e., properties that everything has) (Ibid., 66). Roughly, modes are reified ‘ways’ that objects can be, and can change their substrate. Chisholm lays out the following desiderata for a definition of mode: Our definition … should allow us to say that the statue is a mode of the piece of metal—and that the piece of metal is not a mode of the statue … [and] should also allow us to say that a house is a mode of a heap or aggregate of such as Whitehead’s. As we’ll see, Chisholm’s ‘modes’ are quite akin to processes or homogenous activities in many respects.

50

building materials—and that the heap or aggregate of building materials is not a mode of the house (Ibid.).

Chisholm wants to divide the (physical) world clearly into modes and substrates/substances. Substances or substrates are bare objects, and modes are commonsense objects which are modes of the substances. His definition of mode is: (BD1) x is a mode of y =df (1) neither x nor y is an abstract object; (2) there is a z which is such that y, but not x, is necessarily such that it has z as a part; and (3) there is a P which is such that (a) x exemplifies P and (b) x is the only thing other than y which is necessarily such that it has P iff y has P. (Ibid.) It is best to turn to examples to clarify. Shipping is a mode of an aggregate just so long as neither shipping nor the aggregate are abstract, the aggregate has a part such that the aggregate, but not the shipping mode, necessarily has that part as a part, and, there is a property SHIPPING such that shipping exemplifies it, and shipping is the only thing other than the aggregate which is necessarily such that it has the property of SHIPPING iff the aggregate has the property of SHIPPING. It is very difficult to understand this clearly. It is especially hard to know if, in describing this view, that one may be imputing to Chisholm views or motives he did not have, since Chisholm is very brief here. It seems that shipping is not a universal or a property, but rather a reified activity. Shipping is not a property, SHIPPING is, and both the aggregate and the shipping have that property. The shipping activity, however, has the property indirectly and derivatively, while the aggregate has the property directly. The activity somehow ‘borrows’ the property from the aggregate. But, the aggregate is not necessarily SHIPPING, whereas the shipping is. Modes can move from substrate to substrate: If a substrate is a ship, then, it has a mode which is also a ship. If the substrate ceases to be a ship, and if the mode does not transfer to another substrate, then that mode ceases to be. And if, in such a case, the substrate continues to be, then it would have other modes (Ibid., 66–67).

Chisholm says more about what it means for shipping, for instance, to cease to be. Chisholm offers an answer as follows:

51

(BD2) P is a modal essence of x =df There is a y such that: (i) x is a mode of y; (ii) x has P and y has P; (iii) y is possibly such that it does not have P; and (iv) x is necessarily such that, if y ceases to have P, and if x does not become a mode of anything else, then x ceases to be (Ibid., 67). Using our previous example, SHIPPING would be a modal essence of shipping if there is an aggregate y such that shipping is a mode of y, shipping has the property of SHIPPING and y has the property of SHIPPING, but, y might not be SHIPPING, but, shipping is necessarily such that, if y ceases to be SHIPPING, and if shipping does not become a mode of another aggregate z, then shipping would cease to be. It still is quite mysterious what these modes are. Perhaps Chisholm can clarify things by saying some more about substrates or substances that modes are modes of. He gives the following definition: (BD3) x is a substance =df x is a contingent thing; and there is no y such that x is a mode of y.80 He also asserts that ME is true of substances (Ibid., 67, principle (BA1)). So, in contrast to modes, substances are not modes of anything, but modes are modes of substances. Furthermore, if substances have parts, they have them necessarily. Modes, by contrast, can change their parts: If the ship W is a mode and not a substance, we need not hesitate to say that it changes it parts from one day to the next. But the various aggregates that the situation involves never change their parts. For they are substances (Ibid., 68).

We have the outlines of a research program. I call, perhaps controversially, Chisholm’s modes activities partly because of a process of elimination. Modes are not essential properties of anything.81 Modes are not substances, or things. But, particulars like a ship or a table are modes. Modes have properties and modal essences,82 and are reified particulars. Aggregates have ships and tables as modes, but aggregates which have a table or a ship as a mode do not count as a ship or a table,83 only the modes do. I 80. Ibid. 67. Chisholm notes that this definition of substance won’t square well with those who think that God is a substance and a necessary being. 81. Ibid., 66. 82. Ibid., 66–67. 83. Ibid., 70.

52

cannot interpret what these modes could be except concrete activities, e.g. ‘a shipping’, such that different substances undergo it. If this is the proposal, then it has interesting implications. It certainly seems to dissolve some of the puzzle cases. In the case of Ship of Theseus, the set of aluminum planks which constitute the replacement ship (S2) is the substrate of the same mode, or ship, as the aggregate that originally made up Theseus’ ship (S1). The shipwright’s ship (S3), has a mode qualitatively similar to but distinct from the original ship.84 There is no puzzle of coincidence with the statue and the clay. There is just an aggregate of stuff which is statueing. The substance—the aggregate of stuff, persists throughout being shaped into a statue and being flattened, and happens for a while to be statueing—in which case we often say that there is a statue there. We are right that there is a mode or an activity of statueing there, but we are wrong that there is a substance or thing in addition to the hunk of matter to coincide with it. I will not criticize Chisholm’s modes view here, partly because I think it is roughly correct.85 Elsewhere I have offered an account very similar to Chisholm’s modes view,86 although actually more inspired by Karmo (1977) and Zimmerman (1995). Chisholm’s later account seems to me to be a reasonable alternative to solve the metaphysical puzzles of ordinary objects—one that deserves more attention. *** My main point in the foregoing was to elucidate Chisholm’s changing view of ordinary objects, and to point out certain strengths that the latter account has over the former. Also, I hoped to have shown how Chisholm’s consistent mereological essentialism has some additional motivating factors 84. Ibid., 67. (S2) is not (S3) by transitivity, since Chisholm contends that (S2) is the substrate of the same mode as (S1) after plank replacement, whereas (S3) is not. “Consider a ship that transfers from a substrate y to a substrate z. If z became a ship as a direct result of altering y, and if nothing else also then became a ship as a direct result of altering y, then we may say that there was a mode x which transferred from y to z. What if more than one ship was thus a direct result of altering y? Then we may say that the substrate having the most parts in common with y is the one that received the mode of y.” Ibid. 85. I should note that there are elements of the account that I did not cover, but with which I disagree. But these are quibbles. Overall, though, I agree with the spirit of Chisholm’s enterprise. 86. “Bare Objects, Ordinary Objects, and Mereological Essentialism.” Manuscript.

53

that need stressing. While somewhat speculative, I also, I hope, have shed some light on what might have been some reasons that Chisholm changed his mind about the status of ordinary objects.87 BIBLIOGRAPHY Allaire, Edwin B. 1963 “Bare Particulars,” Philosophical Studies, Vol. XVI. Reprinted in Loux 1970, 235–244. — 1965 “Another Look at Bare Particulars,” Philosophical Studies, Vol XVI. Reprinted in Loux 1970, 250–257. Bergmann, G. 1953 “The Identity of Indiscernibles and the Formalist Definition of ‘Identity’,” Mind, LXII, 75–79. — 1958 “Individuals,” Philosophical Studies IX, 78–85. Black, Max 1952 “The Identity of Indiscernibles,” Mind LXI, 153–64. Burge, Tyler 1977 “A Theory of Aggregates,” Nous 11; 97–117. Burke, Michael 1994, “Preserving the Principle of One Object to a Place: A Novel Account of the Relations Among Objects, Sorts, Sortals, and Persistence Conditions,” Philosophy and Phenomenological Research, 54, 591–624. Cartwright, Helen Morris 1974 “Heraclitus and the Bath Water,” Philosophical Review O 65; 74: 466–485. — 1979 “Quantities,” Philosophical Review JA 70; 79: 25-42. Chappell, V. C. 1964 “Particulars Re-Clothed,” Philosophical Studies, Vol. XV. Reprinted in Loux 1970, 245–249. Chisholm, Roderick 1969 “On the Observability of the Self,” Philosophy and Phenomenological Research 30: 7–21. — 1973 “Parts as Essential to Their Wholes,” Review of Metaphysics, 26: 581–603. — 1975 “Mereological Essentialism: Some Further Considerations,” Review of Metaphysics, 28: 477–484. — 1976 Person and Object. La Salle, Illinois: Open Court Publishing. — 1986 Roderick M. Chisholm, “Self-Profile,” Bogdan, Radu. J. (ed.), Dordrecht: D. Reidel. — 1989 On Metaphysics. Minneapolis, MN: University of Minnesota Press. Eklund, Matti 2007 “Fictionalism”, Stanford Encyclopedia of Philosophy, an online encyclopedia. The entry can be found at http://plato.stanford.edu/entries/ fictionalism/. 87. I am very grateful to Scott Berman, André Gallois, Mark Heller, Kris McDaniel, Thomas McKay, Irem Kurtsal Steen, Dean Zimmerman, and especially a referee for GPS, for reading earlier drafts and giving helpful comments. I am also thankful to Peter van Inwagen for some helpful correspondence.

54

Gallois, Andre 1998 Occasions of Identity. Oxford: Clarendon Press. Hawthorne, John O’Leary 1995 “The Bundle Theory of Substance and the Identity of Indiscernibles,” Analysis, 191–6. Henry, D. P. 1972 Medieval Logic and Metaphysics. London: Hutchinson University Library. Hume, David 2000 A Treatise of Human Nature. Norton, David F., and Norton, Mary J. (eds.) Oxford University Press. Jolley, Nicholas 1999 Locke: his Philosophical Thought, OUP, 70–8. Jubien, Michael 1993 Ontology, Modality, and the Fallacy of Reference. Cambridge: Cambridge University Press. — 2001 “Thinking About Things,” Philosophical Perspectives, 15, Metaphysics. Karmo, Toomas 1977 “Disturbances” Analysis, 37, 147–48. Lewis, David 1983 “New Work for a Theory of Universals,” Australasian Journal of Philosophy, 61, 343–77. — 1984 “Putnam’s Paradox,” Australasian Journal of Philosophy, 62, 221–36. — 1986 On the Plurality of Worlds. Oxford: Basil Blackwell. — 1991 Parts of Classes. Oxford: Basil Blackwell. Locke, John 1975 An Essay Concerning Human Understanding Nidditch, Peter H. (ed.), Oxford: Clarendon Press. Long, D. C. 1970 “Particulars and their Qualities,” in Universals and Particulars: Readings in Ontology, Loux, Michael (ed.), 264–84. Loux, Michael J. 1970 Universals and Particulars: Readings in Ontology, (ed.). Doubleday, New York. — 1998 “Beyond Substrata and Bundles: A Prolegomon to a Substance Ontology,” in Contemporary Readings in the Foundations of Metaphysics, Laurence, Stephen; Macdonald, Cynthia, (eds.) 233–47, Basil-Blackwell, Oxford. — 2002 Metaphysics: A Contemporary Introduction, 2nd edition, Routledge, New York. — 2003 “Individuation,” in The Oxford Handbook of Metaphysics, Loux, Michael; Zimmerman, Dean (eds.), 75–98, Oxford University Press. — 2006 “Aristotle’s Constituent Ontology,” in Oxford Studies in Metaphysics, vol 2, Zimmerman, Dean (ed.), 207–50, Oxford University Press. Lowe, E. J. 1998 “Form Without Matter,” Ratio XI 3 December, 214–34. — 2000 “Locke, Martin and Substance,” The Philosophical Quarterly, Vol. 50, No. 201, 499–514. — 2003 “Individuation,” in The Oxford Handbook of Metaphysics, Zimmerman, Dean, and Loux, Michael (eds.), 75–98, Oxford University Press. Markosian, Ned 1998b “Brutal Composition,” Philosophical Studies, 92, 211–49. Martin, C. B. 1980 “Substance Substantiated,” Australasian Journal of Philosophy, Vol. 58, No. 1, March, 3–10.

55

Mourelatos, Alexander 1978 “Events, Processes, and States,” Linguistics and Philosophy, 2, 415–34. Plantinga, Alvin 1975 “On Mereological Essentialism,” Review of Metaphysics, 28: 468–76. Quine, W. V. O. 1959 Methods of Logic. New York: Holt Dryden. — 1960 Word and Object. Cambridge, Massachusetts: The MIT Press. — 1963 From a Logical Point of View. New York: Harper & Row. Rea, Michael 1997 Material Constitution: A Reader. Rowman & Littlefield. Rescher, Nicholas 1996 Process Philosophy. Albany, NY: State University of New York Press. Sider, Theodore 1999 Review of Michael Jubien, Ontology, Modality, and the Fallacy of Reference, Cambridge: CUP, 1993. Nous 33:2, 284–94. — 2001 Four-Dimensionalism. Oxford: Clarendon Press. — 2006 “Bare Particulars,” Philosophical Perspectives 20, 387–97. Steen, Mark “Bare Objects, Ordinary Objects, and Mereological Essentialism,” manuscript. Stout, Rowland 1997 “Processes,” Philosophy, no. 72. Van Inwagen, Peter 1990 Material Beings. Ithaca, New York: Cornell University Press. — 2004 “A Theory of Properties,” in Oxford Studies in Metaphysics, vol. 1, Zimmerman, Dean (ed.), 107–39, Oxford University Press. — 2006 “A Materialist Ontology of the Human Person,” Chapter 8 in Persons: Human and Divine, van Inwagen, Peter; Zimmerman, Dean (eds.), 199–215, Oxford University Press. Wallace, Megan ms “On Composition as Identity.” This can be found at http:// www.unc.edu/~megw/OnCompAsID.pdf. Whitehead, Alfred North 1978 Process and Reality. An Essay in Cosmology, ed. by D. R. Griffin and D. W. Sherburne. New York: The Macmillan Co. Wiggins, David 1979 “Mereological Essentialism: Asymmetrical Essential Dependence and the Nature of Continuants”, Grazer Philosophische-Studien, 1979; 7/8, 297–315. — 1980 Sameness and Substance. Cambridge, Massachusetts: Harvard University Press. Wolterstorff, Nicholas 1991 “Divine Simplicity”, Philosophical Perspectives 5, Philosophy of Religion. Zimmerman, Dean 1995 “Theories of Masses and Problems of Constitution”, Philosophical Review vol. 104, No. 1, 53–110. — 1997 “Distinct Indiscernibles and the Bundle Theory”, Mind, Vol. 106, 305–9. Reprinted, with additional material, in Metaphysics: The Big Questions, ed. by van Inwagen and Zimmerman (Oxford: Basil Blackwell, 1998), 58–66.

56

Grazer Philosophische Studien 76 (2008), 57–77.

STRUCTURAL UNIVERSALS AND THE PRINCIPLE OF UNIQUENESS OF COMPOSITION Javier KALHAT

University of Zürich Summary Lewis has objected to Armstrong’s notion of a structural universal on the grounds that it violates the Principle of Uniqueness of Composition (PUC), which says that given some parts, there is only one whole that they compose. This paper reviews Armstrong’s case for structural universals, and then attempts to reconcile structural universals with PUC by arguing for the existence of arrangement universals. The latter are not only a key to defending structural universals against Lewis’ objection, but are in fact essential to Armstrong’s conception of structural universals in general. Three objections to my proposal are deflected, and two alternative proposals are shown to be inferior to it.

David Armstrong’s notion of a structural universal plays a central role in his accounts of structural resemblance, the nature of properties, property resemblance and incompatibility, and laws of nature (see, e.g., 1978b: chs. 22, 24; 1983: 77–110; 1989: chs. 6–7; 1997: chs. 4, 15). David Lewis has objected to structural universals on the grounds that they violate the Principle of Uniqueness of Composition (PUC), according to which given some parts, there is only one whole that they compose (Lewis 1986a: 36–8, 1986b).1 My aim in this paper is to advance the debate between Armstrong and Lewis. Specifically, I aim to provide Armstrong with the resources that he needs (and can accept) to handle Lewis’ objection. I shall assume therefore Armstrong’s basic metaphysical framework. Four assumptions of that framework, in particular, are worth mentioning at the outset:

1. Lewis also objects to structural universals on the ground that they involve the repetition of universals within their structure, but I will not address that objection here. For Armstrong’s attempt to meet it, see his 1986 and also 1997: 34-7.

(1) Universals can be composed of universals (see Armstrong 1997: 31–7) (2) Ordinary or ‘thick’2 particulars are states of affairs (see Armstrong 1997: 123–6) (3) States of affairs are partly composed of universals (see Armstrong 1997: 126–7) (4) Thick particulars are partly composed of universals (follows from (2) and (3); see also Armstrong 1997: 123–6) Lewis may not himself endorse (1)–(4), but neither does he directly challenge them in his debate with Armstrong. Lewis does not in fact have a problem with the idea per se of some universals being composed out of other universals. He does not, for example, impose a blanket restriction on conjunctive universals (1991: fn. 13). Instead, Lewis has a problem with a particular kind of complex universal, namely structural ones, on the grounds that their mode of composition is radically unclear.3 Assumptions (2)–(4) rest on Armstrong’s notion of a state of affairs, and it is true that Lewis also rejects them on the grounds that they violate PUC (Lewis is unhappy with structures generally). But the rejection of states of affairs is not a premise in Lewis’ argument against structural universals, and it is his argument against structural universals that is the subject of the present paper. Furthermore, my suggestion as to how to reconcile structural universals with PUC extends to states of affairs, as I shall indicate in footnote 14 below. As I see it, therefore, defending claims (1)–(4) is not only beyond the scope of this paper, but it is also dialectically unnecessary. In adjudicating a disagreement we are surely allowed to assume whatever the parties to the disagreement themselves assume (if only for the sake of argument). I appreciate of course that some philosophers might wish to subscribe to structural universals without committing themselves to claims (1)–(4), or to Armstrong’s metaphysical framework more generally. Still, what I say should be of interest to them. Armstrong’s conception of structural universals is the standard against which any alternative conception must be judged, and his debate with Lewis over structural universals is the locus classicus for a discussion of the topic. Consequently, any contribution to 2. Armstrong draws a distinction between ‘thin’ and ‘thick’ particulars (1997: 123–6). A ‘thin’ particular is a bearer of attributes, which as such has no attributes of its own. A ‘thick’ particular, on the other hand, is a thin particular taken together with its (non-relational) attributes. Examples of thick particulars include: methane molecules, apples, whales, planet earth. 3. For the distinction between conjunctive and structural universals, see section 1 below.

58

either Armstrong’s conception of structural universals or to his debate with Lewis is potentially significant. Section 1 of this paper reviews Armstrong’s reasons for believing in structural universals. I argue that only one of his ‘official’ reasons is plausible, but that his commitment to universals in general gives him a compelling reason for countenancing the existence of structural universals in particular. Section 2 then examines Lewis’ objection, and argues that it fails on the grounds that both he and Armstrong fail to realise that the structural universals they consider have arrangement universals as component parts. Indeed, it is suggested that the latter are key not only to defending structural universals against Lewis’ objection, but also for Armstrong’s conception of structural universals generally. In section 3, I consider and reject three objections to my proposal, and in section 4 I examine two alternative proposals and show them to be inferior to mine. 1. Structural universals: their nature and raison d’etre According to David Armstrong, properties can be either simple or complex.4 A property is simple if it does not have further properties as parts, and complex otherwise.5 Given his rejection of negative and disjunctive properties (e.g., being not red, being either red or green), a property can fail to be simple only if it is conjunctive (e.g., being red and circular) or structural. Suppose that to be an F, a particular must be composed of two non-overlapping particulars, one of which instantiates the property G and the other of which instantiates the property H, with the G part and the H part linked together by an external relation R. To be an F is thus to have a certain sort of structure; we may therefore call F a structural property (Armstrong 1997: 32; see also 1978b: 68–71). The main difference between structural and conjunctive properties is that in a conjunctive property, the parts (properties) are instantiated by the very same particular that instantiates the conjunctive property as a whole, while in a structural property, the parts (properties and relations) are instantiated by particulars which are only proper parts of the particular that instantiates the structural property as a whole. A plausible example of a real structural universal is the 4. In this paper, I assume that properties are universals, and though universals also include relations, ‘property’ and ‘universal’ will be used interchangeably unless otherwise stated. 5. Armstrong prefers to talk of (complex) universals as having ‘constituents’ rather than (mereological) ‘parts’. I will return to this terminological issue in section 2 below.

59

property being hydrogen. A hydrogen atom has a part that instantiates the property being an electron and a part that instantiates the property being a proton. Moreover, the electron-part and the proton-part jointly instantiate the relation of bonding.6 Being hydrogen is thus a structural property; it is the property of being structured hydrogen-wise. Why does Armstrong postulate structural universals? Following Lewis, we can discern three main reasons (see Lewis 1986a: 29–31). (1) For Armstrong, a law of nature that all Fs are Gs is a second-order lawmaking relation N holding between first-order universals F and G. But if F, G, … were all simple, then we could only get the simplest of laws, and it is unlikely that we could in this way cover all the laws of nature that there are or might have been. Hence we need structural universals. It is not the mere need for complexity, however, that secures the existence of structural universals, since the first-order universals F, G, … might be complex by virtue of being conjunctive instead. It is rather the need for a particular kind of complexity, namely, structural complexity that calls for the positing of structural universals. The examples of universals to be considered in section 2 of this paper will, I think, underwrite this claim. For they are examples of the sorts of universals that would figure in laws of nature, and they are undoubtedly structural if complex at all. (2) According to Armstrong, treating ranges of properties as structural universals yields a straightforward account of the resemblances and incompatibilities that hold among them. Take the universals being four metres long and being three metres long. These two universals resemble one another more closely than do (say) being four metres long and being one metre long. They are also incompatible with one another, since no single object can instantiate both. Now, construed as structural universals, being four metres long and being three metres long will have parts. Being four metres long, for example, will have a part that is being three metres long and a part that is being one metre long.7 But if so, the reason why being three metres long resembles being four metres long more closely than does being one metre long, is that being three metres long is a bigger proper part of being four metres long than is being one metre long. Resemblance turns 6. Of course, at the sub-atomic level, a hydrogen atom will have other parts as well. But for the purposes of this paper, we will confine ourselves to the atomic level of description in this and other examples. 7. There is, of course, an indefinite number of ways of cashing out the parts of being four metres long. I choose this one because of the particular resemblance and incompatibility in need of explanation.

60

into a case of partial identity. And the reason why being three metres long and being four metres long are incompatible with one another, is that for an object to instantiate both, that object would have to be identical with one of its proper parts—viz., the one that instantiates the lesser length property—and this is impossible. While Lewis rejects structural universals, he accepts that their affording an explanation of property resemblance and incompatibility counts as a point in their favour. However, that explanation requires that resembling and incompatible properties stand in the part/ whole relation to one another. And while this may be true for properties such as being three metres long and being four metres long, it is not true for all such properties. In particular, the account will not work for resembling but incompatible properties that are arguably simple, such as being negatively charged and being positively charged, and properties that are complex but stand to each other in the relation of overlap instead, such as being triangular and being rectangular. This weakens Armstrong’s account of property resemblance and incompatibility, and hence the view that the latter provides a reason for postulating structural universals. (3) We do not know whether there are in fact any simple properties. Perhaps all universals are structures of structures ad infinitum. If so, we cannot rule out the possibility of structural universals. Surprisingly, Lewis takes this to be Armstrong’s most compelling reason for postulating structural universals (1986a: 30). I am not sure that Armstrong and Lewis are right in doubting that we know some properties to be simple. But even if they are, the fact that we cannot rule out the possibility of structural universals surely does not by itself constitute evidence to rule them in either. At best, it provides us with a reason for remaining agnostic about their existence. Pace Lewis, however, I think that Armstrong has a fourth, and better reason for postulating structural universals. (4) The theory of universals explains objective resemblances between particulars in terms of shared universals. Two particulars can resemble one another in being alike in their structure. Therefore, the theory of universals must postulate a universal, a structural universal, to account for their resemblance. Lewis does not find this reason compelling, on the grounds that it is one thing to explain resemblance in terms of shared universals, and quite another to think that whenever two particulars resemble one another, those particulars must themselves share a universal. Thus, if A and B are

61

alike in their structure, we might be able to account for their resemblance in terms of the resemblances that hold among the corresponding parts of A and B, rather than by postulating an overall, structural universal which they share. But suppose that A and B are each composed of, say, three non-relational parts, and two relational parts. Then a specification of the resemblances that hold among those parts—say, that the non-relational parts of each are respectively Fs, Gs, and Hs, and the relational parts are respectively Rs and R*s—will simply leave unspecified the determinate way in which these various relational and non-relational parts are themselves related (are the Fs related by the Rs to the Gs, or by the R*s to the Hs, etc.?). In other words, a specification of the various partial resemblances between A and B will leave undetermined the overall structural resemblance between A and B. In such cases, therefore, we need structural universals to account for structural resemblance.8 To sum up, then, of Armstrong’s reasons for postulating structural universals, I think that (2) is a bad one, while (3) is at best a reason for remaining agnostic about their existence. (1) is a plausible reason for believing in structural universals, though it does depend upon the success of Armstrong’s account of laws of nature. But in my view, it is (4) that provides Armstrong with the most compelling grounds for believing in structural universals. If, as seems plausible, one accepts both the existence and irreducibility of objective structural resemblances in reality, then the 8. It might be thought that a better counterexample to Lewis’ claim here is a case where two structurally alike particulars have qualitatively different non-relational parts, linked by qualitatively different relations. Thus, consider the case of a ‘shmydrogen’ atom, which is structurally just like a hydrogen atom but which is such that each of the simpler properties and relations entering into its structure are different. To be a shmydrogen atom, F*, a particular must be composed of two other particulars, one of which instantiates G*, being a shmoton, and the other instantiates H*, being a shmelectron, with the part that is G* and the part that is H* being related by the (symmetric) external relation R*, shmonding. The claim would be, then, that there is an important resemblance between these two atoms that can only be explained in terms of their overall structure. Such a challenge is ineffective against Lewis, however, for it trades on a different notion of structure from that employed by him (and by Armstrong). The notion of structure employed in the shmydrogen case is a thin notion of structure, one which involves no reference to the nature of the parts involved, but only to their quantity and distribution (spatial, etc.). The notion of structure employed by Lewis and Armstrong, on the other hand, is a thicker notion, one which does make reference to the nature of the parts involved. The thicker notion of structure will be further elucidated in section 2, where structural universals will be distinguished from what I call arrangement universals. (I thank the referee for this journal for suggesting the shmydrogen case.)

62

rationale for postulating universals in general forces one to recognise the existence of structural universals in particular. I now turn to Lewis’ objection to structural universals. 2. Are there structural universals composed of exactly the same parts? Lewis objects to structural universals on the grounds that they afford prima facie counter-examples to the Principle of Uniqueness of Composition: PUC There is only one mode of composition; and it is such that, for given parts only one whole is composed of them (1986b: 92). An examination of PUC is beyond the scope of this paper. It is the mereological analogue of the principle of extensionality in set theory (which says that sets with exactly the same members are identical), and it is one of the axioms of classical mereology—the theory of the part/ whole relation originally developed by Lesniewski, and later defended by Goodman and Quine, as well as by Lewis (see Simons 1987; Lewis 1986a: 36–7 and 1991:72–87). I take it that for those who subscribe to PUC, the intuitive force behind the principle is the belief that unless non-identical composite objects differed in at least one of their parts, there would then be nothing to ground their difference. “No difference without a difference maker” —so runs the slogan.9 9. PUC might nevertheless seem straightforwardly implausible. Surely, one might insist, two different wholes can have the same parts at different times—as in the case of a Lego house that is taken apart and then put back together as a Lego car—and surely the same whole can have different parts at different times—as in the case of a Lego house which undergoes the replacement of (say) one of its windows for a door. These counterexamples, however, rest on an ‘endurantist’ conception of how objects persist through time, according to which objects are wholly present at every moment of their existence. And Lewis has forcefully argued against such a conception, opting instead for ‘perdurantist’ view, according to which objects persist through time by having different temporal parts at the different moments of their existence (see Lewis 1986c). On Lewis’ view, then, the Lego house and car are composed of different temporal parts of the same persisting Lego units. And the house-with-window and house-with-door are themselves different temporal parts of the same persisting Lego house. The perdurantist view of persistence is controversial, of course, but inevitably I cannot examine it here. In any case, while one way of meeting Lewis’ objection would be to argue against PUC (by e.g., rejecting perdurantism), doing so seems to be strategically in error if the objection can be met more immediately—i.e., within the terms of the debate—by showing that structural universals are in fact consistent with PUC. It is the latter option that I am pursuing in this paper.

63

Now, structural universals supposedly run counter to PUC because it seems that two different structural universals can be composed of exactly the same universals. Take the case of being butane and being isobutane. According to Lewis, they are both composed of the non-relational parts being carbon and being hydrogen, and of the relational part bonding. Since PUC is a principle of mereology, if structural universals have any composition at all, they must then have a non-mereological mode of composition. But this also runs counter to PUC, which specifies that mereological composition is the only form of composition. Thus, structural universals must have no composition, i.e., they must be simple. But they cannot be simple for they necessarily involve simpler universals. Thus, if PUC is correct there can be no structural universals. While Lewis rejects structural universals because they violate PUC, Armstrong rejects PUC because it rules out structural universals, and structures more generally (Armstrong 1986: 85–6). Their disagreement, however, is based on a premise which they both accept, namely, that if there are structural universals, then different structural universals can be composed of exactly the same parts. In my view this premise can be shown to be false. And once it is rejected, there is no obstacle to believing in both structural universals and PUC. Before proceeding with the argument, however, I need to address a terminological issue already mentioned in section 1. While Armstrong rejects PUC because he denies that there is only mereological composition, he does agree with Lewis that ‘part’ is a mereological notion, and as such that the following is true of it: for given parts, there is only one whole that they compose. At the same time, Armstrong does not think that the latter is true of the entities that make up his structural universals, and hence prefers to call those entities ‘constituents’ rather than ‘parts’. Since in this paper I am precisely trying to resist the claim that the entities that make up a given structural universal can in principle compose more than one such universal, I shall refrain from drawing Armstrong’s distinction between ‘parts’ and ‘constituents’. I will thus continue to talk about structural universals as having parts, and insofar as they do not conflict with PUC (which is yet to be shown, of course), ‘part’ will be used in the good old mereological sense of that word. Now back to the argument. Consider, first, molecules of butane and isobutane. On the face of it, they differ qualitatively despite being composed of an equal number of qualitatively identical parts: four carbon parts, ten hydrogen parts and thirteen bonding parts. Chemists, however, can explain

64

the qualitative difference that remains between these molecules by noting that they are structural isomers, that is, they are chemical compounds that have the same molecular formula (C4H10) but different bonding arrangements (Brown et al. 1994: 960–1). Butane (CH3CH2CH2CH3), for example, is a hydrocarbon with a straight-chain bonding arrangement, while isobutane ((CH3)3CH) is a hydrocarbon with a branched-chain bonding arrangement (see Figure 1 below).

Butane (CH3CH2CH2CH3)

Isobutane (CH3)3CH Figure 1. Butane and isobutane molecules [Source: http://classweb.gmu.edu/sslayden/graphics/graphics.htm]

We can think of the bonding arrangement of a molecule as the spatial arrangement of its relational and non-relational parts. It is determined by what chemists call ‘bond angles’, which are the angles at which the nuclei of the various non-relational parts of the molecules stand (Brown et al. 1994: 286-97). From the point of view of ontology, I suggest that we think of a bonding arrangement as a second-order relation, i.e., a relation that includes a first-order relation among its relata. A bonding arrangement is the second-order relation in which the non-relational and first-order relational parts of a molecule stand. Thus, just as each hydrogen part in a butane molecule stands in the relation of bonding to a carbon part, so the

65

four carbon parts, the ten hydrogen parts and the thirteen bonding parts stand in the second-order relation of being butane-wise arranged. Molecules of butane and isobutane qualitatively differ, then, because they have different bonding arrangements. At this point, it might be wondered what the difference is between a structural and an arrangement universal. For I have said that the structural universal being butane is the universal of being structured butane-wise, while its arrangement part is the universal of being butane-wise arranged. But what is the difference between being structured butane-wise and being butane-wise arranged? The difference is that the structural universal being structured butane-wise includes the non-relational and first-order relational parts, whereas the arrangement universal does not. The point can be made clearer by appealing to Armstrong’s own conception of universals as ‘statesof-affairs types’ (1997: 28-29, 34–7). According to Armstrong, universals are ‘gutted’ states of affairs, i.e., they are what remain from states of affairs once the particulars involved are abstracted away. So, consider first of all a molecule of (say) butane: H

H

H

H

H—C—C—C—C—H H

H

H

H

Armstrong takes a molecule of butane to be identical with a state of affairs. Letting a–n stand for particulars, ‘H’ for the property of being hydrogen, ‘C’ for the property of being carbon, and the lines connecting the various non-relational parts for the relation of bonding, we can represent a particular molecule of butane as the following state of affairs S: Hb

Hd

Hc

He

Ha — Ck — Cl — Cm — Cn — Hf Hj

Hh

Hi

Hg

Armstrong next tells us that the structural universal being butane is what remains of the state of affairs S, once the particulars are abstracted away.

66

Replacing a–n by variables x1–x14 then, we obtain the structural universal being butane: Hx2

Hx3

Hx4

Hx5

Hx1 — Cx11 — Cx12 — Cx13 — Cx14 — Hx6 Hx10

Hx9

Hx8

Hx7

I now suggest that we can construe the arrangement universal being butane-wise arranged as what remains of the structural universal being butane once the nonrelational and first-order relational universals have been abstracted away. Let us replace, then, the non-relational part H by the variable F and the non-relational part C by the variable G. And let us replace the relational parts—viz., the bonding parts represented by the lines connecting the non-relational parts of the structural universal—by double-lines (‘ ’). What we get, then, is the arrangement universal being butane-wise arranged:

Fx1

Fx2

Fx3

Fx4

Fx5

Gx11

Gx12

Gx13

Gx14

Fx10

Fx9

Fx8

Fx7

Fx6

Arrangement universals, then, are arrived at through the same process of abstraction that yields the structural universals of which they are parts. They differ from the latter, however, in their degree of incompleteness.10 Earlier, I argued that the qualitative difference between butane and isobutane molecules is grounded in a difference in their respective bonding arrangements. From this suggestion there is but a short step to the further suggestion that those bonding arrangements are themselves parts of the structure of butane and isobutane molecules. For there is a perfectly good 10. My earlier claim that arrangement universals are second-order relations might now come under pressure on the grounds that since they are parts of structural universals, and the latter are instantiated by particulars, arrangement universals are really only first-order relations. But as we have just seen, the relata of an arrangement universal include first-order universals; this makes them by definition second-order.

67

and ordinary sense of the part/ whole relation according to which a part of the whole is anything that constitutes the whole. It is in this sense, for example, that the property being red is a part of the conjunctive property being red and circular, and that the relation of bonding is a part of the structural universal being methane. Now, that the bonding arrangements of butane and isobutane molecules are constitutive of their respective structures is undeniable since it is they after all that explain the qualitative difference between those structures. Accordingly, bonding arrangements are parts of the respective structures of butane and isobutane molecules. A simpler route to this conclusion is the conjunction of the following two claims: (1) relations are unproblematically taken to be parts of the structure of molecules (insofar as they are constitutive of their structure); (2) bonding arrangements are (second-order) relations.11 We have so far been talking about butane and isobutane molecules, but Lewis’ objection concerns the structural universals being butane and being isobutane. It should be clear, nevertheless, how our findings regarding the molecules relate to the structural universals they instantiate. Just now, I suggested that the bonding arrangements of butane and isobutane molecules are parts of their respective structures—they are, indeed, the parts that qualitatively distinguish between those structures. Now, structural universals are universals of structure. The structural universal being 11. It might be thought that a different kind of structure, namely a state of affairs such as Fa constitutes a counterexample to the claim that anything that constitutes the whole is thereby a part of it. a and F are constitutive elements of Fa, but they are not parts of it, since if they were, the mere existence of the elements would guarantee the existence of the state of affairs, which it does not. It is far from obvious, however, that the composition of Fa is exhausted by a and F. Armstrong himself has observed that there is a long tradition of metaphysical thinking that takes there to be a further constitutive element involved, one that binds the particular and the attribute together (e.g., a relation of exemplification or instantiation, a ‘copula’, a ‘non-relational tie’, etc.; see Armstrong 1997: 115ff). It is true that for many this suggestion is undermined by Bradley’s regress, but that depends on how the further, binding element is construed. My own view of the matter is that a state of affairs such as Fa does indeed contain a binding element, namely a particularised relation. It misunderstand the nature of the relation in question, however —indeed the nature of any relation as such—to think that it itself require a further relation to bind it to a and F. It is in the nature of a relation to relate, and as such it does not require anything additional to do so (cf. section 3 below). Suppose, then, that Fa contains the particularised relation of binding the particular a to the attribute F, and suppose too that Armstrong is right in thinking that there are no uninstantiated attributes (see 1997: 38–43). The binding relation can then only be instantiated by a and F, and only exists if it is in fact so instantiated. It now does seem plausible to take the existence of a, F and the binding relation to guarantee the existence of Fa. If so, the state of affairs Fa is not a counterexample to the claim that anything that constitutes a whole is thereby a part of it.

68

butane, for example, is the universal of being structured butane-wise, while the structural universal being isobutane is the universal of being structured isobutane-wise. It is to be expected, therefore, that any constitutive parts of a structure will register in some way or other in the universal of being that particular structure. One way for a part of a structure to register within the universal of being that structure is for it to be a part of the structural universal itself. Relations constitute a notable case of such ‘dual citizenship’. The relation of bonding, for example, which is a constitutive part of the structure of individual hydrogen atoms, is also a part of the structural universal being hydrogen, which those atoms instantiate. Now, since the bonding arrangements being butane-wise arranged and being isobutane-wise arranged are constitutive parts of the structure of butane and isobutane molecules respectively, they must also register within the structural universals being butane and being isobutane. At the same time, as we have already seen, these bonding arrangements are themselves relations. The precedent of other relations, such as bonding, makes it plausible therefore to take the bonding arrangements being butane-wise arranged and being isobutane-wise arranged to be themselves parts of the structural universals being butane and being isobutane, respectively. The relevance of this finding to Lewis’ objection is now obvious: since being butane and being isobutane differ in their bonding arrangement parts, it is simply not true that these structural universals violate PUC. Lewis’ claim to the contrary rests on an inadequate characterisation of the structure of butane and isobutane molecules, and a fortiori of the structural universals the latter instantiate. In light of our discussion so far, I think that two plausible principles emerge concerning the composition of Armstrongian structural universals generally. The first is: (I) Every structural universal has an arrangement universal as a part For every structural universal will be constituted, not merely by its nonrelational and (first-order) relational parts, but also by the second-order relation in which those parts themselves stand. And that relation is what I have called their arrangement universal.12 12. Of course, not all structural universals will have bonding arrangements as their arrangement parts (though those drawn from chemistry typically will). Bonding arrangements are spatial second-order relations, but other arrangements might be second-order relations of a different kind.

69

If every structural universal has an arrangement part, then I think it is plausible to hold that any difference between two structural universals must indeed amount to some difference in their parts. Hence, the second principle: (II) The difference between any two structural universals is accountable in terms of a difference in (a) their non-relational parts, and/ or (b) their (first-order) relational parts, and/ or (c) their arrangement parts If these two principles are correct, then both Lewis and Armstrong are wrong in thinking that if there are structural universals, different such universals can be composed of exactly the same parts. 3. Three objections If I am right in thinking that postulating arrangements parts will adequately meet Lewis’ objection to structural universals, then it is rather surprising that when writing about another kind of structure, namely, states of affairs, Armstrong rejects the view that the arrangement of the parts is itself a part of the whole. He writes: in the case where distinct states of affairs have exactly the same constituents, nevertheless the arrangement of the constituents is different. Arrangement of constituents is not a further constituent (1991: 192).13

It is not clear to me why Armstrong rejects arrangements as parts of states of affairs, but he may have in mind the threat of a Bradley-type regress. If the arrangement of the parts of a whole were itself a part of the whole, then will we not also require an arrangement of the parts and the first arrangement in order to get the whole, and in turn, an arrangement of the parts, the first arrangement and the second arrangement, and so on to infinity? However, I think that acceptance of arrangements as parts leads to no more of an infinite regress than does acceptance of relations as parts. To see this, consider the schematic state of affairs a’s having the relation R to b, or S for short. Armstrong takes such a state of affairs to be composed of the non-relational parts a and b, and of the relational part R. Now, Armstrong 13. For reasons already given, I ignore here Armstrong’s terminological distinction between ‘parts’ and ‘constituents’.

70

would reject as confused the objection that if the relation R were a part of S, then we would require a further relation R1 to relate R to a and b, and in turn a relation R2 to relate R1 to R, and so on ad infinitum. Taking R to be a part of S does not lead to an infinite regress of relations, because R is a part that relates other parts, and simply does not require further relations to do so (Armstrong 1997a: 30). But if so, why not say a similar thing about the arrangement parts? Why not say that an arrangement is a part that arranges other parts, and does not require further arrangements to do so. If Armstrong accepts relations as parts, then he equally ought to accept arrangements as parts. Perhaps, however, Armstrong is thinking of a different sort of regress when he rejects arrangements as parts. Consider this time our butane and isobutane molecules. They have qualitatively identical non-relational and first-order relational parts, but different arrangements of those parts. Hence, I have suggested, butane and isobutane molecules differ in their arrangement parts. But now, aren’t those arrangement parts themselves composed of the same non-relational and first-order relational parts? And if so, won’t we now need to postulate further arrangement parts to distinguish between the different (first-order) arrangement parts, and so on ad infinitum? We only get this regress, however, if we take the arrangement parts of butane and isobutane molecules to be composed of their non-relational and first-order relational parts. But I have already rejected that view. On my conception, arrangement universals result from the removal of the non-relational and first-order relational parts. One might be less inclined to think of arrangement parts as having composition if the only available ‘constituents’ are gaps. And if arrangement parts have no composition, i.e., if they are simple, then it seems plausible to take their difference as primitive. I find this way of providing a terminus for the regress attractive, but since it raises thorny issues about the composition of relations in general, I want to leave it as a fallback position only. A less controversial way of stopping the regress is this. Recall that the arrangement parts of butane and isobutane molecules are determined by bond angles, i.e., the angles at which the nuclei of the various non-relational parts of the molecules stand. Since the arrangement parts are determined by those bond angles, the difference between the arrangement parts must ultimately be grounded in a difference between the bond angles. We can therefore stop the regress by invoking those different bond angles. (There may still be a need to account

71

for the difference between the bond angles, but it is unclear that they will themselves turn out to be composed of the very same parts.) Before concluding this section, I wish to consider a third and final objection to the proposal I have defended here. The objection is that while positing arrangement universals might allow us to qualitatively distinguish between some molecules that are otherwise composed of qualitatively identical parts, it does not enable us to distinguish between all such molecules. Consider the case of enantiomers, i.e., isomers that are non-super-imposable mirror images of one another. In a pair of enantiomers, one isomer will exhibit the property of being right-handed and the other the property of being left-handed. (This is determined by the effect that each enantiomer has on the plane polarised light passing through it. Thus, one enantiomer in the pair will rotate the light to the right by n degrees, while the other will rotate the light by the same number of degrees to the left.) Examples of enantiomers include molecules of lactic acid and alanine. The latter is illustrated below:

Figure 2. Enantiomers of alanine [Source: http://www.daviddarling.info/encyclopedia/E/enant.html]

Now, enantiomers such as those of alanine instantiate structural universals that are (according to my proposal at any rate) composed of the same arrangement universals in addition to the same relational and non-relational parts. Yet, it is claimed, they nevertheless differ in violation of PUC. The objection fails. Enantiomers differ in their spatial orientation, and accordingly instantiate different universals of spatial orientation. But the fact that the enantiomers differ in their spatial orientation does not entail

72

that the structural universals they exemplify themselves thereby differ (despite being composed of identical parts). This is because the spatial orientation of the enantiomers is no part of their structure at all. Otherwise we would have to say that every time we rotate an object in space we modify its structure! The only thing that enantiomers show, as far as we are concerned, is that molecules that exemplify one and the same structural universal can nevertheless differ in other respects, notably by instantiating different universals of spatial orientation. 4. Two alternative proposals I wish to conclude my paper by comparing my proposal as to how to deal with Lewis’ objection against structural universals with two other proposals that can be found in the literature. In 1997, Armstrong modified his original account of universals (1978b) by construing them as types of states of affairs (1997: 28–9). Take, for example, all the existing butane molecules in the universe. As we have seen, Armstrong identifies each one of them with a particular state of affairs. Since all these states of affairs are qualitatively identical, we can think of them as tokens of a particular type of state of affairs, namely, the butane type. Armstrong’s suggestion, then, is that we identify the universal being butane with that type. What goes for being butane also goes for being isobutane (and every other universal). Thus, being isobutane just is the type of state of affairs of which each isobutane molecule is but a token. Armstrong’s new account of universals in terms of types of states of affairs might be thought to give him the resources to deal with Lewis’ objection. For it is clear that butane states of affairs are different from isobutane states of affairs. And that difference, you might think, must surely register at the level of the types of state of affairs of which they are tokens. The suggestion is intuitive enough, I think, but the trouble is to see how exactly the difference registers at the level of the types. It is, of course, not the numerical difference between butane and isobutane molecules that needs registering, but rather their qualitative difference—the difference in structure, to be more specific. But it is simply unclear how that difference registers at the level of the types. The point can be reinforced by noting that, according to Armstrong, we derive the type of state of affairs (say butane) from the token states of affairs (butane molecules) by removing all particulars from the latter. But if we do so, we are merely left with the

73

constituent properties and relations, viz., the property of being carbon, the property of being hydrogen and the relation of bonding. And if those are the constituents of the type butane, then Lewis’ problem is as much a problem for the types as it is for the universals with which they are identified. I do not think, therefore, that Armstrong’s account of universals in terms of types of states of affairs helps with Lewis’ problem, and hence I do not think that the account obviates the need to postulate arrangement universals as parts of structural universals. A different proposal for dealing with Lewis’ objection to structural universals has been developed by Joan Pagès (2002: 215–21). Pagès’ suggestion is that the difference between, say, being butane and being isobutane can be explained by postulating certain ‘formal relations’ which are constitutive parts of these structural universals (2002: 215–21). To illustrate, consider Pagès’ own schematic example of the structural universals being a molecule of D and being a molecule of E. Both structural universal are composed of the properties being an atom of A and being an atom of C, and the relation of bonding. We can represent them as follows: D

A C

E

A

C A

A

Pagès now attempts to distinguish between D and E by postulating certain first-order formal relations. D contains the formal relation R, which is designated by the open formula

(1) Ax & Cy & Az & B(x, y) & B(y, z) In Pagès’ own words: ‘a sequence of three particulars instantiates R if and only if the first particular is an A-atom, the second particular is a C-atom, the third particular is an A-atom, the first particular is bonded to the second particular and the second particular is bonded to the third particular’ (2002: 218). E, on the other hand, contains the different formal relation S, which is designated by the open formula

74

(2) Cx & Ay & Az & B(x, y) & B(y, z) In Pagès’ own words: ‘a sequence of three particulars instantiates S iff the first particular is a C-atom, the second particular is an A-atom, the third particular is an A-atom, the first particular is bonded to the second particular and the second particular is bonded to the third particular’ (2002: 218). Since R and S are evidently different formal relations, it appears that we can distinguish the two structural universals being a molecule of D and being a molecule of E by appealing to the difference between R and S. Mutatis mutandis for the structural universals being butane and being isobutane. My proposal is similar to Pagès’ in two respects: we both think that a further element of the relevant structural universals has to be postulated in order to distinguish between them, and we both take that element to be a relation. But my proposal differs from Pagès’ in the following four respects: first, his formal relations are first-order relations while my arrangement universals are second-order ones. Secondly, while I take arrangement universals to be parts of structural universals, he takes his formal relations to be only constituents of them. His proposal thus makes use of a theoretical distinction that I have no need for, namely, the part/constituent distinction. In view of the problematic nature of this distinction14, a proposal that makes no use of it is, other things being equal, to be preferred. Thirdly, my proposal is firmly rooted in chemistry whereas his is not. For it will be recalled that the arrangement universals I appeal to in order to distinguish between being butane and being isobutane are the bonding arrangements constitutive of the respective structures of butane and isobutane molecules. And those bonding arrangements are discovered by the chemist; they are certainly not postulated by the philosopher. Given Armstrong’s pervasive naturalism, a proposal that is firmly rooted in scientific finding has surely an advantage over one that is not, other things being equal. Fourthly, and following on from this, my proposal appears to be more economical than Pagès’. For assuming that Pagès has no good empirical reason for rejecting the chemist’s notion of a bonding arrangement, he is already committed to bonding arrangements in addition to his formal relations, whereas I am only committed to the bonding arrangements. In view of these four differences between Pagès’ proposal and mine, I conclude that the latter is to be preferred. 14. See Lewis 1986a and 1986b; Scaltsas 1990.

75

To sum up, then, both Lewis and Armstrong are wrong in thinking that structural universals violate PUC. Their claim to the contrary rests on a failure to realise that the structural universals they consider—and indeed Armstrongian structural universals generally—include arrangement universals as component parts.15,16 REFERENCES Armstrong, D. 1978a. Nominalism and Realism, Cambridge: Cambridge University Press. — 1978b. A Theory of Universals, Cambridge: Cambridge University Press. — 1983. What is a Law of Nature? Cambridge: Cambridge University Press. — 1986. “In Defence of Structural Universals”, Australasian Journal of Philosophy, 64: 85–88. — 1991. “Classes are States of Affairs”, Mind 100: 189–200. — 1997. A World of States of Affairs, Cambridge: Cambridge University Press. Brown, T. L., Lemay, H.E., and Bursten, B.E. 1994. Chemistry: The Central Science, USA: Prentice-Hall. Lewis, D. 1986a. “Against Structural Universals”, Australasian Journal of Philosophy 64: 25–46. — 1986b. “A Comment of Armstrong and Forrest”, Australasian Journal of Philosophy 64: 92–3. 15. Though the topic of this paper has been structural universal, it is worth noting (especially in view of the assumptions (2)–(4) stated at the outset of this paper) that arrangement universals can also be used to reconcile states of affairs with PUC. Lewis objects to states of affairs on the grounds that in some cases, given some parts, two different states of affairs can be composed out of them, which is impossible. Thus, given four carbon parts, ten hydrogen parts and thirteen bonding parts, two different molecular states of affairs can be composed, a butane state of affairs or an isobutane state of affairs. But just as in the case of the structural universals being butane and being isobutane, Lewis here overlooks the existence of an additional part in each state of affairs which grounds the difference between them, namely, their respective arrangement parts. Butane states of affairs have a butane-wise arranged part while isobutene states of affairs have an isobutenewise arranged part. Butane and isobutane states of affairs do not violate PUC, therefore. As in the case of structural universals, the suggestion that different wholes with seemingly identical parts can be differentiated by their arrangement universals can plausibly be extended to states of affairs generally. The difference between any two states of affairs will thus be accountable for either in terms of a difference among their non-relational and/ or first-order relational parts, or in terms of a difference in their arrangement parts. 16. For comments and suggestions on earlier drafts, I am grateful to the referee for this journal, Stephan Blatti, Philip Goff, Sebastian Kalhat, David Oderberg, Galen Strawson, Daniel Whiting, and especially Hanjo Glock.

76

— 1986c. On the Plurality of Worlds, Oxford: Blackwell. — 1991. Parts of Classes, Oxford: Blackwell. Pagès, J. 2002. “Structural Universals and Formal Relations”, Synthese 131: 215– 221. Scaltsas, T. 1990. “Is a Whole Identical to its Parts?”, Mind 99: 583–598. Simons, P. 1987. Parts: A Study in Ontology, Oxford: Oxford University Press.

77

Grazer Philosophische Studien 76 (2008), 79–90.

THE ‘GREAT STRUGGLE’ BETWEEN CANTORIANS AND NEO-ARISTOTELIANS: MUCH ADO ABOUT NOTHING Miloš ARSENIJEVIĆ & Miodrag KAPETANOVIĆ University of Belgrade Summary Starting from the generalized concept of syntactically and semantically trivial differences between two formal theories introduced by Arsenijević, we show that two systems of the linear continuum, the Cantorian point-based system and the Aristotelian interval-based system that satisfies Cantor’s coherence condition, are only trivially different. So, the ‘great struggle’ (to use Cantor’s phrase) between the two contending parties turns out to be ‘much ado about nothing’.

1. Introduction

According to what Aristotle called Zeno’s axiom,1 no n-dimensional entity consists solely of (n-1)-dimensional entities. Accepting Zeno’s axiom and rejecting atomism at the same time, Aristotle established an interval-based conception of the continuum that ‘involves indeterminate parts’ and was therefore later called ‘indefinitism’.2 This conception was the official conception of the continuum till the end of the nineteenth century, although mathematicians often used infinitesimals as ‘useful fictions’,3 and physicists sometimes endorsed atomism (but not as an analysis of the continuum). Because the competing ‘theorists […] either leave ultimate elements of matter totally indeterminate, or […] they assume them to be so-called atoms of very small, yet not entirely disappearing space-contents’, the ‘great struggle’ among the followers of Aristotle and Epicurus so scandalized Georg Cantor that he was unwilling to let the matter go unresolved.4 Boldly rejecting Zeno’s axiom, Cantor established the point-based conception 1. 2. 3. 4.

See Aristotle, Metaphysics 1001 b 7. See Leibniz G., II, pp. 281–83. See ibid. GM, IV, p. 91–95. Cantor 1962, p. 275.

of the continuum, stating that a linearly ordered set of null-dimensional points actually makes up a continuum if the set is perfect and coherent (zusammenhängend),5 which means that each element of the set is an accumulation point of an infinite number of elements of the set, whereas each accumulation point of an infinite number of elements of the set is an element of the basic set itself. The revival of infinitesimalism6 and the formalization of the non-Archimedean system of the continuum7 did not prevent Cantorians from dominating twentieth-century mathematics just as Aristotelians had dominated the subject till the end of the nineteenth century. Cantor’s theory has been enormously influential. Logicians have formalized it, mathematicians have accepted it as a basis for Standard Analysis, and because of it philosophers have changed their mind about the structure of the physical world. Even physicists haven’t quantized space and time, in spite of the fact that they acknowledge the existence of the quantum of action. But though the majority of mathematicians and scientists sided with Cantor’s view, and many prominent philosophers did considerable work to defend it as the ontology of the physical world,8 in the last three decades of the last century a number of authors revived the Aristotelian stretchbased approach.9 However, in his 2003 article Arsenijević10 argued that there are interesting cases in which two axiomatic systems, or two formal theories, which are syntactically and semantically non-trivially different in the standard sense of these terms should be rather classified as only trivially different. What we now want to prove is that the point-based and the interval-based system of the continuum represent a remarkable instance of such a case, so that the ‘great struggle’ between Cantorians and Neo-Aristotelians turns out to be ‘much ado about nothing’. Arsenijević has formulated two sets of translation rules that he has shown11 to be sufficient for the mutual translatability of the formulas of 5. Ibid. p. 190. 6. See Ehrlich 2005. 7. Robinson, A. 1974. 8. For instance, Russell 1903 and 1914, Carnap 1928, Grünbaum 1952 and 1974, Salmon 1975, Robinson, D. 1989, Lewis 1994, Earman and Roberts 2006. 9. Hamblin 1969 and 1971, Humberstone 1979, Foldes, 1980, Needham 1981, Burgess 1982, Comer 1985, White 1988, Bochman 1990a and 1990b, Benthem van 1991 and 1995, Roeper 1997, and 2006. 10. Arsenijević 2003, pp. 2–4. 11. Ibid. pp. 8–9.

80

the point-based and the interval-based systems when their axioms are so selected that they implicitly define linearly ordered dense structures. The leading idea of Arsenijević’s proof that these systems are trivially different consists in using as the basis of the translations the so-called Felix Bernstein’s mapping12 between the two sets of formulas, which is a 1-1 mapping of all the formulas of each of the two sets into (and not onto) the set of formulas of the other set.13 From a purely syntactical point of view, it is sufficient that, in both directions, each theorem, but no non-theorem, is translated by a theorem. From a semantic point of view, however, it is not sufficient that, in both directions, each truth, but no falsehood, is translated by a truth, since it is necessary that each translation also be structure preserving. This means that both the elements and relations of a model of one of the systems must be unequivocally spoken of in terms of the elements and relations of a model of the other system. In the case at hand, the foregoing requirement will be satisfied because any element of the interval structure will be unequivocally identified and spoken of as a stretch between two distinct points of the point structure, whereas each element of the point structure will be unequivocally identified and spoken of as an abutment place of two abutting stretches of the stretch structure. The identity and precedence relations of either of the two structures will be unequivocally definable via the identity and precedence relations of the other structure in the way in which it is done below (see translation rules C1 and C2, and also C*1 and C*2). Now, the main problem of using the translation rules formulated by Arsenijević lies in the fact that those rules are tailored to first-order languages, whereas the continuity axiom (defining implicitly the coherence of the set of points and the set of stretches) is normally formulated in a second-order language. In order to avoid this problem, we shall use the Lω1ω1 language to express the continuity axiom in both systems. These formulations will make it possible to extend the applicability of the translation rules formulated for the first-order languages without any modification. As for the difference between translation rules concerning quantifiers (C5 and C*5) as originally formulated by Arsenijević and as they are formulated below,14 the new formulation has an obvious advantage in simplicity, 12. See Cantor 1962, p. 450. 13. See Arsenijević 2003, pp. 3, 8–9. 14. Compare the two original formulations in Arsenijević 2003, pp. 8–9 with the two given below.

81

but it will also enable us to translate the continuity axioms of the two systems into a considerably shorter form (see the comments below about the shorter form of translating the closed formulas). In the next Section, we shall desribe the two formal theories and cite the re-formulated translation rules in full, so that the article will be selfcontained. 2. Comparison between the point-based and the interval-based systems of the continuum according to the generalized definition of the trivial differences between formal theories Let SP contain —besides the logical constants ¬, ⇒, ∧, ∨ and ⇔ —individual variables α1, α2,…, αi,…, β1, β2,…, βi,…, γ1, γ2,…, γi,…, δ1, δ2,…, δi,…, quantifiable by universal and existential quantifiers. The variables are supposed to range over a set of null-dimensional points. Also let SP contain relation symbols ≡, < and >, to be interpreted as the identity, precedence, and succession relations respectively. Let the elementary wffs of SP be αm ≡ αn, αm < αn and αm > αn, where am > an ⇔def. an < am. Finally, let axiom schemes of SP be the following ten formulas, which we shall refer to as (AP1), (AP2),…, (AP10): (αn)¬αn < αn (αl)(αm)(αn)(αl < αm ∧ αm < αn ⇒ αl < αn) (αm)(αn)(αm < αn ∨ αn < αm ∨ αm ≡ αn) (αl)(αm)(αn)(αl ≡ αm ∧ αl < αn ⇒ αm < αn) (αl)(αm)(αn)(αl ≡ αm ∧ αn < αl ⇒ αn < αm) (αm)(∃αn)αm < αn (αm)(∃αn)αn < αm (αm)(αn)(αm < αn ⇒ (∃αl)(αm < αl ∧ αl < αn)) (α1)(α2)…(αi)… ((∃β1)(∧1≤i<ω αi < β1) ⇒ ⇒ (∃γ1)(∧1≤i<ω αi < γ1 ∧¬(∃δ1)(∧1≤i<ω αi < δ1 ∧ δ1 < γ1))) 10. (α1)(α2)…(αi)… ((∃β1)(∧1≤i<ω αi > β1) ⇒ ⇒ (∃γ1)(∧1≤i<ω αi > γ1 ∧ ¬(∃δ1)(∧1≤i<ω αi> δ1 ∧ δ1 > γ1)))

1. 2. 3. 4. 5. 6. 7. 8. 9.

Let SI contain — besides the logical constants ¬, ⇒, ∧, ∨ and ⇔ — individual variables a1, a2,…, ai,…, b1, b2,…, bi,…, c1, c2,…, ci,…, d1, d2,…, di,…, quantifiable by universal and existential quantifiers. The variables are supposed to range over one-dimensional stretches. Also let SP contain

82

the relation symbols =, ≺, , ⎨, ∩ and A, to be interpreted as the identity, (total) precedence, (total) succession, abutment, overlapping and inclusion relations respectively. Then, the elementary wffs will be am = an, am ≺ an, am an, am ⎨ an, am ∩ an, and am A an. But though it will be sometimes useful and much easier to use all these relations, it is important to note that not only , but also ⎨, ∩ and A, are definable via = and ≺ in the following way: am an ⇔def. an ≺ am, am ⎨ an ⇔def. am ≺ an ∧ ¬(∃al)(am ≺ al ∧ al ≺ an), am ∩ an ⇔def. (∃al)(∃ak)(al ≺ an ∧¬al ≺ am ∧ am ≺ ak ∧ ¬an ≺ ak), am A an ⇔def. ¬am = an ∧ (al)(al ∩ am ⇒ al ∩ an). Finally, let axiom schemes of SI be the following ten formulas, which we shall refer to as (AI1), (AI2),…, (AI10): (an)¬an ≺ an (ak)(al)(am)(an)(ak ≺ am ∧ al ≺ an ⇒ ak ≺ an ∨ al ≺ am) (am)(an)(am ≺ an ⇒ am ⎨ an ∨ (∃al)(am ⎨ al ∧ al ⎨ an)) (ak)(al)(am)(an)(ak ⎨ am ∧ ak ⎨ an ∧ al ⎨ am ⇒ al ⎨ an) (ak)(al)(am)(an)(ak ⎨ al ∧ al ⎨ an ∧ ak ⎨ am ∧ am ⎨ an ⇒ al = am) (am)(∃an) am ≺ an (am)(∃an) an ≺ am (am)(∃an) an A am (a1)(a2)…(ai)…((∃u)(∧1≤i<ω ai ≺ u) ⇒ ⇒ (∃v) (∧1≤i<ω ai ≺ v ∧ ¬(∃w) (∧1≤i<ω ai ≺ w ∧ w ≺ v))) 10. (a1)(2)…(ai)…((∃u)(∧1≤i<ω ai u) ⇒ ⇒ (∃v)(∧1≤i<ω ai v ∧ ¬(∃w)(∧1≤i<ω ai w ∧ w v)))

1. 2. 3. 4. 5. 6. 7. 8. 9.

Now, let f be a function f : αn ⎯→ 〈a2n−1, a2n〉 (n = 1, 2,…) mapping variables of SP into ordered pairs of variables of SP, and let C1–C5 be the following translation rules providing a 1-1 translation of all the wffs of SP into a subset of the wffs of SI (where =C means “is to be translated according to syntactic constraints C as”):

83

C1: C2: C3: C4:

C5:

αn ≡ αm =C a2n−1 ⎨ a2n ∧ a2m−1 ⎨ a2m ∧ a2n−1 ⎨ a2m, αn < αm =C a2n−1 ⎨ a2n ∧ a2m−1 ⎨ a2m ∧ a2n−1 ≺ a2m ∧ ¬a2n−1 ⎨ a2m, ¬FP =C ¬C(FP), where FP is a wff of SP translated according to C1–C5 into wff C(FP) of SI, FP′♥FP″ =C C(FP′)♥C(FP″), where ♥ stands for ⇒ or ∧ or ∨ or ⇔, and FP’ and FP” stand for two wffs of SP translated according to C1–C5 into two wffs of SI, C(FP′) and C(FP″) respectively, (αn)Ω(αn) =C(a2n−1)(a2n)((a2n−1 ⎨ a2n) ⇒ Ω*(a2n−1, a2n)) and (∃αn)Ω(αn) =C(∃a2n−1)(∃a2n)((a2n−1 ⎨ a2n) ∧ Ω*(a2n−1, a2n)), where Ω(αn) is a formula of SP translated into formula Ω*(a2n−1, a2n) of SI according to C1–C5 .

Let f* be a function f* : an ⎯→ 〈α2n−1, α2n〉 (n = 1, 2,…) mapping variables of SI into ordered pairs of variables of SP, and let C*1–C*5 be the following translation rules providing a 1-1 translation of all the wffs of SI into a subset of the wffs of SP (where =C* is to be understood analogously to =C): C*1: C*2: C*3: C*4:

C*5:

an = am =C* α2n−1 < α2n ∧ α2m−1 < α2m ∧ α2n−1 ≡ α2m−1 ∧ α2n ≡ α2m, an ≺ am =C* α2n−1 < α2n ∧ α2m−1 < α2m ∧ ¬α2m−1 < α2n, ¬FI =C* ¬C*(FI), where FI is a wff of SI translated according to C*1–C*5 into wff C(FI) of SP, FI’♥FI” =C* C*(FI’)♥C*(FI”), where ♥ stands for ⇒ or ∧ or ∨ or ⇔, and FI’ and FI” stand for two wffs of SI translated according to C*1–C*5 into two wffs of SP, C*(FI’) and C*(FI”) respectively, (an)Φ(an) =C*(α2n−1)(α2n)((α2n−1 < α2n) ⇒ Φ*(α2n−1, α2n)) and

84

(∃an)Φ(an) =C*(∃α2n−1)(∃α2n)((α2n−1 < α2n) ∧ Φ*(α2n−1, α2n)), where Φ(an) is a formula of SI translated into formula Φ*(α2n−1, α2n) of SP according to C*1–C*5. These translation rules (C1–C5 and C*1–C*5) constitute an effective mechanical procedure for translating any formula of either of the two systems, be it open or closed, into exactly one formula of the other system. However, since by translating closed formulas the condition occurring in the translation of quantifiers reoccurs necessarily either as a conjunct or as a part of the consequent of the translation (depending on whether the existential or the universal quantifier is involved in a given translation), we can always obtain an equivalent but shorter version of the resulting formula. Since it is obvious why it is so if the quantifier of the original formula is existential, let us take an example when it is universal. Let Φ(an) be (AI1), that is (an)¬(an ≺ an). According to C*1–C*5, the translation Φ*(α2n−1, α2n) of (AI1), to be denoted as (AI1)*, reads as follows: (α2n−1)(α2n)(α2n−1 < α2n ⇒ ¬(α2n−1 < α2n ∧ α2n−1 < α2n ∧ ¬α2n−1 < α2n)). However, this is equivalent to (α2n−1)(α2n)(α2n−1 < α2n ⇒ ¬¬α2n−1 < α2n), since α2n−1 < α2n ∧ α2n−1 < α2n occurring in the consequent can be erased, given that in propositional calculus p ⇒ ¬(p ∧ p ∧ q) is equivalent to p ⇒ ¬ q. Now, by using this device for obtaining the shorter form of a translation, we get (AP9)* as a translation of the axiom (AP9): (AP9)* (a1)(a2)…(ai)… (∧1≤i<ω a2i−1 ⎨ a2i ⇒ ((∃b1)(∃b2)( b1 ⎨ b2 ∧ ∧ (∧1≤i<ω ai ≺ b2)) ⇒ ⇒ (∃c1)(∃c2)(c1 ⎨ c2 ∧ (∧1≤i<ω ai ≺ c2) ∧ ∧¬(∃d1)¬(∃d2)(d1 ⎨ d2 ∧ ((∧1≤i<ω ai ≺ d2) ∧ d1 ≺ c2 ∧ ¬ d1 ⎨c2)))).

85

In the same way, we get (AI9)* as a translation of the axiom (AI9): (AI9)* (α1)(α2)…(αi)… (∧1≤i<ω α2i−1 < α2i ⇒ ((∃β1)(∃β2)( β1 < β2 ∧ ∧ (∧1≤i<ω ¬β1 < α2i)) ⇒ ⇒ (∃γ1)(∃γ2)(γ1 < γ2 ∧ (∧1≤i<ω ¬γ1 < α2i) ∧ ∧¬(∃δ1)¬(∃δ2)(δ1 < δ2 ∧ ((∧1≤i<ω ¬δ1< α2i) ∧ ¬γ1 < δ2)))). The main thing to do is to prove that both (AP9)* and (AP10)*, as well as (AI9)* and (AI10)*, are theorems of SI and SP, respectively. Proof for (AP9)* Let us assume both ∧1≤i<ω a2i−1 ⎨ a2i and (∃b1)(∃b2)( b1 ⎨ b2 ∧ (∧1≤i<ω ai ≺ b2)), which are the two antecedents of (AP9)*. Now, since for any i (1≤i<ω), ai ≺ b2, it follows directly from (AI9) that there is v such that ai ≺ v and, for no w, both ai ≺ w and w ≺ v. Let us now assume, contrary to the statement of the consequent of (AP9)*, that for any two c1, c2 such that c1 ⎨ c2 and for any i (1≤i<ω) ai ≺ c2, there are always d1 and d2 such that d1 ⎨ d2 and for any i (1≤i<ω) ai ≺ d2, so that d1 ≺ c2 and ¬ d1 ⎨ c2. But then, if we take c2 to be just v from the consequent of (AI9) (and c1 any interval such that c1 ⎨ c2), the assumption that for any i (1≤i<ω) ai ≺ c2 but d1 ≺ c2 and ¬ d1 ⎨ c2 contradicts the choice of c2, since if c2 = v, then, according to (AI9), for any d1 and d2 such that d1 ⎨ d2 and for any i (1≤i<ω) ai ≺ d2, it cannot be that d1 ≺ c2 and ¬ d1 ⎨ c2. (Q.E.D.) Proof for (AI9)* Let us assume both ∧1≤i<ω α2i−1 < α2i and

(∃β1)(∃β2)( β1 < β2 ∧ (∧1≤i<ω ¬β1 < α2i)),

which are the two antecedents of (AI9)*. Now, since for any i (1≤i<ω), ¬β1 < α2i implies ¬β1 < αi, it follows directly from (AP9) that there is γ such that ¬γ < αi and for no δ, both ¬δ < αi and δ < γ. Let us now assume, contrary to the statement of the consequent of (AI9)*, that for any two γ1, γ2 such that γ1 < γ2 and for any i (1≤i<ω) ¬γ1 < α2i,

86

there are always δ1 and δ2 such that δ1 < δ2 and for any i (1≤i<ω) ¬δ1 < α2i, so that ¬δ1 < α2i and ¬γ1 < δ2. But then, if we take γ1 to be just γ from the consequent of (AP9) (and γ2 any point such that γ1 < γ2), we get first ¬γ1 < δ2, and then δ1 < γ1 (since δ1 < δ2), which contradicts the choice of γ1, since if γ1 ≡ γ, there is no δ (and so also no δ1) such that both ¬δ < αi and δ < ¬γ1. (Q.E.D.) It can be proved analogously that (AP10)* and (AI10)* are also theorems of SI and SP, respectively. 3. Conclusion Because of the intuitive similarity between the point-based and the interval-based systems of the continuum, it is hard to believe, in spite of the ‘great struggle’ between Cantorians and Aristotelians, that nobody else had the idea that the two systems of the continuum are not so radically different that there wouldn’t be some sense in which one could say that they are merely trivially different. And yes, by speaking about instant-based and period-based time systems, van Benthem has proclaimed that ‘systematic connections between point structures and period structures enable one to use both perspectives at will’.15 But, van Benthem says nothing concrete about how these ‘systematic connections’ are to be formally defined and how it is to be proved that the point-based and the interval-based systems conform to such a definition. Now, in Section 2 it is established that after translating (AP9) into SI and (AI9) into SP the obtained formulas (AP9)* and (AI9)* are theorems of SI and SP respectively. The same holds for (AP10)* and (AI10)*. Together with Arsenijević’s analogous 2003 result concerning the first eight axioms of the two systems, this is sufficient for suggesting that SP and SI are syntactically only trivially different. However, more needs to be said about the alleged semantically trivial difference between them, for there is a seemingly striking discrepancy between the entities of their corresponding models. The basic elements of the intended model of SP are points, whereas intervals are continuous sets of points. The basic elements of the intended model of SI are stretches, whilst points are the abutment places of abutting stretches. But aren’t stretches (of SI) and intervals (of SP) hope15. Van Benthem 1991, p. 84.

87

lessly different entities, given that stretches are neither open nor closed nor half-open (half-closed), whereas intervals are necessarily either open or closed or half-open (half-closed)? The solution to this problem can be found in the way in which in (AI9) and (AI10) the suppositions for the existence of the least upper bound and the greatest lower bound are introduced. Namely, an infinite set of stretches having the least upper bound correspond to an interval open on the right side in the point-based structure, whereas, analogously, an infinite number of stretches having the greatest lower bound correspond to an interval open on the left side. Consequently, an infinite number of stretches having both the least upper and the greatest lower bound correspond to an open interval. And then, curiously enough, stretches themselves, which are originally neither closed nor open, turn out to correspond to closed intervals in the point-based structure. There is another big philosophical question that has to be answered. According to Quine’s famous slogan ‘To be assumed as an entity is to be reckoned as the value of a variable’16, the two formal theories are not trivially different if there is no model in which their variables range over the elements of one and the same basic set, and in the case of SP and SI their variables can never do this. But why should it be so important what variables do, if the set of entities — elementary and non-elementary — is the same in any intended model of the two theories? So, pace Quine, there are good reasons for saying that SP and SI are only trivially different. Syntactically, it is sufficient that there are two sets of translation rules that, though not inverses of one another, are theorem preserving. Semantically, it is sufficient that when speaking about points we cannot avoid automatically saying something unequivocal about stretches, and vice versa.17 Finally, let us mention a benign asymmetry between the two systems. When we speak of a least upper bound in a continuous point-based structure, it is always just a single point. However, the least upper bound in a continuous interval structure is not a single interval but an equivalence class of intervals. This trivial fact is a consequence of another trivial fact. Contrary to a stretch, which is always a single interval of an interval structure, the place of the abutment of any two intervals, which defines a point of a point structure, is always an abutment place of an infinite number of 16. Quine 1961, p. 13. 17. See Arsenijević 2003, pp. 10–11.

88

intervals. But we can always choose one pair from the equivalence class of abutting intervals to represent a given point.18 REFERENCES Aristotle, in: Bekker, I. 1831: Aristotelis opera, edidit Academia Regia Borussica, vol. i , Berolini. Arsenijević, M. 1992: “Logik der Punkte und Logik der Intervalle”, Philosophia naturalis 29/2, pp. 160–179. — 2003: “Generalized concepts of syntactically and semantically trivial differences and instant-based and period-based time ontologies”, Journal of Applied Logic 1, pp. 1–12. Benthem, J. van 1991: The Logic of Time, Kluwer. — 1995: “Temporal logic”, in: Handbook of Logic in Artificial Intelligence and Logic Programming, Vol. 4, Clarendon Press, Oxford, pp. 241–350, D. M. Gabbay, C. J. Hogger, J. A. Robinson (Eds.). Bochman, A. 1990a: “Concerted instant-interval temporal semantics I: Temporal ontologies”, Notre Dame Journal of Formal Logic 31, pp. 403–414. — 1990b: “Concerted instant-interval temporal semantics II: Temporal valuations and logics of change”, Notre Dame Journal of Formal Logic 31, pp. 580–601. Burgess, J. P. 1982: “Axioms for tense logic II: Time periods”, Notre Dame Journal of Formal Logic 23, pp. 375–383. Cantor, G. 1962: Gesammelte Abhandlungen, Hildesheim. Carnap, R. 1928: Der logische Aufbau der Welt, Felix Meiner. Comer, S. C. 1985: “The elementary theory of interval real numbers”, Zeitschrift für Mathematische Logik und Grundlagen der Mathematik 31, pp. 89–95. Earman, J. and Roberts, J. 2006: “Contact with the nomic: a challenge for deniers of Humean supervenience about laws of nature”, Philosophy and Phenomenological Research (forthcoming). Ehrlich, P. 2005: “The rise of non-Archimedean mathematics and the roots of a misconception 1: The emergence of non-Archimedean Systems of magnitudes”, Archive for History of Exact Sciences (forthcoming). Foldes, S. 1980: “On intervals in relational structures”, Zeitschrift für Mathematische Logik und Grundlagen der Mathematik 26, pp. 97–101. Grünbaum, A. 1952: “A consistent conception of the extended linear continuum as aggregate of unextended elements”, Philosophy of Science 4, (pp. 288-306). 18. Acknowledgements: We are grateful to Allen Janis for helpful discussion, and to Nuel Belnap and Gerald Massey for discussions and comments on an earlier version of this paper.

89

Grünbaum, A. 1974: Philosophical Problems of Space and Time, Reidel. Hamblin, C. L. 1969: “Starting and stopping”, The Monist 53, pp. 410–425. — 1971: ‘Instants and intervals’, Studium generale 24, pp. 127–134. Humberstone, I. L. 1979: “Interval semantics for tense logic: some remarks”, Journal of Philosophical Logic 8, pp. 171–196. Leibniz, G. W. 1849–63: Gesammelte Werke, G. H. Pertz (Ed.), Halle. Lewis, D. 1994: “Humean supervenience debugged”, Mind 103, pp. 473–490. Needham, P. 1981: “Temporal intervals and temporal order’, Logique et Analyse 24, pp. 49–64. Quine, W. O. 1961: From a Logical Point of View, Harvard University Press. Robinson, A 1974: Non-Standard Analysis, Revised Edition, North-Holland Publishing Company, Amsterdam. Robinson, D. 1989: “Matter, motion and Humean supervenience”, Australasian Journal of Philosophy 67, pp. 394–409. Roeper, P. 1997: “Region-based topology”, Journal of Philosophical Logic 26, pp. 251–309. — 2006: “The Aristotelian continuum. A formal characterization”, Notre Dame Journal of Formal Logic 47, pp. 211–232. Russell, B. 1903: The Principles of Mathematics, George Allen & Unwin. — 1914: Our Knowledge of the External World, George Allen & Unwin. Salmon, W. C. 1975: Space, Time, and Motion, Dickenson. Venema, Y. 1990: “Expressiveness and completeness of an interval tense logic”, Notre Dame Journal of Formal Logic 31, pp. 529–547. White, M. J. 1988: “An ‘almost classical’ period-based tense logic”, Notre Dame Journal of Formal Logic 29, pp. 438–453.

90

Grazer Philosophische Studien 76 (2008), 91–107.

MATHEMATICAL KNOWLEDGE IS CONTEXT DEPENDENT Benedikt LÖWE Universiteit van Amsterdam, Universität Hamburg & Rheinische Friedrich-Wilhelms-Universität Bonn Thomas MÜLLER Rheinische Friedrich-Wilhelms-Universität Bonn Summary We argue that mathematical knowledge is context dependent. Our main argument is that on pain of distorting mathematical practice, one must analyse the notion of having available a proof, which supplies justification in mathematics, in a context dependent way.

‘But a proof is sometimes a fuzzy concept, subject to whim and personality.’ Kenneth Chang, New York Times (April 6, 2004)

1. Introduction Mathematical knowledge appears to be of a special, privileged form. When somebody knows a mathematical fact, we say that she knows ‘with mathematical certainty’, and it is commonly assumed that nothing can be more firmly grounded than that. Not surprisingly, in philosophical contexts, mathematics is often used as an epistemological role model. Mathematical knowledge is assumed to be absolute and undeniably firm. The main reason for that special status lies in the fact that mathematicians prove their theorems: Mathematical knowledge is proven knowledge (‘more geometrico demonstrata’). What has been proven is established beyond all doubt. Thus, mathematical knowledge stands out as knowledge with a uniform witness, the notion of mathematical (deductive) proof. This close connection between mathematical knowledge and the priviledged form of epistemic justification via mathematical proof leads to a broad consensus of how to analyse mathematical knowledge. The standard

view of mathematicians and philosophers alike (which is in agreement with the common perception of the educated public) can be described the following way: (*) S knows that P iff S has available a proof of P. Of course, (*) is vague with respect to the two key notions of the explanans (“proof ” and “having available”). We shall discuss the notions of proof in detail in § 2. Assuming for a moment that we agree on what “proof ” is, what does it mean to have available a proof? A literal reading in terms of having access to a material copy of the proof is inappropriate. It is too narrow, because there just aren’t enough copies of proofs to back even a fraction of true mathematical knowledge claims (especially if one demands derivations, of which there are hardly any around).1 But it is also too wide: A mathematical illiterate on the first floor of UC Berkeley’s Evans Hall (the math library) has available lots and lots of proofs, but it would be odd to say that the mere location could affect any change in mathematical knowledge (genius loci nonwithstanding). Thus, “having available” cannot be spelled out in terms of actual physical access; it needs to be given a modalised reading in which the epistemic subject S plays an active role. A reformulation of (*) that makes that modalisation explicit is the following: (†)

S knows that P iff S could in principle generate a proof of P.

Of course, (†) continues to have vague terms, viz. “could in principle”, “generate”, and “proof ”. As mentioned before, we shall discuss “proof ” in § 2; the notions of “could in principle” and “generate” will be discussed in § 3. Following the tradition of standard (context independent) epistemology, many philosophers would like to interpret (†) by giving necessary and sufficient conditions for the right-hand side to hold, independent of the context. The general perception of the absolute nature of mathemati1. E.g., no living mathematician has seen a derivation of the Feit-Thompson Theorem, yet there are (many) mathematicians who know that every group of odd order is solvable. The original paper, Feit and Thompson (1963), has over 250 pages. Only specialists in finite group theory will know even an informal proof. On the other hand, the theorem is rather well known.

92

cal knowledge makes such a project appear more promising than in other fields of knowledge. In such a context independent or invariantist reading, the vague notions “could in principle”, “generate”, and “proof ” would be replaced by distinguished sharp notions “could in principle♯”, “generate♯” and “proof ♯”, leading to (†♯)

S knows that P iff S could in principle♯ generate♯ a proof ♯ of P.

In this paper, we shall argue that no reading of (†♯) is adequate as an analysis of mathematical knowledge. In our argument, we shall proceed from a mildly naturalistic philosophical methodology: In philosophising about mathematics, mathematical practice must be taken seriously. If certain expressions, such as “knowledge”, or “… knows that …”, are used in the mathematical community, then that usage cannot be dismissed without good arguments. This is not to say that mathematical practice has the last word—but it certainly has to have the first word. Thus, we will not be satisfied with an epistemology for mathematics according to which there is no (or hardly any) mathematical knowledge in the world—mathematical practice asserts that on the contrary, there is a lot of mathematical knowledge. On the other hand, we will also not be willing to accept an epistemology that identifies all true mathematical statements as the necessary proposition 2 + 2 = 4 in disguise. Such a position consequently grants that every epistemic subject knows all mathematical truths (but may not be aware of them). There is certainly less mathematical knowledge than that! The paper is structured as follows: In § 2, we start by briefly discussing the status of mathematics as an epistemic exception and the nature of mathematical proof. In § 3, we then move on to consider possible interpretations of “could in principle” and “generate” in connection with various notions of proof. This section contains our argument against (†♯). Furthermore, not even the weakest notion of “proof ” is necessary for mathematical knowledge: in § 4, we discuss inductive reasoning and knowledge by testimony in mathematics. Having debunked (†♯), we propose an alternative. In § 5, we briefly describe Lewis’s contextualist analysis of knowledge and give it a mathematical reading (#′). In § 6, we tie (#′) to the Dreyfus-Dreyfus skill model to arrive at our final analysis (‡).

93

2. Standard mathematical epistemology and the notion(s) of proof Mathematics is an epistemic exception2 as compared to the other sciences. This point has been implicitly or explicitly observed by a large number of philosophers ancient and modern. Plato in the famous paij example (Meno, 82b–84a) shows how the slave, guided by Socrates, without any prior education or empirical data arrives at a mathematical truth. Kant, who holds that mathematical truths are synthetic a priori, limits the use of “knowledge” generally to proven certainties and even claims that mere belief has no place in mathematics at all (Kritik der reinen Vernunft, A823/B851). Frege conceives of mathematics as a branch of logic. His project of logicism makes a purely formal method the hallmark of secure knowledge (Begriffsschrift, (Frege, 1879, IXf.)). And to bring in a contemporary contextualist, Lewis remarks in passing that in ‘the mathematics department, […] they are in confident agreement about what’s true and how to tell, and they disagree only about what’s fruitful and interesting’ (Lewis, 2000, 187f.). The distinguishing feature of mathematical epistemology that underlies the observed exceptional status is the robust notion of mathematical proof: in mathematics, there is deduction, in the sciences there is only induction. Moreover, as we learn early on in our education, a mathematical proof is either correct or incorrect and does not admit degrees of correctness. To use Keith Devlin’s polemic words: ‘Surely, any math teacher can tell in ten minutes whether a solution to a math problem is right or wrong! […] Come on folks, it’s a simple enough question. Is his math right or wrong?’ (Devlin, 2003)3 It is an empirical fact that there seem to be no lasting disagreements in mathematics. Whether someone has available a proof of P is almost never a serious matter of dispute; and thus it is natural to assume that the vague terms “has available” and “proof ” in (*) can be made precise without contest. This lends intuitive support to a context independent reading à la (†♯). 2. This has been an important topic in the sociology of science, discussed, e.g., by Mannheim, Bloor (1976) and Livingston (1986). Cf. also Heintz (2000, Chap. 1) and Prediger (2001, 24f.). 3. Just for the record: Of course, Devlin is playing the advocatus here, arguing that even checking proofs is not as trivial as is often believed. An observation along these lines can already be found in Locke’s Essay (Locke, 1689, IV.ii.7): ‘In long deductions […] the memory does not always so readily and exactly retain; therefore it comes to pass, that this is more imperfect than intuitive knowledge, and men embrace often falsehood for demonstrations.’

94

Let us now discuss the notion of “proof ” that is so seemingly unequivocal. It is undeniably robust, but is this robustness realised in the same form everywhere? In actual mathematical practice, a wide range of texts or activities is called “proof ”. The guiding idea of proving something is to arrive at the result through a number of secure steps, but one needs to specify which steps may be used. Frege in his Begriffsschrift proposed that the steps should be so small that a mechanical procedure was available for checking each step. This led to a mathematically precise definition of formal proof which was then available for metamathematical investigations leading, e.g., to Gödel’s completeness and incompleteness results. We will use the term derivation to stand for formal proof in a mathematically well-defined system.4 Outside meta-mathematical investigations and a few very specialised areas,5 one will not find derivations in mathematical publications. Mathematical journals and textbooks (as well as lectures, research notes and conference talks) instead contain informal proofs.6 The notion of informal proof does not have a mathematically precise definition—if it did, it would be just another version of derivation. From the point of view of derivation, informal proofs contain gaps and appear to be unfinished. It is therefore tempting to see an informal proof just as an imperfect stand-in for a derivation. However, mathematical practice strongly supports the view that the important notion of proof in mathematics is not derivation, but informal proof. One reason for this is communication: ‘The point of publishing a proof […] is to communicate that proof to other mathematicians. […] [T]he most efficient way […] is not by laying out the entire sequence of propositions in excruciating detail’ (Fallis, 2003, 55). Instead, mathematicians publish informal proofs. However, there is more to informal proof than ease of communication. It just isn’t the case that mathematicians have a derivation in mind and transform it into an informal proof for publication in order to reach a wider public—the entire procedure of doing research mathematics rests on doing informal proofs. The proofs in mathematical research papers are so far removed from derivations that only a few experts could produce a derivation from them even if they wanted to, 4. There are various competing notions of derivation, but their differences do not matter for our purposes. For first-order logic, the competing formal systems are equivalent in allowing one to prove exactly the same theorems. 5. E.g., the Journal of Formalized Mathematics, which focuses on derivations in the specific proof system MIZAR, cf. http://www.mizar.org/JFM/; or the publications of the Coq group discussed in § 3. 6. Cf., e.g., Rav (1999) for discussion of the distinction between derivations and informal proofs.

95

and only a minority considers that a worthwhile enterprise. We need to take seriously the fact that derivations are hardly ever used. Subscribing to the tempting image of the derivations as the real objects of mathematical study to which informal proofs are imperfect approximations would be a violation of our maxim of taking mathematical practice seriously.7 Informal proofs come in many flavours. One can, e.g., distinguish semiformal textbook proofs for beginning students, graduate-level textbook proofs, journal proofs, informal research notes, and proof sketches. Each of these types is pragmatically fairly well delineated—try submitting a textbook-style proof to a mathematical journal, or presenting research note-style proofs to beginning students, and you will feel the force of the boundaries. It is often possible to compare proofs for one and the same P with respect to the level of detail they exhibit. One proof may give more details than 7. Derivation is often called the gold standard of mathematical proof. That metaphor is quite telling. First, a few historical facts. Implementing a gold standard means making a fixed weight of gold the standard economic unit of account. This can, e.g., be established by using coins made of gold. More practically, gold is stored in some central reservoir, and paper money is issued as certificates entitling the holder to a fixed amount of gold. Such systems were established in the late 19th century in many Western countries, and there were earlier, similar systems in many places. A positive aspect of an international gold standard is free convertibility of currencies, which was important in boosting international trade. A negative aspect of such a system is that even though gold is nice stuff, what people actually need isn’t gold (except in some cases related to dentistry), but other goods, and the scarcity or otherwise of gold dictates in effect the price of other goods. The successor of the early international gold standard, the Bretton Woods system, collapsed in the early 1970ies. Since then, many countries have sold off much of their gold. Other mechanisms of establishing trust between trading parties have proved to be more practical and more efficient. We would like to draw a rather strict analogy between the rôle of gold for the exchange of goods and the rôle of derivation for the exchange of mathematical knowledge. Historically, of course there never was a period in the development of mathematics during which derivation was the generally accepted currency, but the logicist movement of the early 20th century surely was an attempt at establishing that currency. Just like gold vs. goods, derivation is neither the only store of value for mathematics, nor the most useful. If anything, trading in derivations is more impractical than trading in gold. (Given the scarcity of gold and the expanded international trade today, a return to an international gold standard would mean increasing the current price of gold more than tenfold. But given the scarcity of derivations, establishing derivations as the sole vehicle of mathematical justification would at present completely stop the development of mathematics.) Other mechanisms for establishing trust in the mathematical community are well established, and they are working. Of course that does not mean that derivations are worthless. Quite on the contrary—the belief in the possibility to generate, at least in principle, a derivation corresponding to any given informal proof, may well be one of the strongest sources of mutual trust in the mathematics community. It is just that actual derivations aren’t really needed—except, if you allow, for exercises in mathematical dentistry.

96

another, even though both are valid and complete proofs of certain types. E.g., a textbook proof may contain a whole page of details for a certain inductive construction where a research note would just say “by induction”. Thus it makes sense to say of a proof that it contains gaps relative to another. However, we do not subscribe to an absolute notion of “gaps in proofs”, because that would presuppose an absolute standard of a “gap-less” proof.8

3. Having available a proof In the previous section, we discussed possible readings of the vague term “proof ”, in this section, we shall now focus on the terms “could in principle” and “generate” used in the modalised analysis (†). A classical example of modalisation for mathematical knowledge is Brouwer’s idealised mathematician, the creative subject who creates his choice sequences.9 Kitcher (1984, Chap. 6.III), in a similar vein, employs the notion of an ‘ideal agent’ to account for the fact that actual operations of actual agents do not suffice to establish the truths of arithmetic as he conceives it.10 Steiner (1975, Chap. 3) explicates the modal idealisation of (†) via the following thought experiment: In order to check whether a mathematician has available a proof of P and thus, knows that P, she is asked to transform her (informal) proof into a derivation with the aid of a logician who as a Socratean ‘midwife’ works out the formal details, but is not otherwise mathematically creative. “If the two can bang out a formal proof, then the mathematician is said to have known the proof all along, on the basis of the informal argument” (Steiner, 1975, 100). Thus: (†♯1)

S knows that P iff with the help of a logician, S can generate a derivation of P.

Brouwer, Kitcher, and Steiner give quite specific readings of modal aspects of mathematics, and Steiner gives an explicit test for ‘could in principle generate a proof. This is what one needs to do if one is after an 8. Note that there is a different notion of gap in proof, which Fallis (2003) calls ‘untraversed gaps’ in contrast to the ‘enthymematic gaps’ that we just discussed: If in proving one fails to note a certain special case, the proof will be incomplete—it won’t even belong to the intended class of informal proof. Here the gap terminology is appropriate in an absolute sense. 9. Cf. Brouwer (1929); for an historical overview of the notion, cf. Troelstra (1982). 10. Cf. Chihara (1990, Chap. 11.2) for decisive criticism of Kitcher’s approach.

97

invariantist version of (†). However, mathematical practice provides counterexamples against any fixed notion—there is even knowledge without proof (cf. § 4 below). We will now explore the modal dimension of (†) in three steps, starting with a critique of Steiner’s approach. (i) Steiner’s model (†♯1 ) is open to a number of criticisms, some already voiced in the original publication.11 The envisaged test for knowledge only replaces one form of modalisation (‘has available’ or ‘could in principle generate’) with another, not much clearer one (‘can produce, with the aid of a logician, …’)—and the kind of logician that is needed may well turn out not to exist. The logician’s powers play a crucial role. Steiner rightly stresses that “we cannot envision a superhuman, because such a being would discover a completed proof despite the ignorance of the mathematician” (Steiner, 1975, 101f.), rendering the test useless. In practice, even if two persons cooperate in producing a derivation, the rôles will never be as clearly delineated as the test suggests. It may be fine to say that the pair who succeeded in writing down a derivation had available a proof (and thus, knew that P), but that is of course no good as a test of the mathematician’s knowledge. Let us now consider two variants of Steiner’s modalisation. In both variants, the dubious logician is replaced by a direct appeal to the subject’s capabilities. The first variant is based on derivation, the second, on informal blackboard proof. (ii) Suppose that we want to read (†♯) by fixing “proof ♯” to mean “derivation”. The task then is to try to find a good explication of ‘could in principle generate’. The successes of formalised mathematics have shown that it is possible to provide derivations for many important mathematical statements, however doing so requires a long time: e.g., the Coq community worked for over ten years before Geuvers, Wiedijk, and Zwanenburg were able to formalise the fundamental theorem of algebra (Geuvers et al., 2001). Now, this suggests reading ‘could in principle generate’ as follows: (†♯2 )

S knows that P iff, given ten years, she could write a formal derivation in the language Coq. ♯

11. It must be said in fairness to Steiner that he does not subscribe to († ) in the end. Rather, he gives an example of mathematical knowledge without proof and then argues for a Platonist understanding of mathematical knowledge. 1

98

But compared to these ten years, the time we need to learn mathematical facts is short: many mathematicians could be in the situation that they don’t know anything about P, but are able to learn within ten years both the mathematics needed to understand why P is true and then formalise it in Coq. These mathematicians would satisfy our fixed reading of (†♯2), but by assumption do not know P. For particularly bright beginning students, the time of ten years might be enough to study mathematics, enter graduate school, finish a doctoral degree in mathematics, and learn Coq. Thus, the reading (†♯2)would grant almost indefinite mathematical knowledge to everyone who has the intellectual capacity to finish a mathematics degree. Clearly, not an intended reading. The invariantist readings (†♯1 ) and (†♯2) face another difficulty. As soon as derivations or a system like Coq play a role, we need to concede that there was no mathematical knowledge prior to a certain point in time: e.g., before the Begriffsschrift, nobody could give a derivation of anything, because the concept of derivation had not yet been invented.12 But mathematics is commonly taken to be the prime example of historically stable knowledge—the ancient Greeks already knew the Pythagorean theorem. (iii) At the other end of the spectrum let us read (†♯) by fixing “proof ♯” to mean “informal proof on the blackboard”. For many research situations in mathematics, the relevant notion of ‘could in principle generate’ is something like the following: S knows that P iff, given a blackboard and (†♯3 ) a piece of chalk, she is able to produce an acceptable blackboard proof within an hour. Note that we cannot restrict the timeframe for producing the proof to the time physically needed to produce the chalk markings on the blackboard, as many research mathematicians do not have all of the proofs they need for their work at their immediate cognitive disposal. They need to try one or two standard approaches to tackle the problem, remember the important details, and only after that are they able to provide an acceptable proof. If one makes this time frame too short, then one arrives at too strict a crite12. If you are not satisfied with taking the Begriffsschrift as the beginning of derivation, supply your favourite reading instead. The consequences are practically the same.

99

rion for knowledge. The analysis (†♯3) is an excellent description of mathematical knowledge among researchers meeting in an office for joint work, but is inadequate for other situations. Consider a student in an oral exam asked whether P or non-P is true. Suppose that the student erroneously believes that non-P is true but given a blackboard, a piece of chalk and one hour of time, this particular student might be able to create a blackboard proof of P, first trying to prove non-P, failing, getting some ideas from the failed attempts, then remembering some facts and ideas from lectures, and finally proving P. However, in the oral exam, the examiner will not wait for an hour, the student has to rely on his belief, says ‘non-P’ and fails. Does this mean that the oral exam is not testing knowledge? In view of our methodological maxim, that would be absurd. The analyses (†♯1 ), (†♯2 ), and (†♯3 ) are just three possible examples for a reading of (†♯). For other readings, it is easy to come up with more examples or contexts of knowledge attributions that show that they are problematic. We would like to add two relevant remarks: First of all, our examples show that the temporal component in “could in principle” is immensely important, and that it seems hopeless to try to fix a single reading for all contexts. If one gives the subjects too much time to generate a proof, then one ends up with knowledge assertions that shouldn’t be true, but if one gives them too little time, then some true knowledge assertions dissolve. Secondly, one way to avert the move to full contextualism would be to allow the meanings of “could in principle”, “generate”, and “proof ” to depend on S, but not on the general context. This would give an analysis (†S)

S knows that P iff S could in principleS generateS a proof S of P,

where “could in principleS”, “generateS”, and “proof S” are assignments of meanings of the vague terms to S. For instance, “proof S” could be a formal derivation if S is a member of the Coq programming project, a blackboard proof if S is a research mathematician, and a textbook proof if S is a student. This obviously won’t work either: Coq programmers are typically reseach mathematicians as well and may need to switch between contexts; students face different situations, e.g., our student in the exam mentioned above will be assessed differently than in a tutorial session. 100

One step further, one could make the meanings of the vague terms dependent on S and P, leading to an analysis (†SP). The same argument shows that this cannot deal with the multitude of different contexts either. 4. Mathematical knowledge without proof We have seen that proof comes in many flavours. In this section, we shall discuss examples of proper knowledge attributions in mathematics without cognitive access to any form of mathematical proof.13 A good historical example is reported in Pólya’s study of Euler (Pólya, 1954, Chap. 2.6): It is certainly true to say that Euler knew that 1 + 1/4 + 1/9 + 1/16 + 1/25 + … = π/, but Euler didn’t have available (and knew that he didn’t have available, nor could in principle generate) a proof of that fact—he had established it via generally shaky generalisations from finite to infinite sums, and his evidence was to a large part inductive (i.e., the first 20 or so decimal places coincided). Still, it would be ahistorical to say that Euler had just guessed.14 Cases of knowledge without proof are not rare at all, nor are they a thing of the past. Even more important than inductive generalisations is knowledge via testimony for which proof plays hardly any rôle at all— and yet, in many mathematical contexts it is fine to base a knowledge claim on testimony. That is obvious enough for claims to mathematical knowledge in the general public: Most people haven’t actually seen any mathematical proofs at all, and yet the majority of the public has mathematical knowledge of some kind, e.g., elementary algorithms of arithmetic, the Rule of Three, etc. For beginning math students, a similar observation holds: While we certainly urge them to try to learn and understand the proofs, we also concede that the students do acquire knowledge (though not a very deep kind of knowledge) by just learning theorems by heart, and that may be enough to pass a first exam. And even in the context of research mathematics, some 13. Knowledge without proof points to a vexing question in the philosophy of mathematics: Is it possible to have (a high degree of ) knowledge of P by pure intuition without any formal proof in mind (the Ramanujan phenomenon)? Cf. Thurston (1994) for discussion of this point. 14. Cf. Steiner (1975, Chap. 3.IV) for a similar assessment.

101

knowledge is just based on trust. If one works in cooperation with others, it will not normally be possible, nor required, to learn and check all proofs. It could be said that these examples are so vastly different from those given in § 3 that they constitute a violation of (*) or (†). In § 5, we shall develop a context dependent reading of (†); given that the meanings of the vague terms “could in principle”, “generate”, and “proof ” will vary according to context anyway, it will allows us to understand examples of inductive knowledge and knowledge via testimony as readings of (†), e.g., by relaxing the notion “proof ” (for the Euler-Ramanujam example) or “could in principle” (for cases of knowledge by testimony). 5. Contextualism in mathematics Contextualism is a fairly recent attempt at answering one of the long-standing problems of epistemology, viz., the problem of skepticism. In spelling out contextualism, we follow David Lewis’s general analysis, given in his 1996 classic, ‘Elusive knowledge’. Lewis analyses the statement ‘S knows that P’ context dependently as follows: (#)

S knows that P iff S’s evidence eliminates every possibility in which not-P—Psst!—except for those possibilities that we are properly ignoring (Lewis, 1996, 554).

The option of ‘properly ignoring possibilities’ allows for a spectrum of knowledge contexts from the loose standards of every-day usage (in which, e.g., I know that my cat Possum is not in the study without checking the closed drawers; cf. Lewis (1996, 562)) to the demanding standards of epistemology (Cartesian Doubt), in which (almost) all knowledge claims are defeated. Consequently, a switch of context may destroy knowledge. According to Lewis, this both explains the force of skeptical arguments and points a way to a cure. Lewis’s paper and a number of other related works have given rise to a huge debate about details and technicalities of his version of con-textualism, dealing with important questions about the specification of ‘properly ignoring possibilities’ and the context changes in communicative acts. This

102

paper is not intended to be a direct contribution to that debate.15 Contextualism has not been employed in the epistemology of mathematics so far. There is certainly a number of reasons why this is so. For Lewis, the main reason seems to be that he treats all true mathematical statements as the necessary proposition in disguise, thus blocking any way of distinguishing among them epistemologically. This is a consequence of Lewis’s modal epistemology: A semantics for knowledge claims for Lewis must be based on possible worlds. As all mathematical statements are true in all possible worlds, modal semantics must treat all mathematical statements as the necessary proposition, modelled as the set of all worlds. As we pointed out in the introduction, given our methodology, we cannot follow Lewis here.16 It seems obvious to us that Lewis’s modal approach to epistemology can be separated from his contextualist stance, and thus we will employ a contextualist analysis of knowledge along Lewisian lines. Our discussion in § 3 has revealed that the vague terms “could in principle”, “generate”, and “proof ” in (†) need to be interpreted depending on the context. No fixed notion (†♯) of “could in principle generate a proof ” yields an adequate analysis for all cases of mathematical knowledge. Thus, contextualism wins the day. But how? Our task now is to link the general contextualist analysis of knowledge (#) to the specific case of mathematics where S’s evidence and the ignored possibilities must be linked to the proof or other justification that is required according to (†). As we saw, a context generally specifies a type of proof (or other justification) as appropriate. Very few contexts in mathematics demand derivations. Blackboard proofs are typical of research mathematics, and mathematical knowledge claims in the general public typically do not need to be backed by any form of proof at all. Similarly, S’s evidence may be interpreted as the dispositional state of mind of S with respect to the required form of proof of P. Above we gave one explication by linking that disposition to a time frame and other resources that would be required to generate a written version of the proof in question. Thus one way of writing out (#) for the case of mathematics is the following: S knows that P iff S’s dispositional state of mind allows her (#′) to produce the required form of proof or justification for P with the resources allowed by the context. 15. Important recent contributions include MacFarlane (2005), Schaffer (2004), and DeRose (2002). 16. Incidentally, in Lewis (1993, 218), he supports something very close to our methodology, so there may be a slight tension in his position.

103

This analysis may be all that is needed, but it also comes with a certain problem: There does not seem to be an independent standard from which to assess the allowed resources. Thus, (#′) might be accused of being empirically void. We suggest in § 6 that the notion of mathematical skill can help to improve the analysis. 6. Mathematical knowledge and mathematical skills The notion of mathematical skill links the “dispositional state of mind” of (#′) with actual performance: Skill is both a modal notion (what somebody is able to do even while not doing it) and has an empirical side (skills can be tested). Our motivation for bringing skills into the picture is that through the Dreyfus-Dreyfus model of skill acquisition there is available a semi-formal theory of skill levels that has been fruitfully applied, e.g., to chess skills and nursing skills (Benner, 1984). In the Dreyfus-Dreyfus model (Dreyfus and Dreyfus, 1986), there are five levels of skill ranging from novice to expert. These levels are delineated by their different relation to explicitly formulated rules. While a novice needs to stick to explicit rules in a step-by-step fashion, experts have internalised and transgressed such rules and are able to proceed intuitively. Certainly the link between mathematical knowledge and mathematical skills merits further investigation, which will need to be left for a future occasion. Here we merely wish to argue that a skill-based analysis is plausible. Using the notion of skill, we can reformulate (#′), our preliminary synthesis of contextualism (#) and mathematical knowledge (†), as follows: (‡)

S knows that P iff S’s current mathematical skills are sufficient to produce the form of proof or justification for P required by the actual context.

This analysis, we claim, is adequate as a general explication of mathematical knowledge. It refers to the actual context and is thus flexible with respect to both crucial aspects of mathematical knowledge: Context determines the required form of proof or other justification, and context also sets the standard for the modal component in terms of a required skill level. Skill levels provide the link of our analysis with independent constraints that

104

was lacking in the case of (#′)—unlike counterfactual time constraints, skill levels can be (and, more importantly: are) characterised independently of the conceptual analysis for mathematical knowledge given in this paper. Mathematical practice affirms that the concept of mathematical skill is well entrenched. It is customary to comment on students’ or researchers’ skills, and it is often possible to rank people with respect to their skills.17 Skills are tested in exams and job talks, and it may well be that the aim of mathematics education is best characterised not as instilling mathematical knowledge, but as teaching mathematical skills. 7. Conclusion In this paper we argued that contrary to first appearances, mathematical knowledge is not a fixed, context independent notion. Rather, we showed by appeal to mathematical practice that unless one disregards actual practice—which in our view would be just plain bad methodology—, one is forced to admit that mathematical knowledge is context dependent. Many accounts of mathematical knowledge refer to the need to have available a proof. We concede that proof plays a crucial role in mathematics and in mathematical knowledge, but there is also mathematical knowledge without proof. Nor is proof a fixed notion: There are various forms of proof, and context determines which type of proof, if proof at all, is required. Furthermore, availability of proof is a modal notion that we suggested is best explained by reference to mathematical skills. What then of formal derivation? The concept of derivation and its universal acceptance as a formalization of the intuitive notion of proof is important for the foundations of mathematics, but contrary to folklore, it hardly plays any rôle in determining the truth of “S knows that P”—Psst!—unless the context explicitly demands it.

17. An interesting question which again merits further investigation is the following: How finely do we need to individuate mathematical skills? Will it be enough to ascribe to persons a single “mathematical skill level”, or will we need to be more topic-specific, speaking, e.g., of algebraic vs. geometrical skills?

105

REFERENCES Benner, Patricia (1984), From Novice to Expert: Excellence and Power in Clinical Nursing Practice. Addison-Wesley. Brouwer, Luitzen E. J. (1929), “Mathematik, Wissenschaft und Sprache”. Monatshefte fur Mathematik und Physik 36, 153–164. Bloor, David (1976), Knowledge and Social Imagery. University of Chicago Press. Chihara, Charles (1990), Constructibility and Mathematical Existence, Oxford University Press. DeRose, Keith (2002), “Assertion, Knowledge and Context”, Philosophical Review 111, 167–203. Devlin, Keith (2003), “The shame of it, MAA Online: Devlin’s Angle”, May 2003. URL: http://www.maa.org/devlin/devlin_05_03.html. Dreyfus, Hubert L. and Dreyfus, Stuart E. (1986), Mind over Machine: The Power of Human Intuition and Expertise in the Era of the Computer, Free Press. Fallis, Don (2003), “Intentional gaps in mathematical proofs”, Synthese 134, 45–69. Feit, Walter and Thompson, John G. (1963), “Solvability of groups of odd order”. Pacific Journal of Mathematics 13, 775–1029. Frege, Gottlob (1879), Begriffsschrift. Halle. Geuvers, Herman, Wiedijk, Freek and Zwanenburg, Jan (2001), “A Constructive Proof of the Fundamental Theorem of Algebra without using the Rationals”. In: Paul Callaghan, Zhaohui Luo, James McKinna and Robert Pollack (eds.), Types for Proofs and Programs, Proceedings of the International Workshop, TYPES 2000, Durham, Springer LNCS 2277, 96–111. Heintz, Bettina (2000), Die Innenwelt der Mathematik. Zur Kultur und Praxis einer beweisenden Disziplin. Wien: Springer. Kitcher, Philip (1984), The Nature of Mathematical Knowledge. New York. Lewis, David (1993), “Mathematics is megethology”. Philosophia Mathematica 3, 3–23. Reprinted in his Papers in Philosophical Logic, Cambridge UP 1998, 203–229. Page references are to the reprint. — (1996), “Elusive knowledge”. Australasian Journal of Philosophy 74, 549–567. Reprinted in his Papers in Metaphysics and Epistemology, Cambridge UP 1999. Page references are to the original. — (2000), “Academic appointments: Why ignore the advantage of being right?” In his Papers in Ethics and Social Philosophy, Cambridge University Press, 187–200. Livingston, Eric (1986), The Ethnomathematical Foundations of Mathematics. London: Routledge. Locke, John (1689), An Essay Concerning Human Understanding. Ed. by P. H. Nidditch. Oxford University Press 1975.

106

MacFarlane, John (2005), “The assessement sensitivity of knowledge attributions”. In: T.S. Gendler and J. Hawthorne (eds.), Oxford Studies in Epistemology, Oxford. Prediger, Susanne (2001), “Mathematik als kulturelles Produkt menschlicher Denktätigkeit und ihr Bezug zum Individuum”. In K. Lengnink, S. Prediger, F. Siebel (eds.): Mathematik und Mensch. Sichtweisen der Allgemeinen Mathematik, Darmstädter Schriften zur Allgemeinen Wissenschaft 2, Verlag Allgemeine Wissenschaft, Mühltal 2001, 21–36. Pólya, George (1954), Induction and Analogy in Mathematics, Princeton University Press. Rav, Yehuda (1999), “Why do we prove theorems?” Philosophia Mathematica (3) 7, 5–41. Schaffer, Jonathan (2004), “Skepticism, contextualism, and discrimination”. Philosophy and Phenomenological Research 69, 138–155. Steiner, Mark (1975), Mathematical Knowledge, Cornell University Press. Troelstra, Anne S. (1982), “On the origin and development of Brouwer’s concept of choice sequence”. In: A. Troelstra and D. van Dalen (eds.), The L. E. J. Brouwer Centenary Symposium, Amsterdam: North Holland, 465–486. Thurston, William P. (1994), “On proof and progress in mathematics”. Bulletin of the American Mathematical Society 30, 161–177.

107

Grazer Philosophische Studien 76 (2008), 109–134.

EPISTEMOLOGY AND EMPIRICAL INVESTIGATION1 Kristoffer AHLSTROM UMass Amherst Summary Recently, Hilary Kornblith has argued that epistemological investigation is substantially empirical. In the present paper, I will first show that his claim is not contingent upon the further and, admittedly, controversial assumption that all objects of epistemological investigation are natural kinds. Then, I will argue that, contrary to what Kornblith seems to assume, this methodological contention does not imply that there is no need for attending to our epistemic concepts in epistemology. Understanding the make-up of our concepts and, in particular, the purposes they fill, is necessary for a proper acknowledgement of epistemology’s role in conceptual improvement.

1. Introduction In his book Knowledge and Its Place in Nature (2002), Hilary Kornblith makes an intriguing case for the re-conceptualization of epistemological analysis from a largely non-empirical to a substantially empirical investigation, arguing that knowledge—one of the main targets of epistemological investigation—is a natural kind, open to straightforward empirical scrutiny. Assuming that knowledge is not unique in this respect, which is an assumption that Kornblith, indeed, seems to make, we may generate the following argument: The Argument. (A) All objects of epistemological investigation are natural kinds. 1. Earlier versions of this paper were presented at the University of Massachusetts Amherst as well as at the 2007 meeting of the Danish Epistemology Network in Copenhagen, May 2007. I would like to thank both audiences for valuable discussions and comments, and am particularly indebted to Hilary Kornblith, Klemens Kappel, Joseph Levine, Åsa Wikforss, Helge Malmgren, Radha Vij, Alex Sarch, and Kelly Trogdon.

(B) If (A), epistemological investigation is substantially empirical. (C) Hence, epistemological investigation is substantially empirical (A, B, MP). (D) If (C), a thorough understanding of our epistemic concepts, over and above the phenomena that they pick out, is irrelevant to epistemological investigation. (E) Hence, a thorough understanding of our epistemic concepts, over and above the phenomena that they pick out, is irrelevant to epistemological investigation (C, D, MP). Kornblith has never explicitly stated this argument. Still, I take it that it provides one of the most reasonable rationales for his more general claims about the implications of his results concerning knowledge to epistemological analysis at large.2 The plausibility of this interpretive claim should become more obvious as we go along. That being said, I will, in the following, scrutinize, qualify, and criticize the Argument in two steps. More specifically, §§ 2 through 4 will serve to contest (A) but defend (C), by showing that the latter premise is plausible even given that all objects of epistemological investigations are artifactual (or “socially constructed”) rather than natural kinds. However, §§ 5 through 7 will show that (E), nevertheless, does not follow from (C), since (D) is false and the claim that epistemological investigation is substantially empirical hence, does not imply that an understanding of our epistemic concepts is irrelevant to epistemology. 2. On the implausibility of premise (A) It should be beyond doubt that the Argument is valid. Indeed, it consists in two modus ponens arguments, where the conclusion of the first, i.e., (C), makes up the first premise of the second. However, I would like to contest its soundness. For one thing, it hinges on (A), i.e., the controversial assumption that all objects of epistemological investigation are natural kinds. As already mentioned, Kornblith (2002) has, indeed, argued that knowledge, as it is being studied by cognitive ethologists (cognitive ethology being the study of animal cognition), is a natural kind. However, the crucial question here is whether this claim may be 2. See, e.g., Kornblith (2007, 2006).

110

generalized to other objects of epistemological study, so as to render (A) plausible. Take epistemic justification, for example. What are the prospects for extending Kornblith’s case for knowledge to justification? Unfortunately, unlike knowledge, justification is not an entrenched concept in cognitive ethology. Hence, it is questionable whether Kornblith’s case for knowledge can be extended to justification in any straightforward way. In fact, it is hard to see exactly how justification, together with such related concepts as evidence, understanding, and rationality, at all could correspond to natural rather than artifactual (or “socially constructed”) kinds, the latter of which do not comprise, say, homeostatically structured conglomerates of properties independent of human understanding, but rather a grid whose structure reflects nothing but human intentions.3 Still, as has been noted by Alvin Goldman and Joel Pust (1998), the lack of natural kind status hardly places the topic of justification (or that of evidence, understanding, and rationality) outside the scope of epistemological analysis. So, on pain of radically restricting the scope of epistemological analysis (an option that I will not consider), the defender of The Argument has to face up to the following problem: Problem 1. Unless (A) holds, there is little reason to believe that the applicability of epistemological analysis stretches beyond the analysis of one particular object of epistemological investigation, namely knowledge. And even this particular application is contingent upon the admittedly controversial claim that knowledge, in fact, is a natural kind. The following two sections discuss two solutions to Problem 1, both of which amount to the claim that there is a case to be made for extending the conception of epistemological analysis as substantially empirical to the analysis of artifactual kinds.

3. For an account of natural kinds as homeostatically related properties, see Kornblith (1993).

111

3. A first attempt to save (C): Content externalism about artifactual kind terms The first solution starts out with the observation that it might be plausibly argued—and, indeed, has been argued by Putnam (1975) and, more recently, by Kornblith (forthcoming)—that content externalism provides the correct semantics not only for natural kind terms but also for artifactual kind terms. Rather than directly contesting this line of argument, the second solution (which is the one I will favor) concludes that, as it turns out, the plausibility of extending the claim about empirical investigation to artifactual kinds is largely independent of which semantic theory one accepts for the latter. Before evaluating any of these solutions, however, we need to say something about what constitutes artifactual kinds and, in particular, what distinguishes them from natural kinds. To a first approximation, we may characterize artifactual kinds negatively as not comprising homeostatically clustered properties. However, even disregarding the fact that this characterization is hardly informative, it does not even uniquely pick out artifactual kinds, unless natural and artifactual kinds exhaust the realm of kinds (which they do not). So, by way of a positive characterization, we may say that artifactual kinds are somehow dependent on human intentions. However, this formulation is not only vague but also potentially misleading if not further qualified. Take polyethylene or amphetamine, for example. Since they are synthetic substances, it is plausible to assume that neither of them would be around if it were not for certain human intentions, pertaining to the need for a light, flexible, yet tough material or a substance to fight fatigue and increase alertness among servicemen. Still, this only serves to show that the existence of some (instances of ) synthetic substances are causally dependent on certain human intentions. It does not show, however, that the kinds to which those substances correspond are ontologically dependent on human intentions. That is, it does not show that the identity conditions for polyethylene or amphetamine—i.e., the conditions specifying what makes something an instance of polyethylene or amphetamine—are, in any interesting sense, intertwined with human intentions. Indeed, an acknowledgement of this very fact is implicit in what we take to be the best explanation of why polyethylenes and amphetamines fit into reliable inductive generalizations better than any random motley of properties. This explanation assumes that polyethylenes and amphetamines are endowed with a shared, underlying nature (i.e.,

112

C2H4 and C9H13N, respectively), and that this, furthermore, accounts for the fact that some inductions involving the respective substances are successful (e.g., from “this is amphetamine” to “this will increase stamina but decrease appetite if ingested”) while others are not (e.g., “this is made of polyethylene” to “this is blue”). Hence, they may plausibly be considered natural kinds. Not so for, say, pens—a clear example of an artifactual kind. There is no need to assume that all pens share an underlying nature to explain why certain inductions involving pens are successful (e.g., from “this is a pen” to “this can be used to write with”) while others are not (e.g., from “this is a pen” to “this is warm”). The reason is that instances of artifactual kinds owe their kind membership exclusively to the fact that they fulfill certain purposes. More specifically, I suggest that Identity Conditions for Artifactual Kinds. the identity conditions of artifactual kinds are given by sets of human intentions, pertaining to the fulfillment of certain purposes.4 Clearly, this is not to say that artifactual kinds consist of sets of human intentions and purposes, but that what determines whether or not something is an instance of a particular artifactual kind pertains to whether that something can fulfill certain purposes and, thereby, answer to a specific set of human intentions.5 Thus, a pen is a pen (roughly) by virtue of fulfilling the purpose of drawing and writing and, thereby, answering to certain human intentions regarding creative outlet and communication, just like a key is a key (roughly) by virtue of serving the purpose of locking and unlocking doors, lockers, etc., and, thereby, answering to a set of human intentions regarding controlled access to certain spaces.6 Let us now consider Kornblith’s (forthcoming) claim that the semantic mechanisms of reference for artifactual terms are insensitive to these ontological differences between natural and artifactual kinds. Take an SUV, for example—clearly, an example of an artifactual kind. Unlike the case of 4. See Thomasson (2003) for a more detailed suggestion along these lines. 5. It might be argued that it, for some artifactual kinds, is not sufficient for membership that something merely can fulfill certain purposes and, thereby, answer to certain human intentions but that it has to have come about as the result of an intention to fulfill those purposes. See Thomasson (2003, 594) for a discussion. 6. I am not suggesting that such sets of intentions can be summed up in anything like a clear-cut conjunction of properties. This picture—just like actual categorizations of artifacts—is fully compatible with conceptual fuzziness and in-between cases.

113

water and polyethylene, there is no reason to assume that SUVs share an underlying nature, since an explanation of why we categorize the world and successfully reason in terms of SUVs and non-SUV type vehicles does not need to go beyond factors pertaining to certain (potentially superficial) properties regarding form (e.g., relative size) and function (e.g., performance), answering to certain human intentions concerning traveling and transportation. In fact, I am, personally, not sure what makes something an SUV and, in particular, not what distinguishes it (if anything at all) from a jeep, van or any other fairly big motor vehicle with four wheels. Regardless of whether I, thereby, just happen to be exceptionally uninformed as for motor vehicles, however, I take it to be fairly uncontroversial that I, nevertheless, just succeeded in referring to SUVs. How can that be? To a first approximation, it may be due to the dual fact that (a) there are people in my linguistic community that do know what makes something an SUV and (b) my successful reference to SUVs is parasitic upon their knowledge and ability to discriminate SUVs from non-SUV type vehicles. But are these conditions necessary for successful reference? Is it, in particular, necessary that there is at least one member of my linguistic community that knows what, thereby, constitutes SUVs? Remember that what makes something an SUV pertains to a set of human intentions and purposes—not anything like an underlying nature, shared by all SUVs. So, is successful reference contingent upon there being at least one member of my linguistic community that knows what this set is, i.e., to which intentions SUVs need to answer and what purposes they need to fulfill? Consider the following two responses, corresponding to two variants of content externalism: Social Externalism about Artifactual Kind Terms. Successful reference to artifactual kinds (only) requires that there is at least one member of the relevant linguistic community (i.e., an “expert”) that can correctly delineate the set of human intentions and purposes that provides the identity conditions for the kind in question. Subsequent instances of successful reference to this kind are then parasitic upon the discriminatory competence of this member.7 Causal Externalism about Artifactual Kind Terms. Successful reference to artifactual kinds (only) requires that a sample of the kind has been picked out in an initial act of baptism through an 7. See Burge (1986).

114

ostensive definition, fixing the set of human intentions and purposes that provides the identity conditions for the kind in question and establishing a socially sustained chain of reference upon which subsequent instances of successful reference to whatever bears a certain equivalence relation (most plausibly spelled out in terms of certain potentially superficial properties regarding form and function) to the ostended sample are parasitic.8 Clearly, neither formulation is intended to constitute a full-fledged theory. If anything, they both give rise to further questions. Take Causal Externalism, for example. For one thing, it is somewhat puzzling how the mere causal relation involved in an act of baptism could fix one set of human intentions and, consequently, pick out a single kind. For example, take a wooden, box-like, and fairly heavy object of 35 by 35 inches that we may name b. Picking out b as a sample for an artifactual kind, what determines the relevant set of intentions, given that b can be used to sit on (i.e., as a chair), to sit by (i.e., as a table), to stop doors from closing (i.e., as a door-stop), if pushed down a set of stairs, as a device stopping people from ascending (and possibly hurting them quite badly in the process of doing so), and probably a whole host of other things? One plausible suggestion is that there is something about the mental state of the baptizer that determines the relevant set of intentions, namely that the mental state (at the very least) instantiates that very set of intentions. If so, however, it becomes harder to distinguish Causal from Social Externalism. For example, is successful reference contingent upon (a) the act of baptism and the resulting chain of reference, or (b) the fact that there is a baptizer, carrying the heaviest burden in the division of linguistic labor due to her insight into the relevant set of intentions (granted introspective access, of course)? It is not so clear to me which one is the case here. However, the relevant question for our purposes is how to analyze artifactual kinds and, in particular, whether any of the above considerations render the claim that epistemology is substantially empirical implausible. They do not. The idea that epistemology is substantially empirical can be plausibly extended to artifactual kinds, regardless of whether Social or Causal Externalism holds and for the following reason: Both Social and Causal Externalism are fully compatible with successful reference despite substantial ignorance regarding the actual properties of the entities or 8. See Kornblith (forthcoming).

115

phenomena picked out. Granted, Social Externalism implies that there is at least one member of the relevant linguistic community that has insight into the identity conditions of the kind in question. So, in the general case, the baptizer, clearly, knows something about the artifactual kind she is baptizing, such as that it can be used for certain purposes and, thereby, answer to certain human intentions. However, her knowledge of many of the properties that constitute the artifactual kind in question may be ever so limited—or better said: there is nothing in her role as a baptizer that hinders her knowledge from being limited thus. That is, even if we reject Causal in favor of Social Externalism, the epistemically most privileged user of an artifactual kind term, i.e., the baptizer herself, may have an ever so limited insight into the properties that do or may make up instances of the kind in question. As the reader surely suspects, it is exactly in this gap between successful reference and an insight into the properties of the referent picked out that our first solution to Problem 1 gets its foothold, since such a gap makes possible scenarios in which (a) a majority of speakers either are largely ignorant of or have a highly inaccurate conception of many of the properties that make up the kind that they are successfully referring to and (b) even the most epistemically privileged speaker (i.e., the baptizer) may have an ever so limited insight into the multitude of properties that may not in any straightforward way be inferred from the intended purpose of the kind in question. Given such a gap, I take it that it would reasonably follow that epistemology is substantially empirical even given that most (if not all) objects of epistemological investigation are artifactual kinds. 4. A second attempt to save (C): Epistemology and conceptual refinement The problem with this solution, however, is that it is far from controversial that anything like Social or Causal Externalism provides the correct semantics for artifactual kind terms. While remaining essentially neutral on this issue, and in an attempt to develop a somewhat more dialectically sensible rationale for (C), I will now argue that, even if a strong form of internalism turned out to provide the correct semantics for artifactual kind terms, this would in no way undermine the claim that epistemology is substantially empirical. The argument will also indicate that it was not externalism that did the job in the above solution after all.

116

So, consider the following: Strong Internalism about Artifactual Kind Terms. Successful reference to artifactual kinds requires that the speaker can correctly delineate the set of human intentions and purposes that provides the identity conditions for the kind in question. Strong Internalism makes up the other extreme of the semantic spectrum; it is not enough that there is an appropriate causal chain of reference, nor that someone in the linguistic community can delineate the set of intentions in question—the speaker must herself be able to make such a delineation for her to successfully refer. Perhaps this is a more plausible thesis about the semantics of artifactual kinds, or perhaps it is not. What is important to note for our purposes is that even if Strong Internalism turned out to be true, this would only imply that every competent user of artifactual kind terms were in the same predicament as the epistemically most privileged user in the Social Externalism scenario. That is, while being extremely informed as to the relevant set of human intentions and purposes, they may still, qua competent users, have an ever so limited insight into many of the properties that may or may not make up actual instances of the kind in question. More importantly, they may, in the epistemic case, be ever so uninformed as for properties of epistemological significance—or so I will now argue. First, consider the following definition: Conceptual Accuracy. A concept is accurate to the extent that it provides a correct and complete description of its referent. The idea here is two-fold: (a) there are aspects of concepts that do not serve to determine reference, and (b) these aspects may be represented in terms of descriptions. Let us look closer at (a) first. Ever since Kripke (1980), it has been customary to distinguish between factors that serve to fix and factors that serve to determine reference. For example, while whatever conceptual component responsible for my tendency to think of horses as having four legs may serve to fix the reference of horse, i.e., to be a helpful tool in picking out actual horses in my environment, it does not determine reference, for the simple reason that some horses are amputees. As such, factors fixing reference, clearly, play an important cognitive role

117

in our mental life, by significantly facilitating our interaction with the extra-mental world. Contra the descriptivist, however, these factors should not be confused with the factors determining reference. We will return to this point in a second. Let us now turn to (b). On closer inspection, we see that, if there were no conceptual aspects that could be represented in terms of descriptions, it is hard to see how concepts at all could be the objects of any kind of analysis in the first place. On any view of concepts—be it concepts as abilities, Forms, senses or mental representations—concepts serve to categorize the world and the systems of categorization that arise from concept use may be represented in terms of descriptions. Hence, a concept that serves to put all and only blue objects that weigh more than two pounds in one category may be characterized in terms of the description “is blue and weighs more than two pounds,” quite independently of one’s favored ontology of concepts. This is not, of course, to say that whatever mental event that is causally responsible for the categorization takes the form of a description—that would have to be established through empirical research—which is exactly why I am not saying that concepts are descriptions but merely that they may be represented as incorporating descriptions. One particular reluctance to talking about concepts and descriptions in the same sentence stems from an aversion to the traditionally influential idea that reference determination works by way of the referent satisfying a description inherent in the corresponding concept. However, this is certainly not the idea being defended here. If anything, the present notion of conceptual accuracy serves to state in clearer terms the very externalism that served to refute this descriptivist picture of reference fixing: In so far as any form of content externalism holds, having an accurate concept—i.e. a concept whose descriptive aspect provides a correct and complete description of its referent—is not a prerequisite for successful reference. However, since we are, for the moment, assuming that Strong Internalism provides the correct semantics for artifactual kind terms, we will focus on the fact that not even Strong Internalism implies that having an accurate concept of an artifactual kind is a prerequisite for successfully referring to it. The reason may be brought out as follows: Take any object x of epistemological investigation. If x is an artifactual kind, there is a set φ of human intentions and purposes that determines the identify conditions for x. What will φ contain? Given that x is an artifactual epistemic kind, it will most likely contain intentions and purposes pertaining to the attainment of certain epistemic goals, say, true belief in significant

118

matters. Is it possible to say something more specific? Well, if Strong Internalism holds, we may note that, unless we want to radically restrict the extent to which people may successfully refer to x, we have to restrict the richness of information contained in φ since, in the general case, the more information is contained in φ, the rarer is successful reference to x. Given that reference to x is widespread, however—which is, hopefully, the case for most epistemic kinds—we may, at the very least, say that being acquainted with the information contained in φ can not involve a complete apprehension of all properties that make up actual instances of x. Hence, even given Strong Internalism, the semantically most informed person—i.e., the person that has the most complete grasp of what is contained in φ—might still be in the dark as for many of the properties that may or may not make up instances of x. In other words, she may still have an inaccurate concept of the kind in question. I take it that few would deny this claim, if understood in the weak sense of there always being further facts that could be found out that do not flow from what we know just by virtue of being able to successfully refer. For example, just by virtue of successfully referring to pens and keys, I may (at least on the Strong Internalist’s story) know a whole host of things about pens and keys and, in particular, things that I may easily infer from being acquainted with the relevant sets of intentions and purposes. At the same time, there may very well be a lot of things that I do not know about pens and keys, such as its exact mechanical make-up, its molecular constitution, etc. As pointed out by Paul Griffiths (1997), this is to be expected considering that the factors determining the identity conditions are substantially less rich in the artifactual kind case than in the natural kind case: The traditional natural kinds are among the richest. The kindhood of a physical element determines almost all its salient properties. […] In contrast, knowing what sort of thing an artifact is, knowing that it is a bracelet for example, may fix very few of its features. There are just too many ways to skin a cat, or in this case too many ways to decoratively encircle the wrist. (Griffiths, 1997, 190)

119

Of course, it does not follow that being familiar with the multitude of ways that pens or keys may be crafted or wrists may be decoratively encircled is necessarily very significant to me, given that my goals, as far as pens, keys, and bracelets go, are restricted to successful, everyday interactions. What I want to claim, however, is that the same does not hold in the epistemological case. In particular, I want to claim that what is not contained in φ—i.e., what is still to be found out when we have enough knowledge for successful reference to occur—is of epistemological significance. My argument for this claim runs as follows. First, consider the following: One important epistemological desideratum is to guide epistemic inquiry. One element in this desideratum is to see to it that our epistemic vocabulary is as apt as possible, where a vocabulary is apt in so far as it invokes apt concepts, and Conceptual Aptness. a concept is apt to the extent that it serves its intended purpose well. To say that concepts serve purposes is not meant to imply anything controversial. At a very basic level, the purpose of concepts is simply to enable us to think certain thoughts, have certain beliefs, etc., and, thereby, interact with the world in more or less successful ways. Differently put, concepts provide a framework for thinking and believing (and so on), in the sense of a way of categorizing the world. As such, the extent to which one framework is better than another depends on what we wish to attain with it. This point may be illustrated in the epistemic domain by noting that we, as epistemic inquirers, are engaged in a certain project of epistemic evaluation and doxastic revision, (roughly) aimed at attaining and maintaining true belief in significant matters. An integral part of succeeding in this latter task is having an apt epistemic vocabulary, where an apt epistemic vocabulary is a vocabulary that can be used to categorize and, thereby, evaluate fellow inquirers and the world in a way that serves the purpose of attaining and maintaining true beliefs in significant matters. Clearly, some concepts will serve this purpose better than others. In particular, the following seems a reasonable claim: If epistemic vocabulary V1 is more refined than vocabulary V2—i.e., if V1 incorporates accurate concepts to a larger extent than V2 does—then V1 is more apt than V2, ceteris paribus.

120

Furthermore, it is reasonable to assume that getting acquainted with the properties that may or may not make up actual instances of x—and, in particular, those properties that may not be readily inferred from being acquainted with what is contained in φ—would enable us to refine our concept of x, in the sense of pruning it so as to provide a more correct and complete description of its referent. Finally, it seems fairly uncontroversial that the proper method for getting acquainted with those properties will have to be empirical. Take justification, by way of illustration. Construed as an artifactual kind, the relevant set of intentions and purposes would most likely pertain to the flagging of appropriate sources of information, where the appropriateness is understood in terms of truth-conductivity. More than that, on Strong Internalism, every competent speaker would be perfectly familiar with the details of this set. But does this imply that they, thereby, know everything there is to know about justification? Does it, in particular, follow that there is nothing else to find out that is of epistemological significance? That seems unreasonable. In particular, it would be of epistemological significance to find out, among other things, (a) what external phenomena actually satisfy the relevant requirements of truth-conductivity, (b) about the multiplicity of properties that may or may not make up these phenomena, and, perhaps more importantly, (c) how they fit in to the causal fabric of the world and, hence, may not only be better understood but also be manipulated to the benefit of the epistemic inquirer—all of which seems to be things that cannot in any straightforward way be inferred from the relevant set of intentions and purposes, nor be discovered without recourse to an empirical investigation. So, in the general case, and in so far as conceptual refinement may rightfully play a substantial role in epistemology, the claim that epistemological investigation is a substantially empirical investigation is largely independent not only of whether the objects of epistemological investigations are natural or artifactual kinds, but also of any content externalist or internalist considerations with respect to the latter. In other words, given that the above line of reasoning does, indeed, apply to most (if not all) objects of epistemological investigation—and I see no reason why it would not—we may conclude that the claim that epistemological analysis is substantially empirical does not hinge on (A). Indeed, the preceding discussion provides us with reason to take the following argument to be sound:

121

(A*) For every object x of epistemological investigation, x is either a natural or an artifactual kind. (B*) If (A*), epistemological investigation is substantially empirical. (C) Hence, epistemological investigation is substantially empirical (A*, B*, MP). This concludes my solution to Problem 1. Next, we will be looking into why (D) does not hold—i.e., why it does not follow from (C) that an understanding of our epistemic concepts is largely irrelevant to epistemological investigation. In the process of doing so, we will not only provide further evidence to the effect that epistemology is substantially empirical, but also introduce a more radical means to attaining an apt vocabulary: conceptual reconstruction. 5. An alleged rationale for (D): Factual analysis and the AC principle Returning to The Argument, let us now turn to premise (D) and the conclusion (E), stating that a thorough understanding of our epistemic concepts, over and above the phenomena that they pick out, is irrelevant to epistemological investigation. Kornblith’s commitment to this conclusion comes out most clearly in his critique of the idea that the (sole) job of epistemology is to analyze epistemic concepts by way of intuitions about hypothetical cases—a view that he characterizes as follows: Appeals to intuition are designed to allow us to illuminate the contours of our concepts. By examining our intuitions about imaginary or hypothetical cases, we should be able to come to an understanding of our concepts of, for example, knowledge and justification. The goal of epistemology on this view, or, at a minimum, an essential first step in developing an epistemological theory, is an understanding of our concepts. (Kornblith, 2006, 11–12)

Kornblith continues: My own view is that our concepts of knowledge and justification are of no epistemological interest. The proper objects of epistemological theorizing are knowledge and justification themselves, rather than our concepts of them. (Kornblith, 2006, 12)

In light of the reasonable claim that some initial examination of our epistemic concepts might be necessary in order to fix the subject matter,

122

Kornblith makes it clear that his main disagreement with the tradition of epistemology as conceptual analysis concerns the scope of such a semantic investigation. More specifically, he claims that the semantic investigation called for is “utterly trivial” and, thereby, in no way related to the two thousand year old project that, in a tradition stemming from Plato’s the Theaetetus and culminating in the Gettier-inspired literature, typically falls under the heading of the analysis of knowledge and justification. When trying to put this claim in more precise terms, it serves us well to make a distinction between two stages of epistemological investigation. The first one corresponds to the identification of an epistemological object F, i.e., of fixing the subject matter (if only tentatively) through picking out a selection of what we take to be paradigmatic instances of the concept ‘F.’ In doing this, our concepts ‘F’ (and the categorization intuitions they give rise to), clearly, play a vital role, if only in the sense that, in order to identify an F, one needs to have some kind of grasp of what it is to be (an) F. Naturally, depending on the extent to which ‘F’ is ambiguous or imprecise, this process of identification may be more or less demanding. Regardless, the purpose of identification is to pave the way for the more substantial and straightforwardly empirical aggregation of the characteristics that may or may not be found in Fs, in order to reach a satisfactory answer to the question that incited inquiry in the first place: “What is (an) F?”9 Against the background of this distinction, I would like to characterize Kornblith’s notion of analysis as follows: Factual Analysis (FA). Identification: For any concept ‘F,’ Identify a set Q, containing a selection of what we take to be paradigmatic instances of ‘F.’ Aggregation: Against the background of an a posteriori investigation into the (extra-mental) elements found in Q, aggregate a set of characteristics that specify what actually constitutes (an) F. Is FA an empirical of analysis? In so far as aggregation goes, the answer would have to be ‘yes.’ As stated, however, it remains an open question whether the same goes for identification. More specifically, the question 9. This distinction between identification and aggregation is borrowed from Sen’s (1981) treatment of the issue of poverty.

123

remains open until it is specified whether the fixing of a set of paradigmatic examples presupposes an a priori access to conceptual content as opposed to an empirical investigations into what people tend to mean by terms. However, the question does not need to be settled for present purposes since it seems reasonable to assume that either of the two claims would have to be true: Either (a) identification calls for a substantial investigation, in which case an empirical method will be desirable due to a superior methodological rigor, or (b) identification does not call for a substantial investigation, in which case an a priori method would not compromise the claim that epistemological investigation—if understood along the lines of FA—is still a substantially empirical investigation. We have to keep in mind, however, that Kornblith does not only commit himself to the idea that FA as a viable method of epistemological analysis, but to the stronger claim that FA provides a complete method, i.e., that there are no other aspects to epistemological investigation over and above identification and aggregation, as spelled out above. This brings us to (D). If FA is all there is to epistemological analysis, there is no need for a thorough understanding of our epistemic concepts—as in an understanding that goes beyond whatever semantic investigation is needed for identification—since the main component of epistemological analysis will consist in a purely empirical investigation into the epistemic phenomena picked out. In other words, if FA provides a complete methodology, we should accept (D). But why should we take FA to provide a complete methodology? Suppose that it proves possible to successfully conduct a series of FAs, providing an account of what constitutes (an) F for every epistemic concept ‘F.’ This would, undoubtedly, be an impressive accomplishment. But would it mark the end of epistemological investigation? Considering the picture of epistemology as the pursuit of an apt epistemic vocabulary, the question may be reformulated as follows: Would such a set of analyses necessarily yield a fully apt epistemic vocabulary? Considering what was said above in relation to conceptual refinement, it might be tempting to answer the question in the positive, under the assumption that a more apt set of concept simply is a more accurate set of concepts. In other words, the answer would be ‘yes’ if the following principle can be shown to hold: The AC principle. The pursuit of a more apt set of concepts reduces to that of providing a more accurate set of concepts.

124

In other words, we have established the following chain of dependency: If the AC Principle holds, then it is reasonable to assume that FA yields not only accurate but apt concepts and, hence, provides a complete epistemological methodology. Furthermore, if FA provides a complete epistemology methodology, then it is reasonable to assume that (D) holds. However, if the AC Principle does not hold, (D) remains unwarranted since FA, thereby, might yield accurate but not necessarily apt concepts. More specifically, I will argue that Problem 2. unless the AC Principle holds, FA fails to acknowledge epistemology’s role in the particular kind of conceptual improvement involved in conceptual reconstruction. In the following two sections, I will (a) spell out this methodological component of conceptual reconstruction, (b) provide two examples of cases in which the need for it indicates that increased accuracy does not imply increased aptness, even if we assume content externalism for the corresponding concepts, and (c) conclude that the AC Principle is an unviable epistemological assumption and (D) is without warrant. 6. Conceptual aptness in epistemology A strong motivation for content externalism about natural kind terms is that it provides us with a straightforward and intuitive explanation of disagreement in the natural sciences.10 That is, given content externalism, successful reference is completely compatible with inaccurate concepts and theories on part of the person referring. Hence, even if Dalton, Rutherford, and modern physicists were and are working with substantially different theories and concepts of the atom, they were and are essentially talking about the same phenomenon, i.e., the atom. Indeed, only if we say this can we make the further claim that they disagree about the latter’s constitution and that contemporary theories of the atom constitute improvements of the earlier theories of Dalton and Rutherford, in line with the overall progress of science.

10. See, e.g., Putnam (1975).

125

Does this rather neat picture of scientific disagreement and improvement carry over to epistemology? One of the major reasons for doubting that it does is that (a) we have reason to believe that the majority of epistemological objects are artifactual rather than natural kinds, that (b) it remains to be shown that content externalism provides the correct semantics for artifactual kind terms, and that (c) it is not obvious that explaining epistemological disagreement requires assuming referential continuity, rather than that our epistemological project is continuous with that of our epistemological predecessors (e.g., in the sense that we are all concerned with the search for an apt epistemic vocabulary). However, being able to explain disagreement is, clearly, only one motivation for content externalism and there might very well be independent reasons for wanting to defend such externalism in epistemology. For this reason, I will now consider a dialectically more robust strategy by assuming content externalism but showing that the AC principle still does not hold. The strategy will first be demonstrated on an abstract level and then, in the next section, illustrated by way of two examples. So, first consider how referents get assigned to concepts on an externalist story. For dialectical purposes, let us assume the strongest form of externalism, i.e., Causal Externalism. On Causal Externalism, a term or concept gets assigned a referent by way of an initial act of baptism. In other words, the referent of a concept is determined by the (mere) fact that the baptizer stands in a certain causal relation to it. As we have already seen, such a semantic story is perfectly compatible with successful reference in spite of considerable ignorance on part of the speaker, which leaves room for extensive conceptual refinement. Hence, it was suggested above that there is a connection between refinement and aptness. This, furthermore, provides at least part of a rationale for the conceptual refinement of purely descriptive concepts in science, understood as concepts the mere (or at least most central) purpose of which is to categorize, without thereby providing a normative evaluation of whether something is good or bad in relation to a specified set of goals. Hence, H2O may constitute a refinement of water, as far as chemistry goes, and mean molecular kinetic energy a refinement of temperature, as far as kinetic theory goes. However, to fully understand what constitutes an apt epistemic vocabulary, we need to keep in mind that epistemic concepts typically are normative concepts. Epistemic concepts, qua normative concepts, are tools by which we evaluate the conduct of fellow epistemic inquirers—that is, categorize their conduct as being an instance of knowledge, justification,

126

rationality, etc.—but where we, merely by virtue of such a categorization, also make an explicitly normative judgment about the extent to which what is being done is good from an epistemic point of view. This suggests that the purposes of epistemic concepts may be understood in relation to the principles in which they figure and the goals that these principles are designed to meet. Furthermore, it prompts the following question: Is there any guarantee that the referent that was originally fixed to an epistemic concept or term in an initial act of baptism, in fact, provides the best route to our epistemic goals? The answer is no. Granted, it might be possible to construct an argument to the effect that whatever we are referring to with our epistemic concepts does not provide a completely useless route to our epistemic goals, given that having true belief is an integral part of attaining many of our practical goals (and that the latter is something that we tend to do). However, it is hard to see why the referent initially fixed necessarily provides the best path to our epistemic goals. For this reason, epistemology has to consider the possibility that alternative referents may present a better route to our epistemic goals than our present referents. Hence, it makes sense to not only investigate the question of to what extent our concepts provide accurate pictures of their referents—referents that need not provide the best route to our epistemic goal—but also whether there are any alternative referents that might present a better route. Such an inquiry, however, has to take into account not only the causal structure of the world (as revealed through straightforward empirical inquiry) but also the purpose of the original concepts and, in particular, the principles in which they typically figure and the goals these principles are meant to attain, since nothing short of such an investigation will enable us to specify what would constitute a better route to our goals and, consequently, a more apt concept. 7. When accuracy does not increase aptness: Two scenarios To illustrate this point, I will now present two hypothetical scenarios in which increased accuracy does not increase aptness, since the referents in question—fixed in accordance with Causal Externalism—do not present optimal routes to our epistemic goals. I will argue that the proper epistemological strategy in those cases is not refinement but a more substantive conceptual improvement in light of the larger context of principles and

127

goals in which the concept figures. I will refer to such conceptual improvement as conceptual reconstruction, to a first approximation understood as an ameliorative activity located further out on a continuum of increasingly radical conceptual revision. A helpful metaphor here is architectural reconstruction. When reconstructing, say, a house, you start out with certain pre-existing material, i.e., the house that is to be reconstructed. Let us call this house H1. The house that results from the reconstruction—let us call it H2—might look nothing like H1. Nevertheless, H2 will (if everything goes as planned) serve a certain set of purposes better than H1 did. Perhaps H2 is more spacious, has better insulation, has a more attractive design, etc., than H1. Indeed, it is reasonable to believe that the intention to realize those properties provided the very reason for reconstruction. Similarly, conceptual reconstruction starts out with a pre-existing concept. The reconstructed concept might, in the end, look nothing like the original concept. Still, the very point of reconstruction is that the reconstructed concept serves a set of purposes better than the original concept. In the case of epistemic concepts, these purposes will be understood in relation to our epistemic goals—goals that will be spelled out in more detail below but that, to a first approximation, may be understood in terms of true belief in significant matters. However, this raises some questions about the identity conditions for concepts. Is the concept that results from reconstruction “the same concept” as the concept that we started out with? I find this question about as puzzling (and interesting) as the question whether my kitchen remains “the same” over the course of a renovation. Clearly, many of its properties will change—some vanish, some arise—and the reconstructed kitchen will not be identical to the old one. (If it were, the renovation would have failed.) At the same time, it is still my kitchen and it will still serve the same purposes—indeed, it will, hopefully, serve some of the same purposes better than my old kitchen. Analogously, I will say that a reconstructed concept C2 is “the same” concept as an original concept C1 to the extent that they both figure in relation to the same set of purposes. At the same time, however, C2 will, clearly, also be different form C1 in the sense that it has different properties and, due to this, serves the purposes in question to a greater degree. Against the background of a distinction between conceptual refinement and reconstruction, say that we perform a factual analysis of justification and let us, for simplicity’s sake, refer to this concept as our

128

concept of justification.11 Assume, furthermore, that we, in the process of identification, find that the properties by which we typically individuate degree of justification pertain to the fulfillment of epistemic duties. In fact, this would make complete sense, given what we know about the etymology of the term “justification.” As noted by William Alston, the term “has been imported into epistemology from talk about voluntary action,” which “explains the strong tendency to think of the justification of belief in deontological terms, in terms of being permitted to believe that p (not being to blame for doing so, being ‘in the clear’ in so believing)” (Alston, 1993, pp. 532 and 533, respectively).12 What does this tell us about the way that justification was originally endowed with a referent? For one thing, it lends some support to the dual claim that (a) justification was originally introduced as applying to the formation of belief, and that (b) it has traditionally been presupposed that we can form or refrain from forming beliefs by willed action—on pain of denying that ought implies can. Finally, assume that we find, in the process of aggregation and empirical investigation of the phenomenon actually picked out by our concepts, that we have no voluntary control over the formation of beliefs.13 If this turned out to be the case, we seem to have uncovered reason to believe that our concept of justification is off the mark, in that it pertains to something that we, as a matter of fact, cannot have, namely epistemic duties. What would be a proper epistemological response? Two responses are available. On the first response, we reject voluntarism as a mistaken view about the way our mind works, but retain the idea that justification applies to belief-formation. This would correspond to a simple refinement and the most promising candidate for fleshing it out would probably be some form of process reliabilism. However, as it stands, this response 11. It does not matter so much for present purposes whether there is such a thing as our concept of justification or rather a rich multiplicity, since all that the following line of reasoning requires is that we are talking in terms of a specific concept—be it a widely shared one or not. 12. See also Plantinga (1990). 13. The kind of voluntary control at issue here is what Alston (2005, 62) refers to as basic voluntary control. Undoubtedly, this is not the only kind of voluntary control—in fact, Alston distinguishes between three types of (decreasingly extensive) voluntary control as well as different grades of indirect voluntary influence (pp. 62–80). However, Alston also provides convincing arguments to the effect that, even given increased taxonomical complexity, there does not seem to be any voluntary control or influence such that it both (a) applies to the psychology of common epistemic inquirers, and (b) is sufficiently extensive to warrant talking in terms of genuine duties and obligations.

129

suffers from a significant problem: It takes for granted that the referent inherited from our deontological predecessors does, in fact, provide the optimal route to our epistemic goals. While this certainly cannot be ruled out, nor can it be assumed, for reasons brought out in the previous section. This leads us to the second response, on which we engage in an inquiry best described as a continuation of aggregation, with the crucial qualification that it is preceded and guided by an investigation into the intended purpose of the original concept, in an empirical search for properties that may serve that purpose better, given a relevant set of epistemic principles and goals. For example, if such an investigation were to demonstrate that the purpose of justification is to flag certain voluntary acts (previously identified as acts of belief-formation) as appropriate sources of information, and the appropriateness in question is typically understood in relation to a goal of attaining and maintaining true belief in significant matters, one possible route for empirical inquiry would be to identify a kind of voluntary act that tends to yield and support true belief, thereby providing material for a reconstructed concept. The resulting view would retain voluntarism but reject the idea that justification should apply to beliefformation. This would correspond not to a refinement but a reconstruction, where justification gets “re-baptized,” so to speak, and justification, thereby, gets assigned a new referent. This brings us to the second scenario and the traditionally most influential candidate for such a voluntary act: introspection. More specifically, say that we, in the process of identification, find that we tend to determine degree of justification by reference to an introspective evaluation of reasons on part of the allegedly justified (or unjustified) subject. Let us, furthermore, assume that this particular notion of justification originally entered into the discourse of evaluating epistemic subjects some four hundred years ago by way of Descartes’ ideas about what one perceives clearly and distinctly on introspection. What does this tell us about the way in which justification was originally endowed with a referent? At the very least, it lends some support to the dual claim that (a) justification was originally introduced as applying to acts of introspection, and that (b) it has traditionally been presupposed that such acts provide a powerful and reliable access to the grounds for our beliefs. However, suppose that we also find that we, as a matter of empirical fact (say, facts uncovered by cognitive psychology), seldom have access to the epistemic qualities of the processes by which we form beliefs and, furthermore, that the stories (consciously

130

or unconsciously) reconstructed by us as to the epistemic etiology of our beliefs are often quite inaccurate.14 If that turned out to be the case, what would be a proper epistemological response? Again, two responses are available. On the first response, we would try to identify conditions under which we do have reliable access to our reasons and, then, refine our concept accordingly. However, the very same research hinted at in the previous paragraph gives us reason to think that such conditions are quite hard to come by. Hence, the second response: Conduct an investigation into the purpose of our (supposedly inapt) concept, let the result of such an investigation guide further empirical aggregation of candidate properties that may figure in a reconstructed concept that fills the same (or close to the same) purpose, without being committed to the idea that we have a reliable introspective access to the epistemic qualities of our belief-forming processes. For example, if an investigation into the purpose of our concept of justification were to reveal that its purpose is to flag certain voluntary acts (previously identified as acts of introspection) as appropriate sources of information, and the appropriateness is (again) typically understood in relation to a goal of attaining and maintaining true belief in significant matters, one way for empirical inquiry to proceed would be to identify an alternative kind of voluntary act (one candidate being certain acts of reasoning) that tends to yield and support true belief. An empirical aggregation preceded and guided by an investigation into the purpose of our concept would, thereby, provide material for a reconstructed concept. These hypothetical examples serve to illustrate two points with a direct bearing on (D) and the issue of accuracy and aptness: First, there are possible cases in which merely attending to the referents of our current concepts would not enable us to complete the task of identifying more apt concepts, since those referents provide sub-optimal routes to our epistemic goals. For this reason, an epistemological investigation need to attend not only to the referents of our concepts but also to other properties that are not in any obvious way implicated by our current concepts but that may, nevertheless, figure in a more apt vocabulary. Second, this empirical investigation needs to be preceded and guided by an investigation into the purposes of the original concepts, providing the empirical inquiry at issue with a direction in the form of an understanding of what properties to look for. For this reason, the substantially empirical method in no way 14. See Wilson (2002) for some evidence to this effect.

131

eliminates the need for an understanding of our epistemic concepts and, in particular, the particular purposes for which we employ them. To the contrary, such an understanding plays a vital role in the search for a more apt epistemic vocabulary. Unless it can, somehow, be shown that scenarios like the two just considered are impossible (which is different from arguing that they are not actual), we have to leave room for the possibility of conceptual reconstruction when providing a methodological framework for epistemology. More specifically: It is beyond doubt that some instances of conceptual improvement in epistemology might indeed flow from a straightforward conceptual refinement. However, given that epistemology is concerned with explicitly normative concepts, some cases of improvement must take into account not only (a) facts about the referent but also (b) facts about the intended purpose of the concept in question, so as to guide empirical aggregation in the search for (c) properties that might not be implied by the original concept, nor present themselves in any straightforward way through an unconditional empirical investigation into the referent, but that might nevertheless furnish a reconstructed concept with an increased aptness, given the principles and goals that the original concept was supposed to (but failed to) meet. Hence, even if substantially empirical, a proper epistemological methodology needs to leave room for attending to our concepts, and in particular to the purpose for which we employ the epistemic concepts that we do. This is why epistemic concepts—contra the decrees of FA—do not drop out of the epistemological picture as soon as we move beyond the initial stage of delimiting a set of paradigmatic examples. 8. Conclusions By way of recapitulation, we have established two conclusions. The first conclusion is that (C) may be established independently of (A). That is, under the plausible assumption that conceptual refinement plays an important role in epistemological theorizing, the claim that epistemology is a substantially empirical investigation is not contingent upon the admittedly controversial idea that all objects of epistemological investigation are natural kinds. In fact, (C) can be shown to be plausible even under the dual assumption that (a) all objects of epistemological investigation are artifactual kinds, and that (b) what I have referred to as Strong Internalism

132

provides the correct semantics for artifactual kind terms. The second conclusion is that (E) does not follow from (C), since (D) is false. That is, concepts are not only relevant to identification, i.e., the fixing of a (non-exhaustive) set of (what we take to be) paradigmatic examples of the phenomenon under investigation. An insight into the purposes of our epistemic concepts is, in some cases of conceptual reconstruction, also a prerequisite for knowing how to direct the process of aggregation in the improvement of our conceptual apparatus and, hence, answering a question that ought to lie at the heart of any epistemology interested in not only describing but also improving on epistemic inquiry, namely “Given our epistemic goals, what would be a set of epistemic concepts that served us better?”

REFERENCES Alston, W. P., (2005), Beyond “Justification”: Dimensions of Epistemic Evaluation, Cornell University Press. Burge, T., (1986), “Individualism and Psychology,” Philosophical Review 95, 3–45. Goldman, A., and Pust, J., (1998), “Philosophical Theory and Intuitional Evidence,” in M. DePaul and W. Ramsey, eds., Rethinking Intuition: The Psychology of Intuition and Its Role in Philosophical Inquiry, Rowman & Littlefield Publishers, Inc., 179–197. Griffiths, P. E., (1997), What Emotions Really Are: The Problem of Psychological Categories, The University of Chicago Press. Kornblith, H., (1993), Inductive Inference and Its Natural Ground, The MIT Press. — (2002), Knowledge and Its Place in Nature, Oxford University Press. — (2006), “Appeals to Intuition and the Ambitions of Epistemology,” in S. Hetherington, ed., Epistemology Futures, Oxford University Press, 10–25. — (2007), “Naturalism and Intuitions,” Grazer Philosophische Studien 74, 27–49. — (forthcoming), “How to Refer to Artifacts,” in E. Margolis and S. Laurence, eds., Creations of the Mind: Essays on Artifacts and their Representation, Oxford University Press, forthcoming. Plantinga, A., (1990), “Justification in the 20th Century,” Philosophy and Phenomenological Research 50 (supplement), 45–71. Putnam, H., (1975), “The Meaning of ‘Meaning’,” reprinted in his Mind, Language and Reality: Philosophical Papers vol. 2, Cambridge University Press, 215–271.

133

Sen, A., (1981), Poverty and Famines: An Essay on Entitlement and Deprivation, Oxford University Press. Thomasson, A. L., (2003), “Realism and Human Kinds,” Philosophy and Phenomenological Research LXVII (3), 580–609. Wilson, T., (2002), Strangers To Ourselves: Discovering the Adaptive Unconscious, Harvard University Press.

134

Grazer Philosophische Studien 76 (2008), 135–147.

THE USE OF ‘USE’ Daniel WHITING University of Southampton … the meaning of a word is its use in the language. (Ludwig Wittgenstein) Not every use … is a meaning. (Ludwig Wittgenstein)1

Summary Many equate the meaning of a linguistic expression with its use. This paper investigates prominent objections to the equivalence claim and argues that they are unsuccessful. Once one suitably distinguishes the kind of use to be identified with meaning, the two do not diverge. Doing so, however, requires employing terms that are cognates of ‘meaning’ (if not ‘meaning’ itself ). Nonetheless, I stress, this does not count against the equivalence claim. Moreover, one should not assume that the circularity on this occasion is vicious.

1. Introduction The dictum that meaning is use, that for a word to have a meaning is for it to have a use, is typically presented as placing emphasis on the public nature of linguistic activity, as appropriately situating the notion of meaning in its characteristic context of communication, and more generally as dissuading us from a Cartesian conception of subjects as essentially cut off from one another in private realms. According to its proponents, the appeal to use promises to de-mystify meaning by suitably re-connecting talk of meaning with the familiar and concrete linguistic practices into which we are naturally habituated. Since its first airing, the claim that meaning is use has gained considerable currency. While perhaps not as popular as it once 1. Respectively Wittgenstein 1967: § 43; Wittgenstein 1982: § 289. Note that this paper is intended as neither an exegesis nor a defense of Wittgenstein’s views.

was, it is fair to say that, in one form or another, it is accepted by many prominent philosophers. Indeed, Brian Loar claims that contemporary ‘theory of meaning is divided into two: truth theories, and use theories’ (2006: 85; cf. Borg 2004: 4).2 It seems, then, that the ‘Homeric struggle’ Peter Strawson famously identified battles on (2004: 132). The use-theorist does not merely suggest that linguistic meaning is (in some sense) determined by or had in virtue of use, that how expressions are employed fixes what meanings they possess. Whatever their disparate views on the nature of meaning, many philosophers would accept such claims as more or less trivial (see Davidson 2005: 12–3; Glock 1996b: 209; Higginbotham 2006: 75; Lewis 1975). Indeed, one might ask, how else could the terms of a natural language such as English, a conventional sign-system, acquire the meanings they have? The more controversial claim is, rather, that a meaning, that which is thereby determined, just is a use. This putative insight has been elaborated into various full-fledged theories that apart from their common root might otherwise bear little resemblance to one another.3 Given the details of a particular worked-out proposal, various objections to a use-centred theory of meaning might or might not be effective. Among the multitude of concerns is whether such a theory could accommodate the constancy and communicability of meaning, explain the connections between words and reality, or do justice to the productivity, learnability and systematicity of language.4 While these matters are no doubt pressing, I shall address an objection which is in many ways more fundamental and applies equally to any use-theory whatever its details. The objection is simply that the notions of meaning and use are not equivalent.5 Against this claim, I shall argue that, once one specifies the relevant use to be identified with meaning, the two do not come apart. Of course, this response concedes a certain amount of ground to the critic; it admits that 2. For present purposes, this diagnosis will suffice, although the relationship between usebased and truth-theoretic approaches to meaning is somewhat more complex than one of mere opposition. 3. Contrast, for example, the very different accounts in Brandom 1994 and Horwich 2005. 4. For critical overviews of such issues, see Lepore 1994 and Whiting 2006a. 5. To say that the notions of meaning and use are equivalent is, for present purposes, to say that they are co-extensive. Certain proponents or detractors of the claim that meaning is use might have a stronger reading in mind than this. Nevertheless, for the most part I shall stick to this weaker interpretation since, if it is false, any stronger version will also be false.

136

there are cases where the use of ‘use’ and of ‘meaning’ diverge. But, surely, it is never the view of the use-theorist that the terms are interchangeable in all contexts; this is obviously false. One talks of the meaning but not the use of life, and one talks of the use but not the meaning of a hammer. Rather, it is the more modest but potentially illuminating claim that there is a use of ‘meaning’—one which calls for philosophical investigation—which coincides with one of ‘use’. Significantly, it turns out that distinguishing the relevant kind of use to be identified with meaning takes us full circle; that is, it requires employing semantic notions of the same general sort as meaning (if not meaning itself ). While this is quite a different problem to that with which we started, and which is the main concern of this paper, I nevertheless offer some remarks intended to dampen, if not extinguish, its threat. Note that I do not propose, in this paper, to argue for the view that meaning is use. The primary aim is only to show that the prominent, putative counterexamples to that claim are unsuccessful. Needless to say, there might be more. It would be tedious, however, to anticipate and assess each (actual and possible) case in turn. Instead, I shall present particular responses to certain sorts of counterexamples with the hope that it is clear how they might generalize. 2. A preliminary sketch Use is a nebulous notion. Although for present purposes I need not and should not commit to a particular, full-blown account of meaning, some preliminary remarks are in order regarding how the appeal to use is to be understood. While remaining as neutral as possible on the finer details, this will provide sufficient substance to the dictum to begin a critical investigation of the purported similarities and differences between meaning and use. While there might be aspects of what is said that certain use-theorists will object to, I am confident that a representative number of philosophers would accept the sketch that follows. To begin with, one must distinguish an occasion of use from the way of using an expression. By way of illustration, imagine John and Jill fall from a boat into a stormy sea. John utters ‘Please help me’ and Jill utters ‘Please help me’. In this scenario, there are two particular utterances of the word ‘help’, that is, there are two occasions of use. Crucially, however, there are not two meanings here—‘help’ means just the same on both occasions,

137

namely aid or assistance. Hence, one should not equate the meaning of a term with its use on an occasion. Notably, however, there is a clear sense in which Jill and John are using the word in just the same manner, that is, ‘help’ is being used in the same way by both. Since there is just one way of using the expression and one meaning, it would seem that one should equate the meaning of ‘help’ with the way of using it. Next, one must distinguish the actual from the proper use of a word. By ‘actual’ use, I mean how a person (or group of people) happens as a matter of fact to employ an expression and how she is (or they are) disposed to do so. By ‘proper’ use, I mean how a word is correctly employed, how it should, may or ought to be used.6 One might assume that meaning is obviously to be equated with actual use. Doing so, however, appears to deliver the wrong results; specifically, it leads to false attributions of meaning. Imagine that Bill habitually uses ‘refutes’ interchangeably with ‘denies’, and regularly makes transitions from sentences such as ‘Ali denies the theory’ to ‘Ali refutes the theory’. If one were to identify the meaning of ‘refute’ with how it is actually employed, it would seem that one would have to judge that it means denies, when in fact it means disproves. One might note, however, that Bill is using ‘refute’ incorrectly; he is not employing it as he should, i.e. as we use ‘disprove’. Hence, equating meaning with proper use entitles one to judge that, Bill’s actual employment notwithstanding, ‘refute’ means disprove. That meaning is an intrinsically normative notion is controversial.7 Engaging with the discussion surrounding this idea would take us beyond the scope of this paper. Fortunately, doing so is not necessary for present purposes. One can admit that prima facie the meaning of a term is its proper use and yet leave open the questions of whether appearances in this instance are deceptive, and whether this superficially normative notion can be reduced to more basic, non-normative notions. In summary, I suggested that, for the use-theorist, the meaning of a word is to be equated with the proper way of employing it, rather than its actual use or any occasion of its utterance. While this is already to go beyond the 6. In virtue of what a way of using is ‘proper’ is an issue that I cannot resolve here. The idea is, roughly, that linguistic practitioners implicitly or explicitly introduce standards against which the employment of expressions is checked, that is, the proprieties of use are instituted by the users themselves. 7. Kripke popularized the view that meaning is intrinsically normative (1982). For recent criticism, see Boghossian 2005; Glüer 1999; Hattiangadi 2006; Wikforss 2001. For recent defence, see Glock 2005; Whiting 2007b.

138

mere dictum that meaning is use, there remain a myriad of ways in which these ideas might be further unpacked. In this paper, I shall not explore such ways but instead, sticking with this rough-and-ready characterisation, assess the extent to which the notion thereby characterised diverges from that of meaning. 3. Use without meaning The first objection to consider is that there exist expressions that have a use but not a meaning (Lycan 2000: 95–7; Rundle 1990: 190–1).8 Examples include: um bee-bop-a-lula abracadabra tallyho Meaning and use cannot be identical if there can be the one without the other. While Glock (1996a: 207) grants this point—taking it to ‘refute’ the claim that meaning and use—I think it can be challenged. Evidently ‘um’ is used, perhaps when searching for the right word or to signal that one’s pause does not indicate that one’s utterance is complete. Nevertheless, there is no way in which ‘um’ is to be used. Should one litter one’s talk with instances of ‘um’, it would no doubt be irritating but one would not be making a linguistic mistake; one would not be conflicting with its correct use, as none has been laid down for it. Since it is the proper way of using an expression that is to be equated with meaning, and since there are no proprieties governing its use, ‘um’ does not present a counterexample to the equivalence claim (so interpreted). (Parallel remarks apply to ‘bee-bop-a-lula’.) The putative counterexample of ‘tallyho’ is more awkward. As Peter Hacker points out, it does ‘have a rule governed use and [is] not invoked merely for the sake of the jingle’ (2005: 155).9 Thus, Hacker concedes 8. A specific case in point is proper names, which have a use but according to a prominent view no meaning (see Lycan 2000: 94; Rundle 2001: 101). Adequately discussing the phenomenon of proper names is beyond the scope of this paper and I shall not address it here. 9. I attribute this view to Hacker alone although it is expressed in Baker and Hacker 2005. The chapter in which the issue is discussed is an addition to the revised edition, which Hacker

139

a difference ‘between the use of “the meaning of a word” and the use of “the use of a word”, even within the limitations specified’. Nevertheless, he seeks to preserve a link of sorts: ‘Tallyho’ does not readily lend itself to explanation in the form of ‘“Tallyho” means … ’, and one cannot say that the meaning of ‘Tallyho’ is … […] Does this mean that it has a use but no meaning? That would be too swift. For one would be loath to say that it is meaningless. We might say that it has meaning but not a meaning. (2005: 155) 10

Unlike Hacker, I find myself quite ready to judge that ‘tallyho’ is meaningless. Reluctance to do so perhaps arises because that phrase is typically used to contrast words such as ‘exercise’ or ‘piano’ with utter gibberish such as ‘shalbypombadoo’. Evidently, ‘tallyho’ is more akin to the former than the latter. Nonetheless, one can admit this much without admitting (as I doubt one usually would) that it has meaning. A different strategy would be to distinguish different kinds of proper use, equate meaning with one such kind and show that ‘tallyho’ has another. Consider an uncontroversial example of a meaningful word: ‘vixen’. There are certain transitions that one might make involving that term, e.g. from ‘There are vixens’ to ‘There are female foxes’, certain words with which one might combine it, e.g. ‘is a mammal’, and certain words with which one would not, e.g. ‘is a prime number’. Plausibly, ‘vixen’’s having the meaning it does is to be equated with is having this kind of intra-linguistic role. In contrast, ‘tallyho’ has only what one might dub a ‘perlocutionary’ function.11 While it might be uttered with the intention of achieving prepared independently. This is not to suggest, of course, that Baker would not have endorsed Hacker’s position. 10. Hacker also draws attention to words, such as prepositions, of which one would not say ‘The meaning of x is …’ but would readily say ‘The use of x is …’ (2005: 155). This, in Hacker’s view, demonstrates significant differences between the notions of use and of meaning. Interestingly, however, Hacker recognizes contexts in which such a form of words would be suitable, e.g. when providing a translation. One might comfortably say, for example, ‘The meaning of “es” is it’. That one would not readily do so for such an expression in one’s own language surely has more to do with the availability of terms in that language that could be used to give the meaning of the relevant word than with differences between meaning and use. Specifications of meaning of the above form function by displaying a (familiar) term or terms with the relevant meaning (or usage). Words like ‘it’ typically have no counterparts within the language and cannot be decomposed. Thus, sentences of the form, ‘The meaning of “it” is …’ are not ill-formed but of little utility in conveying how an expression is to be understood. 11. The term ‘perlocutionary’ is borrowed, more or less faithfully, from Austin 1976: 100–1. Incidentally, Austin introduces the term in order to distinguish ‘the different uses of

140

some extra-linguistic effect12—specifically that of bringing one’s audience to believe that a fox is sighted—there are no expressions with which one might combine ‘tallyho’ and no transitions to and from sentences in which it is used that one might make. Hence, while ‘tallyho’ has a use, even a proper use, it does not have the kind of proper use possessed by expressions that have meaning, and hence does not constitute a counterexample. (The same story can be told of ‘abracadabra’.) One might consider this response ad hoc. However, the addendum was not introduced only to avoid an otherwise damning objection; it was independently motivated by reflection on the distinctive role of ‘tallyho’ and the respects in which it differs from that distinctive of meaningful expressions. Thus, one can view that objection as providing opportunity to refine how ‘use’ in this context is to be construed. Although it clearly shows that ‘use’ is not in all its applications interchangeable with ‘meaning’, it does not show that there is not a sense of ‘use’ in which it is interchangeable with ‘meaning’.13 4. Differences in use not differences in meaning Regarding the view that the meaning of an expression is its use, Bede Rundle further objects that ‘when, by whom, and to what end’ an expression is used can change without it changing meaning (1979: 384-5; 1990: 9; cf. Katz 1990: 40). That is, words can differ in use in the above respects without differing in meaning. Therefore, meaning cannot be use but, in Hilary Putnam’s words, ‘a coarse grid laid over’ it (1978: 99). Consider, for example, the following scenario. Abraham Lincoln in 1863 in Gettysburg utters, ‘Be sure to walk on the sidewalk’. Gordon Brown in 2007 in London utters, ‘Be sure to walk on the pavement’. Evidently here are differences in when, where and by whom ‘pavement’ and ‘sidewalk’ are used, but nonetheless they mean the same thing. The change in use does not result in a change in meaning. the expression’ ‘use’, which in his view is ‘a hopelessly ambiguous or wide word, just as is the word “meaning”’. 12. Here, ‘extra-linguistic effect’ excludes what Austin calls ‘uptake’, i.e. the mere recognition of which act is being performed (see Hornsby 2006: 900). 13. Recently, opponents of the appeal to use have presented derogatory expressions as a converse counterexample, i.e. a case of words with a meaning but no use (see Hornsby 2001; Williamson 2003). I shall not assess such claims here but do so in Whiting 2007a.

141

As both Hans-Johann Glock (1996b: 207) and Hacker (2005: 153) astutely point out, this criticism fails to distinguish, in the terms given above, between an occasion of use and a way of using. The above differences in speaker, geographical location and date are all differences in the former not the latter; the occasion of use changes but the way of using remains constant. Since it is the latter that is identified with meaning, the above does not constitute a counterexample. Rundle, however, has a reply (2001: 102–3). Words such as the German articles ‘die’ and ‘das’ differ in proper usage—the first is to be coupled with feminine nouns, the other not—but they do not differ in meaning—each means the. In response, one must point out that meaning is to be equated only with a specific kind of proper use and, with respect to that kind, ‘die’ and ‘das’ do not differ (even though they do in other respects). But what kind of proper use is that? I am not sure there is much one can say that is informative here. It is the kind a difference in which bears on what might be said by utterances of sentences in which the relevant expressions occur. A person might, for example, use either ‘Das Gespräch ist interessant’ or ‘Die Unterhaltung ist interessant’ to say that the conversation is interesting. More straightforwardly, one can individuate the use as the kind that is constant between a word and its translation. While the objection points to a difference in use that does not result in a difference in meaning, it is not of the relevant kind. The undeniable divergence in how ‘das’ and ‘die’ are properly employed does not show that their use is not the same according to less fine-grained criteria. One can therefore continue to treat meaning as use so long as one identifies the specific sort of use, and the distinctive proprieties governing it. That said, making the relevant distinctions in this instance evidently requires drawing on semantic concepts of the same sort as that of meaning. I shall return to address this below (§ 6). 5. Features of use not features of meaning A further objection Rundle raises (1990: 9; 2001: 100), and Glock elaborates (1996b: 208), is that there are features of use that are not features of meaning. Specifically, a use but not a meaning might be ill-advised, unjustified, encouraged, prohibited, accompanied by gestures, revealing, widespread, misleading, dying out or occasion disputes. Use and

142

meaning clearly cannot be the same if one can have properties the other cannot. Surely, however, the way a word is to be used cannot be fashionable, although actually uttering an expression in that way can. To say that the use of an expression is fashionable to say that there is a tendency to use a word in a certain way. Such a trend can increase or decrease as fickle attitudes change, but the way a word is to be used does not admit of degrees. The same can be said of the other features listed above. All are features of an occasion of use, or a tendency of use, rather than of the proper way of using an expression. It is the latter, however, that is equated with meaning.14 A remaining case, which Glock raises, is less easily dealt with. The way an expression is to be used might involve gestures, but a meaning could not. Here one must, again, distinguish the way of using equated with meaning from others. One might single it out as that which is conveyed by a definition, is constant between an expression and its translation, or one must grasp if one is to understand it. That way of using does not involve gestures. Hence, while there are ways an expression is to be employed that involve gestures, there remains a sort that does not. It is that sort which is to be identified with meaning. Notably, some of the notions by which the intended use is picked out are semantic notions on a par with that of meaning. For this reason, one might not be prepared to countenance some of the moves made here. In the next section, I shall respond to this concern. 6. Circularity In a rather different context, Donald Davidson writes: It is empty to say meaning is use unless we specify what use we have in mind, and when we do specify, in a way that helps with meaning, we find ourselves going in a circle. (2005: 13; cf. Rundle 2001: 101)

One might think that the preceding discussion only supports such a contention. Throughout, talk of use has been presented as talk of correct 14. Another case is that a use might be ungrammatical while a meaning could not be. However, this point does not apply since, as Hacker says, talk of the use of a word in this context ‘is intended to select something normative. It is use that accords with what is regarded as correct explanation’ (2005: 153).

143

employment. But, intuitively, the ‘correct’ use of a word is just that which accords with meaning. And in further restricting the notion of proper use so as to ensure it coincides with that of meaning, I appealed to notions such as translation, definition, understanding, and what is said. Surely, one might argue, for a theory of meaning to appeal to such semantic notions is circular. In response, there are a number of points to note. First, even if it is true that the relevant sense of use can only be distinguished by appeal to notions of the same kind as that of meaning, this does not show that the notion of meaning itself must play such a role. Instead, one might pick out the distinctive use via notions such as translation or understanding, which might be intelligible independently of that of meaning, if not more basic. One would need to provide an argument as to why this is an illegitimate strategy. Second, even if ultimately the notion of use can only be distinguished by employing the notion of meaning itself, that would not as such undermine the appeal to use. An account might be circular without being viciously so. Indeed, it is far from obvious that philosophical illumination of a given phenomenon can only be achieved when it is couched in independently intelligible and more fundamental terms. Even if it should turn out that grasping the relevant sense of ‘use’ requires a prior or simultaneous grasp of the concept of meaning, talk of use might nonetheless be more perspicuous than talk of meaning and so genuinely contribute to revealing the latter to be epistemologically and ontologically unproblematic. Moreover, by suitably connecting the ‘meaning-laden’ notion of proper usage to other semantic notions, and more widely to the style of intentional and psychological explanation to which it is intimately related, one might find considerable room to maneuver and, thereby, scope for illumination.15 Needless to say, there is much hand-waving here and nothing has been done to show that such a non-reductionist approach would indeed bear fruit.16 The aim, however, was only to take some of the sting out of a seemingly urgent worry. At the least, it should not be accepted without 15. These sketchy remarks are supposed to gesture to the mode of analysis that Strawson dubs ‘connective analysis’ (1992: ch. 2). 16. For more extensive remarks on why a reductionist account of meaning in independently intelligible or more basic terms is not required (and perhaps not feasible), see Whiting 2006b. For a recent and excellent example of how much work can be achieved by a non-reductionist, use-based approach to meaning, see Alston 2000.

144

question that the possibility that, on inspection, the claim that meaning is use takes us full circle counts against it. Finally, and most importantly for present purposes, should the circularity objection stick, it provides no support for those who deny that meaning is use. On the contrary, only if the two notions overlap in the relevant respects, only if they are too close for comfort, so to speak, could the circularity objection get off the ground. Since this paper’s primary aim is to deny that meaning and use diverge, the circularity objection is beside the point. 7. Conclusion Rundle tells us that ‘you cannot readily say that the meaning of a word just is its use’ (2001: 103). Likewise, Glock remarks that ‘the notions of meaning and rule-guided use … diverge in important respects’ (1996a: 378). Again, Hacker speakers of the ‘failure to plot the ragged contour lines of the concept of meaning within the scope of the concept of the rule-governed use of a word’ (2005: 158). I have tried to show that such judgements are unwarranted. There might, of course, be further examples that demonstrate a divergence between meaning and use but those so far considered have not done so. Certainly, the cases examined establish that the notion of meaning equates only to a specific notion of use but, I argued, this does not establish that it does not equate to any notion of use. Picking out the relevant use does, however, appear to require appeal to concepts of the same kind as that of meaning. Nevertheless, this does not show that in principle the appeal to use is of no philosophical worth. Whether or not in fact it is remains to be seen.17

17. I am indebted to the Arts and Humanities Research Council for funding that made writing this paper possible. Versions of it were presented, in various circumstances, at Lancaster, Reading, Southampton and Stirling. I am grateful to the audiences and respondents on those occasions for valuable feedback. Thanks also to Emma Borg, Hanjo Glock, Javier Kalhat, Adrian Moore, Bede Rundle and Galen Strawson for helpful discussion of earlier incarnations of the material, and to an anonymous referee for suggesting changes that greatly improved the paper.

145

BIBLIOGRAPHY Alston, W. 2000: Illocutionary Acts and Sentence Meaning. New York: Cornell University Press. Austin, J. L. 1976: How to Do Things with Words 2nd ed. Oxford: Oxford University Press. Baker, G. P. and Hacker, P. M. S. 2005: Wittgenstein: Understanding and Meaning Rev. ed. Oxford: Blackwell. Boghossian, P. 2005: “Is Meaning Normative?” Philosophy—Science—Scientific Philosophy A. Beckermann and C. Nimtz, eds. Paderborn: Mentis. Borg, E. 2004: Minimal Semantics. Oxford: Oxford University Press. Brandom, R. B. 1994: Making it Explicit: Reasoning, Representing, and Discursive Practice. Cambridge, Massachusetts: Harvard University Press. Davidson, D. 2005: Truth, Language, and History. Oxford: Oxford University Press. Glock, H-J. 1996a: A Wittgenstein Dictionary. Oxford: Blackwell. — 1996b: “Abusing Use.” Dialectica 50: 205–233. — 2005: “The Normativity of Meaning Made Simple.” Philosophy—Science—Scientific Philosophy A. Beckermann and C. Nimtz, eds. Paderborn: Mentis. Glüer, K. 1999: “Sense and Prescriptivity.” Acta Analytica 14: 111–128. Hattiangadi, A. 2006: “Is Meaning Normative?” Mind and Language 21: 220–240. Higginbotham, J. 2006: “Truth and Reference as the Basis of Meaning.” The Blackwell Guide to the Philosophy of Language M. Devitt and R. Hanley, eds. Oxford: Blackwell. Hornsby, J. 2001: “Meaning and Uselessness: How to Think about Derogatory Words.” Midwest Studies in Philosophy (25) P. French and H. Wettstein, eds. Oxford: Blackwell. — 2006: “Speech Acts and Performatives.” The Oxford Handbook of Philosophy of Language E. Lepore and B. Smith, eds. Oxford: Oxford University Press. Horwich, P. 2005: Reflections on Meaning. Oxford: Oxford University Press. Katz, J 1990: The Metaphysics of Meaning. Cambridge, Massachusetts: MIT Press. Kripke, S. 1982: Wittgenstein on Rules and Private Language. Oxford: Blackwell. Lepore, E. 1994: “Conceptual Role Semantics.” A Companion to the Philosophy of Mind S. Guttenplan, ed. Oxford: Blackwell. Lewis, D. 1975: “Languages and Language.” Language, Mind and Knowledge K. Gunderson, ed. Minneapolis: University of Minnesota Press. Loar, B. 2006: ‘Language, Thought, and Meaning.’ The Blackwell Guide to the Philosophy of Language M. Devitt and R. Hanley, eds. Oxford: Blackwell.

146

Lycan, W. 2000: Philosophy of Language. London: Routledge. Putnam, H. 1978: Meaning and the Moral Sciences. London: Routledge. Rundle, B. 1979: Grammar in Philosophy. Oxford: Oxford University Press. — 1990: Wittgenstein and Contemporary Philosophy of Language. Oxford: Blackwell. — 2001: “Meaning and Understanding.” Wittgenstein: A Critical Reader H-J. Glock, ed. Oxford: Blackwell. Strawson, P. F. 1992: Analysis and Metaphysics. Oxford: Oxford University Press. — 2004: Logico-Linguistic Papers New ed. Aldershot: Ashgate. Whiting, D. 2006a: “Conceptual Role Semantics.” The Internet Encyclopedia of Philosophy J. Fieser and B. Dowden, ed. http://www.iep.utm.edu/c/conc-rol. htm — 2006b: “Between Primitivism and Naturalism.” Acta Analytica 21: 3–22. — 2007a: “Inferentialism, Representationalism and Derogatory Words.” International Journal of Philosophical Studies 15: 191–205. — 2007b: “The Normativity of Meaning Defended.” Analysis 67: 133–140. Wikforss, A. 2001: “Semantic Normativity.” Philosophical Studies 102: 203–226. Williamson, T. 2003: “Understanding and Inference.” Aristotelian Society Supplementary Volume 77: 249–293. Wittgenstein, L. 1967: Philosophical Investigations 3rd ed. G. E. M. Anscombe, R. Rhees, and G. H. von Wright, eds. G. E. M. Anscombe, trans. Oxford: Blackwell. — 1982: Last Writings on the Philosophy of Psychology: Volume I G. H. von Wright and H. Nyman, eds. C. Luckhardt and M. Aue, trans. Oxford: Blackwell.

147

Grazer Philosophische Studien 76 (2008), 149–166.

THE FAILURE OF PURE COGNITIVISM Achim LOHMAR Universität zu Köln

Summary According to Humeanism, actions cannot be adequately explained without reference to the desires of an agent. Desires are viewed as sources of motivation or as motivating states and thus as having an indispensable role to play in the explanation of actions. One of the main rivals of Humeanism is pure cognitivsm. According to this view, actions are to be explained exclusively by beliefs. The present paper’s focus is on arguments Jonathan Dancy has put forward in favor of this pure cognitivist picture. His main line of argument tries to convince us of the claim that desires have no explanatory value at all as regards the explantion of an agent’s actions. I argue that none of Dancy’s arguments against Humeanism is successful, and moreover that the pure cognitivists position fails on its own terms because pure cognitivism is unable to provide an account of desires that makes intelligible their role in the mental economy of agents.

In recent decades Humeans have witnessed the emergence of alternative accounts of motivation and action explanation which call into question either the explanatory value of Humeanism’s desire/belief model or the adequacy of the whole picture of rational agents Humeanism provides. Not all of the alternative accounts are unfriendly to Humeanism.1 But some call into question its very core ideas and think of Humeanism as ultimately misguided. Arguably the most radical departure from Humeanism arises from accounts according to which those things that explain actions are not mental states at all but external facts of the world.2 Other accounts stick to a psychological picture but dispute the view of desires 1. See, for example, Frederick Schick, Understanding Action. An Essay on Reasons. Cambridge 1991, who argues that we need in addition to desires and beliefs ‘understandings’ to account for an agent’s reasons. 2. For anti-psychological accounts of reasons see Jonathan Dancy’s Practical Reality, Oxford 2000, or Rüdiger Bittner’s Doing Things for Reasons, Oxford 2001.

as being the only sources of motivation.3 And then there is pure cognitivism—a view according to which beliefs and only beliefs are motivating states. Pure cognitivism (PC) is the target of the present paper. In the first section I lay open a questionable intuition that drives the move to PC. The three sections to follow are devoted to a close examination of arguments Jonathan Dancy has put forward to challenge Humeanism. I shall argue that none of these arguments establish that there is any problem with Humeanism. This is a crucial result because the case Dancy makes for PC relies on the purely negative claim that desires do not motivate. In the last section I will argue that PC fails for another reason, namely its being unable to provide an intelligible account of the role desires play in the mental economy of agents. 1. Desires in need of explanation? Those who think of the desire/belief model as unsatisfactory typically raise a kind of objection along the following lines. If, in explaining an action, we always have to refer to a desire of the agent in the last resort, then every explanation of an agent’s actions has to end in citing a factum brutum about his motivational make-up. But if we cannot but stop with citing facta bruta about the motivational make-up of an agent, we are bound to fail to give an explanation which presents the action as somehow motivated. A suspicion of this kind is voiced by Dancy when he writes: Some desires, of course, cannot be explained. But if they cannot be explained, then neither can the action that, in desiring as we do, we are motivated to perform. If we cannot say why we want to do it, the fact that we want to do it offers nothing by way of explanation for the action. It merely means that we were, incomprehensibly, motivated to do this incomprehensible thing.4

The point Dancy obviously wants to stress is that in order to make an action intelligible we have to make intelligible the desires an agent has when he acts. Otherwise we are left alone with a totally incomprehensible thing—an action we simply cannot make any sense of. The explanation of desires we need to make an action intelligible cannot be provided by a causal story which explains how it comes about that the agent has a desire 3. See Thomas Nagel, The Possibility of Altruism, Princeton 1970. 4. Dancy, Practical Reality, 85/6 (italics mine). All citations from the paperback edition, Oxford 2004.

150

of this or that sort. It has to be a motivational explanation. That is, in order to answer the question why A bought a new car we have to answer the question of why he wanted to buy a new car, where this question asks for the motivating reasons for his desire to have a new car. Thus, Dancy supposes that the “explanation of motivation must be structurally similar to the explanation of action.” (85) This request makes the account provided by Humeanism initially suspect because Humeanism explains actions in terms of desires and beliefs without offering a structurally similar explanation of the desires of an agent. But Humeans need not be bothered by this. For the intuition underlying the alleged constraint on motivational explanations obviously plays on an ambiguity of the term ‘unmotivated’. Sometimes we use this locution to mean that there are no good reasons to behave in a certain way. That is, there is a normative use of ‘unmotivated’ which comes into play when, by calling an action ‘unmotivated’, we want to express our view that it is a stupid thing to do, or that it is out of order, or something like that. But ‘unmotivated’ can simply mean that there is no motivation—and thus no intentional explanation—making the behaviour of an agent an action of his. Now, the normative use of ‘unmotivated’ obviously presupposes the motivatedness of a piece of conduct. A faucial reflex, for example, can neither be deemed to be unmotivated nor to be motivated (in the normative sense) precisely because a faucial reflex is not an intentional action at all. Performing a headstand, on the other hand, may be a very strange thing to do in many circumstances, but it does not follow from its sometimes being very strange that performing a headstand sometimes just happens to an agent in an incomprehensible way. Even if an agent just happens to have certain desires, it does not follow, therefore, that what he is motivated to do by such desires just happens to him and cannot be understood. Thus, in order to explain an action one need not explain the states which figure in one’s explanation.5 This is but an instance of a truth about explanations in general. Continental drift, for example, explains earthquakes regardless of whether we have any idea of what explains continental drift; the moon’s gravity explains tidal phenomena on earth even if gravity itself cannot be further explained. Likewise, one can understand or ‘rationalize’ an agent’s conduct in terms of his desires (and beliefs) even if there is no motivational 5. To think otherwise is erroneously to think that understanding is, in Peter Liptons phrase, “like a substance that the explanation has to possess in order to pass it on to the phenomenon to be explained.” See Peter Lipton, Inference to the Best Explanation, 2nd edition, London 2004, 22.

151

explanation which shows his desires to be somehow motivated. Why an agent has certain desires (and beliefs) may of course be an interesting question in some contexts of inquiry.6 But it is a question of its own, and not a question one is forced to ask if one wants explain actions. So far I have rejected the suspicion that Humeanism’s desire/belief model cannot provide the right story about the explanation of action as unwarranted by arguing, first, that an action can be explained by desires even if the action is to be deemed ‘unmotivated’ (where this use of ‘unmotivated’ is a normative one), and by stressing, second, that action explanations behave like other explanations and do not require the explanans itself to be understood. To cling to the line of objection voiced by Dancy, and to substantiate the thesis that desires cannot explain actions unless they themselves can be explained in terms of motivating reasons, one need to introduce a special constraint on action explanations. Such a constraint must meet two conditions as is obvious from our previous discussion. The first condition is that such a constraint must not presuppose or entail that action explanations are anomalous explanations in requiring the explanation of actions to be explained. That is, the constraint to be introduced must not presuppose or entail that, in the special context of action explanation, a regress of ‘Why?’ questions is generated. The second condition is that the constraint to be introduced must express the idea that there is an interesting link between explanatory and normative talk about an action’s being motivated or comprehensible. That is, the constraint must be designed so as to undo the objection that to call actions explained in terms of desires ‘unmotivated’ or ‘unintelligible’ is but to play on an ambiguity. The special constraint on action explanations then turns out to be a condition of adequacy which can be expressed along the following lines: To be adequate an account of action explanations has to ensure that whatever is designed to do the explanatory work can also be what grounds an assessment of the action so explained as motivated or as comprehensible. 6. We may be interested, for example, in whether an agent is morally responsible for what he has done, and in this context it may be of great interest to explain why an agent has a certain desire. But note that if we take an interest in this question we already take it for granted that there is an intentional explanation for what the agent has done and that a certain desire of his figures as part of this very explanation.

152

This condition clearly captures the intuition that moves and underlies the suspicion that Humeanism must be somehow misguided because it refers to desires as brute psychological facts.7 Against the background of this constraint on action theories we can spell out Dancy’s critical idea as follows: Humeanism is committed to explain actions by brute psychological facts; brute facts about an agent’s motivational make-up cannot, however, ground an assessment of actions as motivated and thus cannot make actions intelligible or comprehensible. Whether or not the mentioned constraint on action explanations can be justified, I will not, and need not, discuss here. For it seems obvious to me that Dancy’s critical comment on the desire theory of motivation can not be warranted even if we accept the constraint. This is because the Humean view according to which desires are, and beliefs are not, sources of motivation does not commit one to take desires as brute facts in any objectionable sense which is strong enough to declare the Humean theory misguided. This would be the case only if Humeanism implied that it is impossible for motivating states to be critically assessed. But there is certainly no reason to think that Humeanism does imply that.8 That desires are, and beliefs are not, motivating states, does not imply that no distinction can be drawn between, for example, reasonable and unreasonable desires. Even if the desire for self-preservation, for instance, were a basic desire of human beings, it does not follow that it is a brute desire the having of which we cannot make any sense of. To lack this desire may be, on the contrary, a sign of a severely unhealthy state of mind. To be deeply concerned about the well-being of their own children may also be a basic motivational fact about the psychology of most parents. But, again, it does not follow from this that most parents are incomprehensibly motivated to 7. I have choosen this formulation in order to capture the idea underlying the criticism suggested by the passage cited above. Dancy himself expresses the idea that an explanation must be capable of grounding assessments of actions in order to count as an action explanation at all as follows: “[…] there is a constraint on any theory about the relation between normative and motivating reasons. This is that the theory show [sic] that and how any normative reason is capable of contributing to the explanation of an action that is done for that reason. Call this the ‘explanatory constraint’.” (101) 8. Michael Smith has offered a theory about practical reasons that combines a Humean theory of motivation with an anti-Humean view about what he calls normative reasons. Whether or not one agrees with Smith’s account of justifying reasons and rationality—the combination of his theory of motivation and justification seems not to be an incoherent one. If I have normative reason for ϕ-ing in situation C iff I would desire to ϕ in C if I were fully rational, then my ϕ-ing in C may be done by me for a normative reason. And thus, Dancy’s ‘explanatory constraint’ is satisfied by Smith’s theory. See Michael Smith, The Moral Problem, Oxford 1994.

153

do a lot of incomprehensible things. Both, the desire for self-preservation and the desire that one’s own children are living well, are certainly among those candidates of motivational states that can reasonably be expected to survive critical reflection or to survive what Richard Brandt has called ‘cognitive psychotherapy’.9 Now, if a Humean theory of motivation is compatible with there being reasonable and unreasonble desires, then there is pretty much room left for distinguishing between actions that are comprehensible and those that are not. If we explain an action by citing a reasonable desire and a justified belief we offer an explanation in the light of which the action so explained is far from being incomprehensible. A constraint that requires that some actions, if explained along the lines of a theory’s favoured model of motivation, must come out as comprehensible one’s will not suffice, therefore, to shake off Humeanism. Anyway, the principal point of Dancy’s criticism actually does not rest on the suspicion that a Humean theory of motivation is not suitable to the possibility of critical assessment of an agent’s conduct. His main arguments which purport to show that pure cognitivism wins the race, if a psychological account of practical reasons is viabel at all, dispute the explanatory value of desires. To these arguments I turn now. 2. Dancy’s overall argument for Pure Cognitivism In the preceding section we have met with a suspicion that motivates the move away from Humeanism. But what is the argument in favour of PC? When Dancy complains that action explanations in terms of desires do not really explain until the desires are explained themselves he does not state requirements of explanations in general. That is, Dancy is not engaged in a kind of argument which says that there is no good explanation of any event or fact unless we come to the point of a necessarily unexplainable explainer. Dancy’s point is, or must be, the rather special one that desires cannot explain why an agent does something. Thus, his argument in favour 9. Philosophers whose views about practical reasoning belong to a Humean or ‘sentimentalist’ tradition will not allow, however, that the possibility of critical reflection on practical stances, attitudes and desires implies that our practical stances are grounded in and ruled by our cognitive capacities. The very possibility of critical reflection on one’s own desires may in fact hinge upon there being practical attitudes which in the course of critical reflection on some desires or stances cannot itself be subjected to it. This is a major theme in Simon Blackburn’s Ruling Passions. A Theory of Practical Reasoning, Oxford 1998. See especially ch. 8.

154

of PC must in the main rest on a particular conception of desires which, in turn, entails that offering an action explanation in terms of an agent’s desires cannot show the action as somehow motivated. That is, Dancy can be interpreted as in the end trying to lay open that Humeanism’s account of action explanation cannot satisfy what he has called the ‘explanatory constraint’10 by trying to show that desires, properly understood, lack explanatory value and cannot, therefore, account for an action’s being motivated or comprehensible. That the desire theory of motivation does not satisfy the ‘explanatory constraint’ would be a trivial result, however, given that desires have no explanatory value at all. The ‘explanatory constraint’ thus plays no substantial role in Dancy’s argument, although it may be viewed as a background assumption which moves Dancy’s effort to show that desires have no substantial role to play in the context of action explanations. According to Dancy, the principal or original fault of the desire theory of motivation is that it misconstrues the nature of desires. Desires, Dancy tells us, are not motivating mental states but states of being motivated. And this is why a reference to a desire of an agent cannot be part of an explanation of his action, given that a motivational explanation is nothing but an explanation of why an agent is in a state of motivatedness regarding a certain action. Thus, Dancy’s main argument in favour of PC can be reconstructed as follows. It involves three steps. The first step is an argument which purports to show that desires do not motivate; the second step argues for the claim that desires cannot explain actions; the third step consists in an argument which tells us that only beliefs motivate and explain actions. The first step goes as follows: (1) A state of being motivated is not a motivating state. (2) Desires are states of being motivated. Therefore: (3) Desires are not motivating states. The second step takes the conclusion of the preceding argument as a premiss and proceeds to state the explanatory vacuousness of desires: (3) Desires are not motivating states. (4) Only motivating states explain actions. 10. See fn. 7.

155

Therefore: (5) Desires do not explain actions. Now, if we take desires to be a broad category of the mental which covers conative states in general, then only cognitive states—beliefs—can explain actions: (5) Desires do not explain actions. (6) What motivates and explains actions are either desires or beliefs. Therefore: (7) Only beliefs motivate and explain actions. The crucial and most interesting step is obviously the first one. If the first step is granted the two other steps follow smoothly (unless, of course, one wants to reject a psychological account of those things that motivate us). In what follows I will concentrate on the first step. Until now, it must be stressed, we have neither seen a reason for the claim that a state of being motivated cannot be a state that motivates, nor have we seen a reason for the claim that desires are not motivating states. All we have heard is a certain suspicion which may be taken as a hint as to problems for the desire account of motivation. So, what, after all, are Dancy’s reasons for believing the premisses of the first argument? 3. Buck-passing explantions? Why should we rule out that a desire can be both, a state that motivates and a state of being motivated? In other words, why should it be incoherent to say of an agent A that his desire to ϕ motivates him to ϕ and that his being motivated by his desire to ϕ is his being in a state of motivatedness directed at ϕ ? Dancy’s reasoning is at follows. He thinks that it can be shown that desires do not motivate. But from this he does not want to draw the conclusion that desires play no role at all in the mental economy of an agent. This is because he takes it to be obvious that if there is action then there is desire. Thus, he is forced to construe desires as states of being motivated, and states of being motivated, in turn, as not to be states that motivate. But why are desires not to be viewed as motivating states? Dancy gives two arguments. The first can be called the argument from vacuousness:

156

Suppose […] that we are explaining an instance of A’s ϕ-ing by appeal to the fact that A desired to ϕ, or to A’s desire to ϕ. This is what one might call a ‘buck-passing explanation’. No explanation has been given unless A’s desire to ϕ can be explained. […] What motivates [A’s ϕ-ing] must therefore be that which underpins the desire [A’s desire to ϕ]. But the only thing that can underpin the desire and explain it appropriately is the nature of what is desired.

As I have said above, the argument from vacuousness cannot be taken as an instance of a general constraint on explanations according to which no fact is explained unless the explainer of the fact is explained. Otherwise, if Humeanism fails for such a reason, then PC would fail as well. Why, then, has no explanation of A’s ϕ-ing been given unless an explanation of A’s desire to ϕ has been given? Dancy seems to think that such explanations are utterly vacuous because the desire to ϕ has ϕ-ing as its intentional object. That is, he seems to think them utterly vacuous because the content of the desire appealed to is the explanandum itself.11 But why should it follow from this that the desire cannot explain the action or cannot be what motivates the action? Two points are to be made here. First, if explanations along the considered line were vacuous, so were explanations appealing to beliefs. Secondly, explanations along the considered line are or can be informative. Both points can be substantiated if we reflect on the following two explanations: (a) Jasper plays piano because he just wants to play piano. (b) Jasper plays piano because he believes it is a good thing to play piano. 11. Anscombe can be taken as holding a similar view when she stresses (Intention, 2nd ed. Oxford 1963, § 12) “that mental causes are seldom more than a very trivial item among the things that it would be reasonable to consider”—i. e. reasonable to consider in the context of a ‘Why?’ question that asks for a man’s reasons or motives for doing something. Anscombe considers answers to the ‘Why?’ question which state that an agent has a certain desire, or that he wanted to do something, as not at all being answers which purport to present his motive or reason, but as answers that merely report what was going on in the agent’s mind. Thus, desiring something, or wanting to do something, comes out, in Anscombe’s view, as belonging to the causal history of actions, but not as belonging to whatever motivates an agent to do something. As regards ‘Why?’ questions that call for motives, an appeal to desires cannot, in her view, provide any substantial information. This view, however, does not follow from her classification of desires as mental causes but, contrariwise, the classification of desires as mental causes (which are not motives) is at least partly forced on Anscombe because of her conviction that answers to ‘Why?’ questions in terms of desires are utterly uninformative.

157

I suppose that if a feeling of vacuousness arouses if you consider (a), then a feeling of vacuousness will or should also arouse if you consider (b). If no explanation has been offered by (a) unless we add an explanation of why Jasper wants to play piano, then no explantion has been given by (b) unless we add an explanation of why Jasper believes it is a good thing to play piano. If this does not convince you, think of it the following way. If someone asks you why Jasper plays piano and is, for some reason or other, not satisfied with your answer that Jasper just wants to play piano, then he will also not be satisfied with the answer that Jasper believes it is a good thing to play piano. The upshot of this is that if an appeal to A’s desiring to ϕ is of no explanatory value it is not because it is an appeal to a desire of A but because the desire appealed to is directed at or specified by ϕ. Therefore, even if we grant Dancy that A’s desire to ϕ cannot explain why A is ϕ-ing, we need not grant him his (implicit) further claim that this explanation is ‘buck-passing’ because it appeals to a desire of A. But now suppose we would reject explanation (a) and say that it is not the case that Jasper plays the piano just because he wants to play it. If (a) were totally vacuous an explanation, a rejection of (a) couldn’t be informative either. If, as Dancy tells us, no explanation has been given by (a), then a rejection of (a) cannot be understood as a rejection of an explanation. The rejection, therefore, must be as uninformative as the rejected item itself. But, obviously, to say that Jasper does not play piano just because he wants to play piano, is to say something informative. It amounts to the claim that Jasper does not play piano for its own sake. Someone who claims that Jasper does not play piano just because he wants to play it, may thus wish to inform us of the fact that today Jasper plays piano because he wants to prepare for a concert which takes place tomorrow. That is, he may wish to point out that, indeed, Jasper often plays just because he has a desire to play but that, in the present circumstances, Jasper is motivated by the desire to be well prepared for a concert. But let us, for the sake of argument, even grant that explanations of the form “A ϕs because A has a desire to ϕ” are utterly vacuous and uninformative. This would amount to a case against Humeanism only if every explanation conforming to the desire/belief model is an instance of this ‘buck-passing’ scheme. But, obviously, as Dancy (as we shall see) himself agrees, Humeans are not committed to restrict action explanations in this way. Most Humeans will, in fact, point out that action explanations, fully spelled out, cite desires and instrumental beliefs about the means to realize

158

what the agent desires.12 To stick to our previous example, the desire to be well prepared for the concert next day and the belief that he needs practicing Bach’s Italian Concert in order to be well prepared for the concert next day, explain why Jasper plays Bach’s Italian Concert today. This explanation differs crucially from the hitherto considered ‘buck-passing’ scheme of action explanation in that the desire invoked here is not at all directed at the explanandum. That Jasper plays the Italian Concert today because he wants to be well prepared for a concert next day appears to convey substantial information, and here a feeling of vacuousness will certainly not crop up. Therefore, if Dancy’s line of argument cannot be broadened so as to meet explanations in terms of desires which are not specified by the action to be explained, no reason has been given to be suspicious of Humeanism, let alone to reject it. In the following section I show that Dancy’s attempt to broaden his line of argument fails. 4. Vacuousness all the way through? Dancy admits that not all explanations of actions in terms of desires are ‘buck-passing’. Thus, we may explain A’s φ-ing by appeal to his desire to ψ and his belief that by φ-ing he can fullfill his desire to ψ. Dancy wants to stress, however, that even this kind of explanation does not really tell us what motivates A to φ: None the less, what motivates A to φ cannot be the desire to ψ. If that desire cannot be what motivates the action of ψ-ing, it cannot be what motivates the action of ‘doing what promotes ψ-ing’, namely φ-ing. The relation of motivation, that is, of motivating, if it cannot hold between ψ-ing and the desire to ψ, cannot hold between ψ-ing and doing what promotes ψ-ing. (86)

Here Dancy relies on his previous claim (which we granted for the sake of argument) that a desire to φ cannot be what motivates φ-ing. If this claim is granted, then the general principle must be granted, too, that for 12. According to Michael Smith, the Humean theory of action explanation is not to be viewed as a theory about structural features exhibited by every explanation which is offered as an answer to the question as to why an agent has done something, but as a theory that purports to give a ‘constitutive answer’ about what explains actions. That is, to explain an action at full length is not only to make intelligible why an agent has done this or that specific thing but is, at the same time, to show that his behaviour is an action at all. See Michael Smith, “Humeanism, Psychologism, and the Normative Story”, in his Ethics and the A Priori. Selected Essays on Moral Psychology and Meta-Ethics, Cambridge 2004, 146–154.

159

every desire δ and every action α, if δ is directed at α, then α cannot be motivated by δ. But this principle by itself does not rule out that A’s φing can be motivated by his desire to ψ. It can be used, however, to rule out that A’s doing something that promotes ψ-ing can be motivated by his desire to ψ. Now, for his argument against Humeanism to work, Dancy has either to construe the explanandum (A’s φ-ing) as falling under the description “doing something that promotes ψ-ing”, or he has to construe Humeanism as committed to construe the explanandum (A’s φ-ing) as falling under the description “doing something that promotes ψ-ing”. Accordingly, we have two ways to reconstruct the argument of the abovementioned passage. On the first interpretation, Dancy offers the following argument: A’s desire to ψ cannot motivate A’s doing something that promotes ψ-ing. (9) A’s φ-ing = A’s doing something that promotes ψ-ing. Therefore: (10) A’s φ-ing cannot be motivated by A’s desire to ψ. (8)

Under this interpretation, we have Dancy simply stipulating (9). But why should anyone accept this claim? Dancy seems to think that if a desire to ψ were to motivate A’s φ-ing then φ-ing must conform to the description “doing something that promotes ψ-ing” because in order to motivate A’s φ-ing the desire to ψ must (in some way) be directed at φ-ing at least in an indirect way. And once φ-ing is identified with “doing something that promotes ψ-ing” it becomes obvious that an explanation of A’s φ-ing in terms of his desire to ψ is as vacuous as an explanation of φ-ing in terms of a desire to φ. Thus, Dancy construes Humeanism as committed to the following claim: (11) Necessarily, if A’s φ-ing is motivated by A’s desire to ψ, then A’s φ-ing is doing something that promotes ψ-ing. Now, the principle according to which an action α cannot be explained by a desire δ if δ is, even indirectly, directed at α, tells us: (12) Necessarily, if A’s φ-ing is doing something that promotes ψ-ing, then A’s φ-ing is not motivated by A’s desire to ψ.

160

(11) rules out that A’s φ-ing is motivated by A’s desire to ψ if φ-ing does not conform to the description “doing something that promotes ψ-ing”, and (12) rules out that A’s φ-ing is motivated by A’s desire to ψ if φ-ing does conform to the description “doing something that promotes ψ-ing”. Thus, from (11) and (12) it follows that it is impossible that A’s φ-ing is motivated by A’s desire to ψ: (13) A’s φ-ing cannot be motivated by A’s desire to ψ. Read this way, Dancy’s claim is that vacuousness is forced upon action explanations in terms of desires by the very requirements of such explanations themselves. A close inspection of Dancy’s argument against Humeanism reveals, however, that it does not succeed and cannot but miss its target. There are at least two points to be made here. The first is that Dancy misconstrues Humeanism in an unfair way, and that Humeans, as we shall see, are not in the least committed to accept proposition (11) which is certainly false.13 The second point is that Dancy’s unfair construction of Humeanism stems from a certain trick as regards the description of the action to be explained. Let me start with the second point. To work out an explanation of some item the target of the explanatory work must at first be fixed. And, moreover, the explanandum must be held fixed as the explanatory work proceeds until an explanatory proposal is offered. Thus, we take the fixity of the explanandum for granted whenever we evaluate rival explanations, because otherwise there simply are no rival explanations. And so for action explanations. To ask why A is φ-ing is to ask for an explanation of the fact that A is φ-ing. It is not to ask for an explanation of the fact that A is doing something that promotes ψ-ing (even if, by φ-ing, A is in fact doing something that promotes ψ-ing). And even if A in fact is doing something that promotes ψ-ing by φ-ing, this is just a further fact which is not guaranteed to be explained by the explanation of A’ s φ-ing. Consider, for example, Tom digging in his garden. And suppose there is a treasure buried in Tom’s garden just where Tom is digging. Given this information, one might claim that by digging Tom is doing something 13. Proposition (12) is, in my view, also false for reasons I have outlined already in the previous section. But I will not focus on (12) here because I have granted, for the sake of argument, that a desire which is directed at φ cannot be taken as a state that motivates φ-ing, and because (12) may be taken as a special case covered by this principle.

161

that promotes or contributes to his getting rich. But if there is no treasure in Toms garden, then Tom, by digging in his garden, does not do something that promotes or contributes to his getting rich. Whether or not by digging in his garden Tom is doing something that promotes his getting rich therefore fully depends on whether the conditions are such that Tom’s digging in his garden can lead to his becoming rich. Thus, it makes a great and deep difference whether our explanandum is the fact that Tom is digging, or whether it is “Tom’s doing something that promotes his getting rich”, or whether it is “by digging in his garden Tom is doing something that promotes his getting rich”. The last mentioned explanandum differs crucially from the other two in that it is not to be explained by facts about Tom’s mental states but by facts about the conditions under which he is digging. Thus, Dancy’s move to identify A’s φ-ing with A’s doing something that promotes ψ-ing is not only arbitrary. It is a wrong move because an explanation of the former is independent of an explanation of the latter and vice versa. The next step is to consider whether Humeans are committed to this wrong move of construing instances of A’s φ-ing as instances of A’s doing something that promotes ψ-ing. So, let us take the fact that Tom is digging, or simply Tom’s digging, as our explanandum. And suppose that Tom has a strong desire to get rich and that he believes that there is a treasure buried in his garden. Then Humeans will be able to give a substantial and very plausible explanation for Tom’s digging by appealing to this desire and this belief of his. And they might also add that because of his desire and his belief Tom has formed the intention to dig in the garden. But there is obviously not the slightest need for Humeans to construe either Tom’s desire to get rich or his intention to dig in the garden in another way, to wit, as the desire or the intention to do something that promotes his getting rich. Thus Humeans can explain A’s φ-ing by appeal to a desire δ that is not identical with A’s desire to φ without being committed to describe A’s φ-ing as “A’s doing something that contributes to the fullfillment of δ”. The same point can also be substantiated by another consideration. For suppose that in fact there is no treasure buried in Tom’s garden. Then, other things being equal, it is simply wrong to say that by digging Tom is doing something that promotes his getting rich. To claim that Humeans are committed to construe Tom’s digging as Tom’s doing something that promotes his getting rich is then to claim that Humeans are committed to descriptions of an agents actions which does not fit the facts. But this claim is simply absurd. Humeanism is a theory about motivation and

162

action explanation which has no implications whatever as regards the description of an agent’s actions. 5. The failure of Pure Cognitivism In the foregoing sections we have reconstructed the arguments Dancy has designed to reject Humeanism. A closer examination of these arguments has shown that they either rest on false premisses or on arbitrary constructions of Humeanism’s commitments. Thus, Dancy hasn’t offered us any reason against the view that desires are motivating states. This result is of highest importance because Dancy’s argument in favour of PC totally relies on the negative claim that desires do not motivate. The route to PC is therefore blocked. In addition to this crucial result I now want to argue that PC fails for another reason. The reason is that PC is not able to provide a plausible and coherent account of the role of desires. To account for the place of desires in his overall picture Dancy stresses two points. He tells us, first, that PC is compatible with the view that desires have a role to play within the causal explanation of an agent’s action. That a desire to φ cannot be what motivates A’s φ-ing, Dancy tells us, “does nothing to undermine the claim of desire to be a necessary part of whatever complex is capable of leading to action”. (86) Furthermore, Dancy subscribes to the claim that the causal conditions of an action necessarily involve desires: “That desire [the desire to ψ] must of course be a necessary part of whatever complex led to A’s φ-ing, but it cannot be what motivated that action.” (86) Both these claims presuppose that there is a distinction between motivational and causal explanations such that the mental items referred to in the causal explanation of an action need not occur within the motivational explanation of the very same action. But what about the items cited in motivational explanations? It seems clear that PC has to take beliefs as figuring in both kinds of explanations. For otherwise motivational explanations in terms of beliefs were but mere narratives: If what is taken to motivate an action (or what is taken to motivate an agent to do something) would not also be part of the causal sequence leading to the action, then motivating states would be epiphenomenal, and then motivational explanations would not capture what really accounts for an agent’s actions. Thus, PC is committed to the view that whereas the role of desires is confined to causal explanations, beliefs figure in both motivational and causal explanations. But this, I claim, is not a stable position.

163

Given what we have said sofar, there must be an intimate relationship between motivational explanations and causal explanations: If a given action is motivated by an agent’s belief, then there must also be a causal explanation in terms of this very belief and a desire of the agent. That is, the presence of a motivational explanation guarantees that there is causal story involving, as parts, the motivating belief and a desire which is supposed to be a state of motivatedness. We thus arrive at a position according to which A’s having a desire to φ is entailed by A’s having a belief (or a set of beliefs) about φ-ing. This view, however, is not compatible with Dancy’s claim that desires contribute causally to an agent’s actions. For if there is a special causal role to play for desires such that desires are indispensible parts of what leads to action, desires and beliefs need to be distinct states. The view under consideration, however, precisely denies that desires are ontologically distinct states. The instability of Dancy’s position as regards the role of desires comes to the fore in greater detail if we take into consideration the fact that PC is committed to the view that actions can be made intelligible regardless of the contents of an agent’s desires. That PC is committed to this claim is easily shown: (14) If A’s φ-ing can be made intelligible regardless of A’s desires, then it can be made intelligible regardless of the contents of A’s desires. (15) A’s φ-ing can be made intelligible regardless of A’s desires. Therefore: (16) A’s φ-ing can be made intelligible regardless of the contents of A’s desires. (16) tells us that what an agent desires is totally irrelevant for understanding actions. But then motivational explanations must not be constrained by what an agent desires. This condition, however, cannot be met. If Tom has a strong aversion towards worms, Tom cannot be motivated to dig in his garden by his belief that digging is a necessary means to pick up worms. A motivational explanation referring to this belief of Tom is obviously false. And thus motivational explanations are constrained by what an agent desires. (16) tells us, furthermore, that which of an agent’s desires is causally efficacious is totally irrelevant for understanding actions. For it cannot matter which of an agent’s desires is efficacious if what an agent desires is supposed to be irrelevant for understanding actions. But then motivational explanations must not be constrained by facts about

164

the causal efficaciousness of desires. Like the first condition, this second one can also not be met. If the fact that Jasper plays piano today is causally explained, inter alia, by his desire to be well prepared for a concert next day but not by his desire to please his mother, his playing piano cannot be motivated by his belief that his mother will be pleased if he plays piano today. Under circumstances like this, a motivational explanation citing Jasper’s belief that his mother will be pleased if he plays piano is simply wrong. Therefore, motivational explanations are constrained by facts about which of an agent’s desires contributes causally to an action of his. To deny that motivational explanations are constrained by what an agent desires and by which of his desires is causally efficacious, one had to admit that any desire could contribute to the causal explanation of any action. This, however, is but to make completely void the conception of desires as causally relevant mental states. Therefore, if Dancy admits that desires have a causal role to play for the generation of action, he is forced to admit, too, that what an agent desires matters for making his actions intelligible. But then, of course, PC has disappeared from the scene. Can a pure cognitivist fare better if he dispenses with desires altogether? To dispense with desires altogether, a pure cognitivist would have to attribute to beliefs those features Humeans attribute to desires in order to construct a plausible view of what explains actions. But this seems to generate a serious quandary for PC. For either there is a substantial difference between motivating states and states of being motivated or there is not. If there is no substantial difference, and motivating states are (as a conceptual matter) states of being motivated, then the pure cognitivst cannot argue, like Dancy, that Humeans have misconstrued desire talk, but is forced to explain away desire talk. But if there is a substantial difference between motivating states and states of being motivated, then the pure cognitivist cannot exclusively rely on construing beliefs as motivating states to account for what explains actions. That is, he has to reintroduce states of being motivated in his theory, and then he either must claim that beliefs are not only motivating states but also states of being motivated, or he must claim that, in addition to beliefs, there are non-belief like states that are indispensible parts of whatever explains action. Either way, the prospects for PC seem not to be very bright. To explain away desire talk (while preserving belief talk) is really heroic an undertaking; that beliefs are states of being motivated seems to be ad hoc and is hardly more plausible than claiming that desire talk is infected with error; and to

165

reintroduce non-belief like states is but to fall back on the desire account of motivation. Of all the alternative routes to, and constructions of, PC I can imagine, Dancy’s strategy to reinterpret desire talk seems to be the most promising one. But, for reasons I have pointed out, even this most promising strategy is not successful.14

14. I thank Thomas Grundmann for intensive discussion on the topic during a joint seminar, the participants of the philosophical club meeting at the University of Bielefeld for a lively discussion of the present paper and Jan Opsomer and an anonymous referee for helpful comments on an earlier draft.

166

Grazer Philosophische Studien 76 (2008), 167–189.

DER SINN FÜR SCHÖNHEIT Gerhard ERNST Universität München Abstract If there is a sense of beauty, what is its nature? The main problem we are confronted with here is that on the one hand a sense of beauty, somehow, has to be connected with perception. On the other hand it does not seem to be reducible to it. I argue that a sense of beauty should be analyzed as a faculty of reason. I try to elucidate the nature of this faculty by identifying similarities between aesthetic, moral and scientific knowledge.

Einleitung Das Vermögen, mit dem wir erkennen können, dass etwas schön ist, nenne ich im Folgenden „den Sinn für Schönheit“. Ich gehe für die Zwecke dieses Aufsatzes davon aus, dass es ein solches Vermögen gibt. Meine Frage lautet: Was ist seine Natur? Diese Frage ist vor allem deshalb schwer zu beantworten, weil ein Sinn für Schönheit zwar einerseits offensichtlich etwas mit „gewöhnlicher“ Wahrnehmung zu tun hat – jedenfalls in zentralen Fällen –, andererseits jedoch nicht auf diese reduzierbar zu sein scheint (vgl. dazu Abschnitt 1 und 2). Die Klärung der Natur eines Sinnes für Schönheit verlangt darum zunächst einmal die genauere Bestimmung des Verhältnisses zwischen diesem Sinn und gewöhnlicher Wahrnehmung. Diese Bestimmung werde ich im Wesentlichen anhand einer kritischen Auseinandersetzung mit dem Ansatz von Dominique Lopes vornehmen (Abschnitt 3). Das Ergebnis meiner Überlegungen wird sein, dass der Sinn für Schönheit, wenn es ihn gibt, als ein Vermögen unserer Vernunft angesehen werden muss. Die Natur dieses Vermögens werde ich durch den Vergleich mit moralischer (Abschnitt 4) und wissenschaftlicher Erkenntnis (Abschnitt 5) zu klären versuchen.

1. Sehen schöne Dinge schön aus? Die Suche nach einem Sinn für Schönheit scheint sehr schnell ans Ziel zu gelangen: Sehen wir die Schönheit vieler Dinge nicht einfach? Anders formuliert: Können wir die Schönheit schöner Dinge nicht einfach deshalb erfassen, weil schöne Dinge schön aussehen? Auf den ersten Blick scheint diese Frage nur eine Antwort haben zu können: Natürlich! Wir können doch von ganz vielen Dingen sagen, dass sie schön sind, weil sie schön aussehen. Kunstwerke, etwa Bilder, Skulpturen und Bauwerke, können, wie es scheint, ebenso gut aufgrund ihres Aussehens schön sein wie andere Artefakte, aber auch wie natürliche Objekte, etwa Landschaften oder Wolkenformationen. Sehen alle schönen Dinge schön aus? Natürlich nicht. Ein Musikstück oder das Rauschen des Windes etwa können schön sein, ohne irgendwie auszusehen, und auch eine Skulptur oder ein Stoffstück können vielleicht schön sein, obwohl sie nicht schön aussehen: Das Musikstück und der Wind können sich schön anhören, die Skulptur und das Stoffstück können sich schön anfühlen. Das Gehör und vielleicht der Tastsinn wären damit ebenso Sinne für Schönheit wie unsere Augen. Sind alle Sinne potenzielle Sinne für Schönheit? Es ist klar, dass sich etwas schön anhören kann; etwas zweifelhafter erscheint es mir, ob sich etwas tatsächlich auch schön (und nicht etwa nur angenehm) anfühlen kann. Dass es irgendwo schön riecht, kann man aber eigentlich nicht sagen,1 und auch das beste Essen schmeckt nicht schön, sondern gut. Ich werde auf diese Beobachtung später kurz zurückkommen. Halten wir zunächst fest, dass primär unser Sehvermögen, das Gehör und vielleicht der Tastsinn als Sinne für Schönheit in Frage kommen. Wir müssen uns dann allerdings auf sinnliche Schönheit beschränken, denn wir sprechen auch in Fällen von Schönheit, in denen unsere Sinne offensichtlich keine Rolle spielen. Wenn beispielsweise überhaupt etwas an einem Roman unseren Sinnen zugänglich ist (und dieser nicht eher als ein abstraktes Objekt zu betrachten ist), dann sind es Lesungen oder gedruckte Ausgaben des Textes. Aber ein Roman kann schön sein, obwohl es nur misslungene Lesungen und hässliche Ausgaben des Textes gibt. Ähnliches lässt sich über andere Literaturgattungen sagen: Auch wissenschaftliche Theorien können, so scheint es, eine eigene Schönheit besitzen. Und das 1. Wir sagen manchmal: „Hier riecht es aber ganz schön!“ Dann meinen wir aber, dass es ganz schön (also ziemlich) stark (und zwar unangenehm) riecht.

168

könnte vielleicht sogar dann der Fall sein, wenn nicht nur alle Ausgaben entsprechender Texte unansehnlich wären, sondern wenn zudem auch noch alle Darstellungen der Theorie als wenig gelungen bezeichnet werden müssten. Manche Philosophen sind der Ansicht, dass man in den soeben genannten Fällen streng genommen überhaupt nicht oder nur in einem metaphorischen Sinn von Schönheit sprechen kann. Schönheit ist für sie etwas, was notwendigerweise an Sinnlichkeit gebunden ist.2 Ich möchte mich auf diese Diskussion hier jedoch nicht einlassen und schränke darum mein Thema entsprechend ein: Die Frage, um die es mir geht, ist allein die nach der Natur eines Sinnes für sinnliche Schönheit, also für die Schönheit der sinnlich wahrnehmbaren Welt, zu der insbesondere viele Werke der Kunst gehören. Mit dieser Einschränkung ist es dann tatsächlich naheliegend, zu behaupten, dass Schönheit unseren Sinnen zugänglich ist. Die genannten Dinge sind nun einmal schön, weil sie schön aussehen oder sich schön anhören. Wie könnte man eine derartige Trivialität leugnen? Umso beunruhigender ist es, dass diese Trivialität falsch zu sein scheint. Die folgende bekannte Überlegung spricht dafür, dass man tatsächlich die Schönheit eines Dinges niemals mit den Sinnen erfassen kann: Nehmen wir an, zwei Betrachter stehen vor einem Gemälde. Der eine kommt zu dem Schluss, dass das Gemälde sehr schön ist, der andere ist dagegen der Ansicht, dass das Bild nicht nur nicht als besonders schön, sondern geradezu als abstoßend bezeichnet werden muss. Was können wir daraus schließen? Eines jedenfalls nicht: dass nämlich einer der beiden nicht weiß, wie das Bild aussieht. Es kann sein, dass beide Betrachter sich völlig einig darüber sind, was es hier zu sehen gibt, und doch hält der eine das Gemälde für schön, der andere nicht. Natürlich gibt es eine Vielzahl von Faktoren, die dazu führen können, dass die beiden Betrachter sich entweder nicht einig darüber sind, was es zu sehen gibt, oder dass sie nur glauben, sich einig zu sein, in Wirklichkeit aber Verschiedenes wahrnehmen. Ich werde auf diese Faktoren noch zurückkommen. Aber die Tatsache, dass ein Betrachter das Gemälde für schön hält, der andere nicht, zeigt noch nicht, dass auch nur einer dieser Faktoren zum Tragen kommt. Was für das Gemälde gilt, gilt für beliebige Dinge, und was für das Aussehen gilt, gilt auch für das, was dem Gehör (und vielleicht dem Tastsinn) zugänglich ist: Man kann sich über die wahrnehmbaren Eigenschaften eines Dinges einig, 2. Vgl. Schmücker (2006), S. 250–253, sowie Zangwill (2001a).

169

über seine Schönheit jedoch uneinig sein. Folglich scheint die Schönheit schöner Dinge unseren Sinnen allein unzugänglich zu sein. Das beschriebene Beispiel wird gewöhnlich herangezogen, um für die Relativität ästhetischer Urteile zu argumentieren.3 Es gibt hier, so scheint es, einen rational nicht auflösbaren Dissens. Und daraus folgert man, dass Urteile über die Schönheit einer Sache zu relativieren sind. Man muss das Beispiel jedoch nicht in dieser Weise deuten, denn man kann ja durchaus der Ansicht sein, dass tatsächlich nur einer der beiden Betrachter recht, der andere unrecht hat. Die Relativierung von Urteilen über Schönheit ist nur dann notwendig, wenn man glaubt, dass möglicherweise keiner von beiden einen Fehler macht. Ich nehme dagegen nur an, dass der Dissens zwischen beiden nicht allein durch den Verweis auf das, was unseren Sinnen zugänglich ist, entscheidbar ist. Von der Relativismusfrage sehe ich dagegen in diesem Aufsatz einmal ab. Immerhin: Auch so scheiden damit unsere fünf Sinne als Sinne für Schönheit aus. Was ist das also: der Sinn für Schönheit? 2. Einige Unterscheidungen Die Behauptung, dass sich die beiden Betrachter über das, was sie sehen, einig sein können, nicht jedoch über dessen Schönheit, setzt einige wichtige Unterscheidungen voraus, die ich in diesem Abschnitt kurz erläutern möchte. Zunächst sollte man zwischen zwei Bedeutungen des Satzes „Das ist schön“ unterscheiden. Zum einen verwenden wir diesen Satz durchaus, um einem Gegenstand bestimmte sinnliche Qualitäten zuzuschreiben. In diesem Sinn kann man beispielsweise von dem Triptychon Mai-Juni 1973, in dem Francis Bacon den Selbstmord seines Liebhabers George Dyer darstellt, kaum sagen, dass es schön ist, wohl aber von Botticellis Geburt der Venus. In der gesamten Kunstgeschichte (und besonders in der Kunst des späteren 20. Jahrhunderts) finden sich zahlreiche Werke, deren Hauptaufgabe es geradezu ist, den ästhetischen Reiz des Hässlichen, ja sogar des Ekelerregenden zu enthüllen.4 Jemand, der diese Kunstwerke als schön bezeichnet, verwendet den Satz „Das ist schön“ vermutlich nicht im hier gemeinten, sondern in einem zweiten Sinn. 3. Bender (1996), Goldman (1995), S. 26–39, Steinbrenner (2005a). 4. Vgl. dazu beispielsweise die Diskussion von Kieran (2005), S. 75–86.

170

Zum anderen bringen wir nämlich mit diesem Satz auch ein Gesamturteil über den ästhetischen Wert einer Sache zum Ausdruck. Statt zu sagen „Das ist schön“ könnte man dann auch sagen „Das ist ästhetisch wertvoll“. Alle ästhetisch wertvollen Kunstwerke sind in diesem abstrakten Sinn schön, auch diejenigen, die kaum als schön im ersten Sinn bezeichnet werden können wie Bacons Triptychon. Umgekehrt sind nicht alle Dinge, die im ersten Sinn schön sind, auch besonders schön im zweiten Sinn, also ästhetisch wertvoll. Manche Kunstwerke sind beispielsweise „nur schön“, sonst aber langweilig anzusehen. Unser Beispiel der beiden Gemäldebetrachter spricht nur dafür, dass der ästhetische Wert einer Sache den Sinnen niemals zugänglich ist. Das Prädikat „ist schön“ kann also zwei Funktionen übernehmen. Es kann zum einen dazu dienen, eine inhaltsreiche ästhetische Bewertung einer Sache zum Ausdruck zu bringen. Zum anderen fällen wir damit auch abstrakte ästhetische Werturteile.5 Im ersten Fall wird die Sache nicht nur positiv bewertet; es werden ihr vielmehr auch bestimmte empirische Eigenschaften zugeschrieben. Im zweiten Fall dagegen geht es allein um die Bewertung.6 Viele ästhetische Prädikate können dazu dienen, inhaltsreiche ästhetische Bewertungen zum Ausdruck zu bringen – „ist langweilig“, „ist lebendig“, „ist verworren“, „ist differenziert“ … Die entsprechenden ästhetischen Begriffe hat man, in Analogie zu bestimmten ethischen Begriffen, als „thick concepts“, als inhaltreiche Begriffe, bezeichnet.7 Diese haben neben einer deskriptiven immer auch eine wertende Komponente.8 Wenn im Folgenden von Schönheit die Rede ist, wird es allerdings nicht um Schönheit im inhaltsreichen Sinn, sondern allein um Schönheit als ästhetischer Wert gehen. Die Frage nach der Natur des Sinnes für Schönheit verstehe ich somit als die Frage nach der Natur der Erkenntnis ästhetischen Wertes. Dass Schönheit im abstrakten Sinn den Sinnen allein nicht zugänglich ist, scheint prima facie nicht so überraschend zu sein. Loben wir bei5. Ich lehne mich hier an Sibleys Unterscheidung zwischen „substantive aesthetic evaluations“ und „aesthetic verdicts“ an. Vgl. dazu Sibley (2001); siehe auch Zangwill (2001b), S. 12ff. 6. Ein Naturalist müsste die Unterscheidung allerdings in anderer Weise charakterisieren, da für ihn wohl auch die abstrakte ästhetische Bewertung der Zuschreibung einer (komplexen) empirischen Eigenschaft dient. Ich danke Jakob Steinbrenner für diesen Hinweis. 7. Vgl. Williams (1985), S. 141–143, Gibbard (1992), Zangwill (2001b), S. 16–20. 8. Die Frage, ob alle, die meisten oder die wenigsten ästhetischen Begriffe von dieser Art sind und was das Spezifikum ästhetischer Begriffe ist, kann hier nicht weiter diskutiert werden. Vgl. Zangwill (2001b).

171

spielsweise Kunstwerke nicht aus den unterschiedlichsten Gründen?9 Und könnte es nicht sein, dass der Wert eines Kunstwerkes immer in Eigenschaften liegt, die den Sinnen nicht (direkt) zugänglich sind, etwa in der politischen Wirksamkeit des Werkes etc.? Hier sind allerdings zwei weitere Unterscheidungen zu beachten: Erstens kann man den nicht-künstlerischen Wert eines Kunstwerks von seinem künstlerischen Wert unterscheiden. Ein Kunstwerk kann für die verschiedensten Zwecke eingesetzt werden: eine Skulptur als Briefbeschwerer, ein Musikstück, um Hühner zum Eierlegen zu animieren, ein Gedicht als Eselsbrücke.10 Ein Kunstwerk kann darum wertvoll sein, ohne künstlerisch wertvoll zu sein. Letzteres ist nur dann der Fall, wenn es als Kunstwerk wertvoll ist. Dass der nicht-künstlerische Wert eines Kunstwerks den Sinnen häufig nicht (direkt) zugänglich ist, ist kaum überraschend. Vielleicht ist nicht einmal die These, dass der künstlerische Wert eines Kunstwerks den Sinnen häufig nicht zugänglich ist, besonders beunruhigend. Aber der künstlerische Wert eines Kunstwerks ist, zweitens, noch einmal von seinem ästhetischen Wert zu unterscheiden.11 Was auch immer ein Kunstwerk zu einem guten Kunstwerk macht, muss als sein künstlerischer Wert angesehen werden. Aber es ist nicht notwendigerweise der ästhetische Wert eines Kunstwerks, der es zu einem guten Kunstwerk macht. Man denke an dieser Stelle noch einmal an das Beispiel der Literatur: Romane etwa haben (häufig) einen künstlerischen Wert, auch wenn sie keinen ästhetischen Wert im hier behandelten Sinn, also keinen sinnlichen ästhetischen Wert haben. Umgekehrt gibt es jedenfalls Dinge, die einen ästhetischen, aber keinen künstlerischen Wert haben: Viele Objekte der Natur haben einen ästhetischen Wert, aber keinen künstlerischen, einfach deshalb, weil es sich bei ihnen (in aller Regel) nicht um Kunstwerke handelt. Wenn wir vom ästhetischen Wert der sinnlich wahrnehmbaren Welt (deren Teil viele Kunstwerke sind) sprechen, meinen wir einen Wert, 9. Vgl. dazu Schmücker (2001) und (2003). 10. Im ersten Fall kann man vielleicht sagen, dass es nicht das Kunstwerk ist, das zweckentfremdet wird, sondern das physische Objekt, in dem sich das Kunstwerk manifestiert. Vgl. zu dieser Unterscheidung Schmücker (2001), S. 16, 17. Im zweiten Fall ist das aber nicht so leicht möglich und im dritten Fall überhaupt nicht. Falls allerdings gerade die künstlerische Funktion etwas erst zu einem Kunstwerk macht (vgl. dazu „When is Art?“ in Goodman (1978)), kann ein Kunstwerk nicht zweckentfremdet werden. Die gerade diskutierte Unterscheidung ist dann hinfällig. 11. Diese Unterscheidung wird häufig übergangen. Goldman etwa beschäftigt sich in seinem bekannten Buch Aesthetic Value allein mit künstlerischem Wert. Vgl. auch Steinbrenner (2005b).

172

der an die wahrnehmbaren Eigenschaften einer Sache gebunden ist. Dass der ästhetische Wert den Sinnen allein dennoch nicht zugänglich zu sein scheint, macht die Natur eines Sinnes für Schönheit so schwer verstehbar. 3. Sinnliche Schönheit Die Natur des Sinns für sinnliche Schönheit beziehungsweise, wie wir jetzt sagen können, für ästhetischen Wert lässt sich nur klären, indem man die Beziehung zwischen ästhetischem Wert und Sinnlichkeit genauer bestimmt. Als Ausgangspunkt dafür wähle ich die Analyse ästhetischer Bewertung, die Dominique Lopes in seinem Buch Sight and Sensibility vorschlägt.12 Lopes bezieht sich speziell auf die ästhetische Bewertung von Bildern; sein Ansatz ließe sich jedoch leicht verallgemeinern. Er schreibt: Letting ‚P’ stand for any picture and ‚F’ for any property (such as ‚square’, ‚insipid’, or ‚beautiful’), the internalist conjecture holds that an evaluation, R, of P as F is an aesthetic evaluation if and only if, were R accurate, (1) being F would be a (de)merit in P, all else being equal; (2) a suitable observer’s experience, E, of P as F is partly constitutive of (1); and (3) R is an experience with the same content as E or R is a representation warranted by E.13

Die entscheidende Bedingung in dieser Analyse ästhetischer Bewertung ist die zweite. Hier wird deutlich, worin die besondere Beziehung zwischen der ästhetischen Bewertung eines Objektes – bei Lopes: eines Bildes – und der Wahrnehmung des Objektes besteht: Die Wahrnehmung ist konstitutiv für den Wert. Lopes erläutert seine zweite Bedingung folgendermaßen (wobei er hier für P das Bild Moonlit Kittens und für F „sentimental“ als Beispiele einsetzt): This language in clause (2) about partial constitution requires caution on two fronts. First, the claim is not that the experience of Kittens as sentimental is part of what it is for the picture to be sentimental. Nor is it that an experience of the sentimentality as a flaw is part of what makes the picture’s sentimentality a flaw. […] The claim advanced in clause (2) is that the experience of Kittens as sentimental is part of what it is for the picture’s sentimentality to 12. Vgl. zum Folgenden Lopes (2005), 3. Kapitel. 13. Vgl. Lopes (2005), S. 107.

173

be a flaw in it. Here are three facts: (i) Kittens is sentimental, (ii) Kittens is experienced as sentimental, and (iii) being sentimental is a flaw in Kittens. The conjecture says that (ii) partly constitutes (iii).14

Die Analyse von Lopes ist ziemlich abstrakt. Betrachten wir darum auch einmal, zu welchen Ergebnissen wir kommen, wenn wir die Beispiele verwenden, die Lopes (im ersten oben angeführten Zitat) selbst vorschlägt: Was liefert die Analyse, wenn man für F „quadratisch“, „geschmacklos“ und „schön“ einsetzt? Nehmen wir an, wir kritisieren ein Bild, indem wir sagen, es sei quadratisch. Nehmen wir weiterhin an, die Bewertung sei korrekt, so dass es tatsächlich ein Fehler des Bildes ist, dass es quadratisch ist. Was diese Bewertung zu einer ästhetischen Bewertung macht, ist nach Lopes, dass die Erfahrung (eines geeigneten Betrachters) des Bildes als quadratisch (teilweise) konstitutiv dafür ist, dass es ein Fehler des Bildes ist, dass es quadratisch ist. Nach der Erläuterung der zweiten Bedingung von Lopes ist die Erfahrung des Bildes als quadratisch nicht konstitutiv dafür, dass das Bild quadratisch ist. Und das ist natürlich plausibel. Weiterhin kommt es nicht darauf an, dass es als Fehler wahrgenommen wird, dass das Bild quadratisch ist. Es ist vielmehr die Erfahrung des Bildes als quadratisch allein, welche die quadratische Form des Bildes zu einem seiner Schwachpunkte macht. Die naheliegende Deutung dieser These ist: Die quadratische Form des Bildes ist ein Schwachpunkt, weil sie Ursache davon ist, dass wir das Bild als quadratisch wahrnehmen, und diese Wahrnehmung keinen (oder negativen) Wert hat. Es ist also gerade die These des sogenannten ästhetischen Empiristen, welche die Bedingung (2) von Lopes verständlich macht: dass nämlich der ästhetische Wert einer Sache in den Erfahrungen liegt, die ein geeigneter Beobachter15 mit der Sache machen kann, dass es also letztlich diese Erfahrungen sind, die eigentlich wertvoll sind.16 Lopes möchte sich jedoch nicht auf einen ästhetischen Empirismus festlegen. Er schreibt: All the same, the internalist conjecture suggests a way to argue for aesthetic empiricism. Why is clause (2) true? Perhaps the best explanation of why an experience of Kittens as sentimental makes its sentimentality a flaw in it is 14. Vgl. Lopes (2005), S. 108. 15. Ein geeigneter Beobachter ist jemand, dem es nicht am Sinn für Realität fehlt. Was das heißt, wird im vierten Abschnitt genauer erläutert werden. Entscheidend im gegenwärtigen Zusammenhang ist, dass nicht nur derjenige, der einen Sinn für Schönheit hat, als geeigneter Beobachter zählt. 16. Diese Auffassung kann als die klassische Auffassung ästhetischen Wertes gelten, die angefangen von den Geschmackstheorien des 18. Jahrhunderts bis in die Gegenwart hinein (vgl. etwa Budd (1995) und Levinson (1996)) vorherrschend ist.

174

that the experience of Kittens as sentimental is intrinsically bad. By the same token, one may accept the conjecture and reject aesthetic empiricism by denying that the latter explains the former.17

Der Grund dafür, dass Lopes keinen ästhetischen Empirismus vertreten möchte, ist, dass er diese Position für zu umstritten hält. Und tatsächlich gibt es gerade in der neueren Diskussion eine Reihe von Kritikern dieses klassischen Ansatzes.18 Diese Kritik basiert jedoch meines Erachtens vollständig darauf, dass viele ästhetische Empiristen ästhetischen und künstlerischen Wert gleichsetzen. Der künstlerische Wert einer Sache kann aber tatsächlich kaum allein in den Erfahrungen liegen, die wir mit der Sache machen können. Es wäre sonst beispielsweise schwer verständlich, warum wir Originale mehr schätzen als (selbst perfekte) Kopien und vielleicht auch warum wir Ready-Mades als Kunstwerke klassifizieren.19 Damit ist jedoch noch nichts über den ästhetischen Empirismus als Theorie des ästhetischen Wertes gesagt. In Bezug auf diesen bleibt er meines Erachtens die einzig plausible Position. Wie auch immer: Lopes muss sich an dieser Stelle festlegen, da die Aussage, dass bestimmte Erfahrungen einer Sache den ästhetischen Wert dieser Sache konstituieren, für sich genommen unverständlich ist. Und es ist schwer zu sehen, wie diese Bedingung ohne ästhetischen Empirismus erläutert werden könnte. Lopes bietet jedenfalls keine Alternative an. Setzen wir den ästhetischen Empirismus voraus, so lässt sich die Analyse von Lopes, wie gesagt, plausibel deuten: Eine Schwäche unseres quadratischen Bildes liegt in bestimmten Erfahrungen, die man (das heißt ein geeigneter Beobachter) mit dem Bild machen kann, nämlich in der Erfahrung des Bildes als quadratisch. Diese Erfahrung ist wertlos oder sogar negativ zu bewerten. Es ist aber nicht die Erfahrung der Form des Bildes als Fehler, worin die Schwäche wurzelt, sondern allein die Erfahrung der Form. Anders gesagt: Der ästhetische Wert liegt in den Erfahrungen bestimmter Eigenschaften des Bildes, nicht in der Erfahrung, dass bestimmte Eigenschaften des Bildes wertvoll/los sind! Das einzige Problem mit diesem ersten Beispiel von Lopes besteht darin, dass die Aussage „Das 17. Vgl. Lopes (2005), S. 118. 18. Vgl. Davies (2004), (2006), Kieran (2005), S. 21–33, sowie die Literaturangaben in Kieran (2006), S. 35. 19. Die (perfekte) Kopie ermöglicht nämlich prinzipiell die gleichen Erfahrungen wie das Original (auch wenn wir vielleicht bestimmte Informationen über das Original brauchen, um diese Erfahrungen machen zu können; vgl. unten). Ob auch das Ready-Made genau die gleichen Erfahrungen ermöglicht wie sein alltägliches Pendent, ist schon eher fraglich.

175

Bild ist quadratisch“ für sich genommen überhaupt keine Wertung zum Ausdruck bringt. Insofern ist es meines Erachtens nicht klar, ob Lopes für F überhaupt „quadratisch“ einsetzen darf. Springen wir zum dritten Beispiel von Lopes und setzen „schön“ für F ein. Jetzt verschwindet das zuletzt genannte Problem. Klarerweise kann man ein Bild als schön bewerten. Allerdings scheint jetzt die Erläuterung der zweiten Bedingung von Lopes keinen rechten Sinn mehr zu ergeben. Denn wie kann man behaupten, dass zwar die Erfahrung des Bildes als schön (teilweise) konstitutiv dafür ist, dass es eine Stärke des Bildes ist, schön zu sein, dass aber die Erfahrung der Schönheit des Bildes als einer Stärke nicht (teilweise) konstitutiv dafür ist, dass es eine Stärke des Bildes ist, schön zu sein? Die Erfahrung der Schönheit einer Sache ist die Erfahrung der Schönheit der Sache als einer Stärke, jedenfalls dann, wenn wir Schönheit im abstrakten Sinn meinen, die ja gerade mit dem ästhetischen Wert gleichgesetzt werden kann. Die Erfahrung des (positiven) Wertes einer Sache ist notwendigerweise die Erfahrung des Wertes als einer Stärke! Es ist darum unklar, was Lopes hier behaupten möchte. Jedenfalls ist es auch hier nicht die Erfahrung des ästhetischen Wertes eines Bildes, die das Bild wertvoll macht, sondern die Erfahrung bestimmter (nichtevaluativer) Eigenschaften des Bildes. Das zweite Beispiel, bei dem wir für F „geschmacklos“ einsetzen sollen, führt entweder zum Problem des ersten oder des dritten Beispiels. Es handelt sich hier wohl nach Ansicht der meisten um einen inhaltreichen ästhetischen Begriff,20 und je nachdem, ob wir seinen evaluativen oder seinen deskriptiven Anteil betrachten, kommen wir zu der einen oder anderen gerade geschilderten Schwierigkeit. Wenn wir ein Bild als geschmacklos erfahren, dann kommen wir meines Erachtens nicht umhin, die Geschmacklosigkeit des Bildes als einen Fehler zu erfahren. Es ist aber nicht die Erfahrung der Geschmacklosigkeit als Fehler, die konstitutiv dafür ist, dass die Geschmacklosigkeit ein Fehler des Bildes ist. Das Schlechte an dem Bild ist vielmehr, dass wir (potenziell) bestimmte nichtevaluative Erfahrungen mit ihm machen. Sehen wir aber von der evaluativen Kom20. Stellt in Bezug auf bestimmte Kunstwerke (etwa solche von Jeff Koons oder Andy Warhol) die Behauptung „Dieses Werk ist geschmacklos“ vielleicht für sich genommen keine Wertung oder sogar eine positive Wertung – „Dieses Werk ist herrlich geschmacklos“ – dar? (Ich danke Jakob Steinbrenner für den Hinweis auf diese Frage.) Da meine gegenwärtige Überlegung nicht davon abhängt, ob das Prädikat „ist geschmacklos“ für sich genommen überhaupt nicht wertend ist (und damit wie das Prädikat „ist quadratisch“ behandelt werden kann) oder ob es sogar für sich genommen positiv wertend sein kann (und damit die folgenden Sätze einfach entsprechend variiert werden müssen), muss ich diese Frage hier nicht entscheiden.

176

ponente ab, so ist das Urteil „Dieses Bild ist geschmacklos“ überhaupt kein Werturteil mehr und kann darum auch nicht Ausgangspunkt für die Analyse von Lopes sein. Wir können aus dieser Diskussion der Analyse von Lopes folgende Schlüsse ziehen: Der ästhetische Wert einer Sache liegt in der (potenziellen) Wahrnehmung der Sache.21 Der ästhetische Wert eines Sonnenuntergangs liegt in der (Möglichkeit der) Betrachtung des Sonnenuntergangs, der ästhetische Wert eines Musikstückes liegt im (potenziellen) Hören des Musikstükkes etc. Anders gesagt: Was den Sonnenuntergang schön macht, ist, dass er schön aussieht, was das Musikstück schön macht, ist, dass es sich schön anhört. Die Wahrnehmung selbst ist dabei jedoch noch nicht mit einer Wertung verbunden. Es ist die Wahrnehmung bestimmter Eigenschaften, die wertvoll ist, nicht die Wahrnehmung bestimmter Eigenschaften als Stärken oder Schwächen. Man kann das auch so auf den Punkt bringen: Das Schöne (und damit ästhetisch Wertvolle) an einer Sache ist nicht die (potenzielle) Erfahrung der Schönheit, sondern die Erfahrung von etwas Schönem. Wir erkennen demnach den ästhetischen Wert einer Sache, indem wir den Wert von (potenziellen) Wahrnehmungen der Sache erkennen. Bevor wir die Natur dieser Werterkenntnis (und damit die Natur des Sinns für Schönheit) weiter klären, sind noch drei Abgrenzungen zumindest anzudeuten. Erstens können wir Wahrnehmungen, wie es scheint, auch in Hinblick auf ihren Nutzen beurteilen. Darum geht es jedoch bei der ästhetischen Bewertung offensichtlich nicht. Man sollte deshalb sagen, dass wir es nur dann mit dem ästhetischen Wert einer Sache zu tun haben, wenn wir es mit dem intrinsischen Wert einer Wahrnehmung zu tun haben. Allerdings bleibt es dann, zweitens, immer noch schwierig, den ästhetischen Wert einer Wahrnehmung von ihrem kognitiven Wert zu unterscheiden, denn auch bei Letzterem kann es sich um einen intrinsischen Wert handeln. Dieses Problem kann im vorliegenden Rahmen nicht gelöst werden, da dazu zunächst eine genauere Charakterisierung des kognitiven Wertes der Wahrnehmung vorzunehmen wäre, was hier nicht geleistet werden kann.22 Drittens schließlich umfasst unsere Charakterisierung ästhetischer Bewertung offensichtlich auch hedonistische Bewertungen.23 Indem wir 21. Das Wort „Wahrnehmung“ ist hier in einem weiten Sinn zu verstehen. Es gibt beispielsweise viele Weisen, ein Kunstwerk wahrzunehmen, die alle erfasst sein sollen. 22. Zum Verhältnis von ästhetischem und kognitivem Wert von Bildern vgl. Lopes (2005), 4. Kapitel. 23. Diese Konsequenz zieht auch Lopes explizit. Vgl. Lopes (2005), S. 118.

177

eine bestimmte Empfindung als angenehm bezeichnen, schreiben wir ihr einen intrinsischen Wert zu, und der Wert angenehmer Dinge liegt stets in der (potenziellen) Wahrnehmung dieser Dinge. Das Angenehme an einem heißen Bad im Winter ist, dass es uns die Erfahrung der allmählichen Erwärmung ermöglicht. Die (potenzielle) Erfahrung ist konstitutiv für den Wert eines heißen Bades. Das Angenehme an einem heißen Bad ist auch hier nicht die (potenzielle) Erfahrung der allmählichen Erwärmung als etwas Positives. Man kann wieder sagen: Das Angenehme (und damit Wertvolle) an einem heißen Bad ist demnach nicht die (potenzielle) Erfahrung von Angenehmheit, sondern die (potenzielle) Erfahrung von etwas Angenehmem. Man kann sich fragen, ob es überhaupt eine Schwäche der bisherigen Analyse ist, dass sie nicht zwischen hedonistischem und ästhetischem Wert unterscheidet. Beide scheinen ja tatsächlich denselben Bezug zur Sinnlichkeit aufzuweisen. Im Hinblick auf die Frage, um die es mir geht, ist die Unterscheidung deshalb nicht unbedingt notwendig. Ich gehe vielmehr davon aus, dass der Sinn für Schönheit prinzipiell von der gleichen Art ist wie der Sinn für das Angenehme. Möchte man dennoch zwischen beiden Bereichen unterscheiden, so hat man verschiedene Möglichkeiten. Man kann etwa bei der Beobachtung ansetzen, auf die ich im ersten Abschnitt hingewiesen habe, dass nämlich typischerweise das Gesehene und das Gehörte als schön bezeichnet werden, während man in erster Linie das Gefühlte als angenehm klassifiziert. Gerüche und Geschmäcker nennt man meistens einfach gut. Es scheint daher eine gewisse Zuordnung zu den verschiedenen Sinnen zu geben: Ästhetische Bewertungen beziehen sich primär auf das Gesehene und das Gehörte, hedonistische Bewertungen auf das Gefühlte.24 Aber die Zuordnung ist nicht strikt: Blendendes Licht kann ebenso wie ein schriller Ton nicht nur unschön, sondern auch unangenehm sein, und eine Skulptur oder ein Stoffstück kann sich, wie gesagt, vielleicht nicht nur angenehm sondern auch schön anfühlen. Man muss zur Unterscheidung beider Bereiche darum zumindest auch den Inhalt der entsprechenden Wahrnehmungen berücksichtigen.25 Da 24. Der entsprechende Sinn ist der Tastsinn, aber auch die innere Körperwahrnehmung (etwa wenn man sich satt fühlt oder einen Muskelkater hat). Letzteres nennt man „Propriozeption“. 25. Natürlich kann man das Problem dadurch verschieben, dass man sagt, Schönheitsurteile bezögen sich, anders als Urteile über das Angenehme, auf ästhetische Erfahrungen. Die Frage nach der Natur ästhetischer Erfahrungen ist meines Erachtens dann aber ihrerseits durch den Verweis auf den Inhalt beziehungsweise Gegenstand dieser Erfahrungen zu klären. Vgl. Carroll (2006).

178

meine weiteren Überlegungen nicht davon abhängen, zu welchen Ergebnissen man dabei kommt, führe ich diesen Gedanken hier jedoch nicht weiter.26 4. Schön und gut Wenn man den ästhetischen Wert einer Sache im intrinsischen Wert bestimmter (potenzieller) Wahrnehmungen dieser Sache sieht, so muss der Sinn für Schönheit ein Vermögen sein, mit dessen Hilfe wir den intrinsischen Wert bestimmter Wahrnehmungen erkennen können. Wie aber erkennt man intrinsische Werte? Hier hilft es meines Erachtens, die enge Verbindung zwischen intrinsischen Werten und Gründen für Handlungen zu beachten: Wenn ich durch meine Handlung etwas intrinsisch Wertvolles erreichen kann, dann spricht das stets für die Handlung. Wenn ich beispielsweise durch eine Handlung erreichen kann, dass beim Kindergeburtstag der Kuchen gerecht zwischen den Kindern aufgeteilt wird,27 dann spricht das dafür, die Handlung auszuführen. Das heißt natürlich nicht, dass ich die Handlung zwangsläufig ausführen sollte – es kann ja auch vieles gegen die Handlung sprechen. Dass der Kuchen gerecht zwischen den Kindern aufgeteilt werden würde, konstituiert aber jedenfalls einen Grund für die Handlung. Was für Gerechtigkeit gilt, gilt für alle intrinsischen Werte: Wenn ich sie durch meine Handlung realisieren kann, dann habe ich einen Grund für meine Handlung. Jemand, der so etwas sagt wie: „Ich kann zwar mit dieser Handlung etwas (in bestimmter Hinsicht) Gutes erreichen, aber es spricht überhaupt nichts dafür, die Handlung auszuführen, anders gesagt: es gibt überhaupt keinen Grund, die Handlung auszuführen“, äußert sich letztlich unverständlich.28 Somit habe ich also beispielsweise auch einen Grund, ins Museum zu gehen, wenn es dort ästhetisch Wertvolles zu sehen gibt. Wenn bestimmte Erfahrungen intrinsisch wertvoll sind, dann habe ich auch Grund, diese Erfahrungen zu ermöglichen – für mich selbst, aber auch für andere. 26. Ebenso verzichte ich auf die in diesem Zusammenhang naheliegende Analyse der Kantischen Position. 27. Das betrachte ich als Beispiel für ein intrinsisch wertvolles Resultat. Wer diese Überzeugung nicht teilt, kann leicht ein anderes Beispiel finden. 28. Externalisten bestreiten (jedenfalls nach einer Deutung) den internen Zusammenhang zwischen dem (in irgendeiner Hinsicht) Guten und Gründen. Dafür, dass der Externalismus nicht haltbar ist, argumentiere ich in Ernst (2006).

179

Der Sinn für Schönheit kann somit als ein Sinn für Gründe identifiziert werden: Jemand, der einen Sinn für Schönheit hat, erkennt, dass er bestimmte Gründe für Handlungen hat. Er erkennt beispielsweise, dass die Tatsache, dass man im Museum bestimmte Erfahrungen machen kann, einen Grund dafür konstituiert, ins Museum zu gehen. So gesehen ist der Sinn für Schönheit nicht nur mit dem Sinn für das Angenehme, sondern insbesondere mit dem moralischen Sinn eng verwandt. Jemand, der einen Sinn für Moral hat, erkennt, dass bestimmte Tatsachen Gründe für Handlungen konstituieren, ebenso wie derjenige, der einen Sinn für Schönheit (und für das Angenehme) hat, erkennt, dass bestimmte Tatsachen Gründe für Handlungen konstituieren. Die Sinne unterscheiden sich darin, welche Tatsachen jeweils als Gründe erkannt werden. Sie kommen darin überein, dass beide als Vermögen der Vernunft zu deuten sind, denn sie ist es, auf die Gründe (als solche) ihrer Natur nach zielen. Gehen wir vor diesem Hintergrund noch einmal zurück zu unserem Ausgangsbeispiel: den beiden Gemäldebetrachtern mit abweichenden ästhetischen Urteilen. Möchte man keine relativistische Interpretation der Situation geben, so muss man davon ausgehen, dass einer der beiden unrecht hat. Wie kann das sein? Wenn die vorliegende Analyse unseres Sinnes für Schönheit korrekt ist, können wir einfach entsprechende Überlegungen aus der Metaethik übertragen. Wenn jemand zu einem falschen moralischen Urteil kommt, dann kann das zwei grundlegend verschiedene Ursachen haben. Eine Ursache besteht darin, dass die Person die zu bewertende Handlung oder die Handlungsumstände falsch einschätzt. Es kann beispielsweise sein, dass die Person sich über die Folgen der Handlung täuscht oder auch über die Frage, welche Handlungen in einer Situation möglich sind und welche nicht. Wenn dass der Fall ist, kann es sein, dass die Person fälschlicherweise zu dem Schluss kommt, dass es richtig ist, Handlung A auszuführen, obwohl es richtig wäre, Handlung B auszuführen. Dieser Person fehlt es dann nicht am Sinn für Moral, sondern am Sinn für Realität. Sie erkennt einfach nicht, was möglich ist beziehungsweise welche Folgen sich ergeben werden. Eine ganz andere Ursache für ein moralisches Fehlurteil ist im Spiel, wenn eine Person die Lage und die Handlungsfolgen völlig richtig einschätzt und dennoch zur falschen Bewertung der Handlung kommt. Nehmen wir an, Helga ist sich völlig darüber im Klaren, dass ein Kind keinen Kuchen bekommen wird, wenn sie den Kuchen in fünf Stücke teilt und diese an die ersten fünf Kinder verteilt. Dass sich diese Verteilung ergeben wird, konstituiert in ihren Augen aber einfach keinen Grund dafür, den Kuchen anders zu

180

schneiden.29 Sie ist nicht blind für die Realität, hat aber offensichtlich keinen Sinn für Gerechtigkeit. Sie ist darum ein (zumindest partieller) Amoralist. Beide Arten von Fehlern kann man auch im Fall ästhetischer Bewertungen finden.30 Zum einen kommt es häufig vor, dass ästhetische Fehlurteile auf einer unzureichenden Wahrnehmung der Gegebenheiten beruhen. Es ist oft schwierig, die Feinheiten zu erkennen, die für ein richtiges ästhetisches Urteil relevant sind. Diese Erkenntnis setzt häufig eine Menge Wissen voraus. Kendall Walton ist beispielsweise der Ansicht, dass die ästhetischen Eigenschaften eines Dinges nicht nur von den intrinsischen Eigenschaften des Dinges abhängen, sondern auch von bestimmten relationalen Eigenschaften, wie etwa der Herstellungsgeschichte oder den Künstlerintentionen bei einem Kunstwerk.31 Die Wahrnehmung ästhetischer Eigenschaften kann so eine Fülle von Hintergrundinformationen verlangen. Auch Nelson Goodman weist beispielsweise darauf hin, dass häufig erst Informationen über den (kunst-)historischen Kontext die Wahrnehmung eines Stils ermöglichen.32 Und es ist völlig klar, dass man eine Anspielung in der Regel nur dann erkennen kann, wenn man das kennt, worauf angespielt wird. Man kann diese Phänomene unter das aus der Wissenschaftstheorie bekannte Schlagwort von der Theoriegeladenheit der Beobachtung subsumieren. So wie man aus einem Röntgenbild die relevanten Informationen nur ersehen kann, wenn man die entsprechenden Hintergrundinformationen (und die entsprechende Übung) hat, so kann man auch die ästhetischen (beziehungsweise die ästhetisch relevanten) Eigenschaften einer Sache häufig nur aufgrund entsprechender Hintergrundinformationen (und entsprechender Übung) wahrnehmen. Wenn in der Ästhetik von idealen Beobachtern die Rede ist, dann möchte man in erster Linie Beobachter auszeichnen, denen keine ästhetisch relevanten Aspekte der Wirklichkeit entgehen. Sie sollen eine geschulte und feine, durch keine Ablenkungen gestörte Wahrnehmung haben und über alle relevanten Hintergrundinformationen verfügen. Erst dann ist ihrem ästhetischen Urteil zu trauen.33 29. Es soll hier also nicht so sein, dass Helga den Grund zwar sieht, aber bessere Gründe hat, den Kuchen dennoch ungerecht zu verteilen. 30. Vgl. zum Folgenden die klassischen Ausführungen von Hume sowie deren Interpretation bei Dickie (Hume (1998), Dickie (1988), S. 141ff.). 31. Vgl. Walton (1970). 32. Vgl. Goodman (1978), 2. Kapitel. 33. Im dritten Abschnitt war genau in diesem Sinn von „geeigneten Beobachtern“ die Rede.

181

Man sollte sich meines Erachtens der These anschließen, dass die meisten moralischen und ästhetischen Meinungsverschiedenheiten durch abweichende Meinungen über nichtmoralische und nichtästhetische Fakten zu erklären sind.34 Wenn zwei Betrachter eines Gemäldes zu unterschiedlichen ästhetischen Urteilen kommen, dann liegt die Vermutung nahe, dass der eine in dem Bild etwas findet (oder zu finden glaubt), was der andere nicht findet (oder nicht zu finden glaubt). Einer von beiden täuscht sich dann, aber nicht aufgrund eines fehlenden Sinnes für Schönheit, sondern aufgrund eines fehlenden Sinnes für die Realität. Dennoch: Es kann eben auch der Fall eintreten, dass beide Betrachter alle relevanten Aspekte erkennen und dennoch nicht zu dem gleichen ästhetischen Urteil kommen. Dann muss der Fehler, den mindestens einer von beiden macht, von anderer Art sein. Wie im Fall der Amoralistin muss man dann davon ausgehen, dass einer von beiden eine bestimmte Tatsache, die er tatsächlich erkennt, nicht in gleicher Weise wie der andere als Grund betrachtet. Ihm fehlt es dann am Sinn für Schönheit. Beide Fehler sind unabhängig voneinander. Es gibt im Ästhetischen wie im Moralischen Menschen, die einen ausgeprägten Sinn für Schönheit und Moral haben, aber leider völlig unsensibel (oder unwissend) sind, was die Wahrnehmung ihrer Umwelt angeht. Sie kommen zu ästhetischen und moralischen Fehlurteilen (und dementsprechend auch zu falschen Handlungen), weil sie die Realität nicht erkennen. Umgekehrt gibt es hier wie dort auch Menschen, die eine feine Wahrnehmung haben, aber das, was sie erkennen, nicht in der richtigen Weise als Gründe erkennen. Wir betrachten solche Menschen als (partielle) Amoralisten und als (partielle) Banausen. Es fehlt ihnen nicht am Sinn für Realität, sondern am Sinn für Moral beziehungsweise Schönheit. 5. Schön und natürlich Die bisherigen Überlegungen sollten deutlich machen, dass die Fähigkeit zur Erkenntnis von Schönheit von der gleichen Art ist wie die Fähigkeit zu moralischer Erkenntnis. In beiden Fällen handelt es sich um „Sinne für Gründe“. Damit ist aber, so scheint es, eher das Problem präzisiert als eine Lösung gefunden. Denn die Natur des Sinnes für Moral ist sicherlich nicht 34. Diese Einsicht wird vor allem von Nonkognitivisten betont. Vgl. etwa Ayer (1990), S. 114, 115.

182

leichter zu klären als die Natur des Sinnes für Schönheit. Die kontroverse Debatte um die Natur moralischer (Tatsachen und) Erkenntnis überträgt sich damit unmittelbar auf die Ästhetik. Dementsprechend findet man hier wie dort nonkognitivistische, naturalistische und nonnaturalistische Positionen.35 Im vorliegenden Rahmen ist es nicht möglich, diese Positionen im Einzelnen zu diskutieren, um ihre Vorzüge und Nachteile aufzuzeigen. Stattdessen möchte ich im Folgenden meinen eigenen Ansatz zur Klärung der Natur ästhetischer Erkenntnis skizzieren.36 Möchte man die Natur ästhetischer Erkenntnis klären, so besteht ein Hauptproblem, wie es scheint, darin, den richtigen Grad von Abhängigkeit und Unabhängigkeit der korrespondierenden ästhetischen Tatsachen vom Erkenntnissubjekt zu erfassen. Hier bietet sich bekanntlich der Vergleich zwischen ästhetischer Erkenntnis und der Erkenntnis sekundärer Qualitäten an:37 So wie die Wahrnehmung etwa von Farben zwar nicht völlig unabhängig von uns (insbesondere von der spezifischen Ausbildung unserer Augen), aber dennoch objektiv (das heißt hier vor allem: eine echte Form der Erkenntnis) ist, so könnte auch ästhetische Erkenntnis als objektive aber subjektabhängige Erkenntnis gedeutet werden. Dieser Vergleich geht meines Erachtens in die richtige Richtung. Auch ich bin der Ansicht, dass sich die Rätselhaftigkeit unseres Sinnes für Schönheit nur beseitigen lässt, indem man deutlich macht, dass sich die Besonderheiten dieser Erkenntnisart auch in Bereichen finden, die wir für weniger problematisch halten. Das beste Vergleichsobjekt wäre natürlich moralische Erkenntnis. Aber diese ist, wie gesagt, nicht weniger problematisch als ästhetische Erkenntnis. Der Vergleich mit der Erkenntnis sekundärer Qualitäten ist diesbezüglich hilfreicher. Allerdings hat auch er eine gravierende Schwachstelle. Was ästhetische (ebenso wie moralische) Erkenntnis so schwer verständlich macht, ist gerade die Tatsache, dass wir es hier mit einem Vermögen zur Erkenntnis von Gründen (als solchen) zu tun haben. Die Erkenntnis sekundärer Qualitäten ist aber genau in diesem Punkt von anderer Art. Die Tatsache, dass etwas schön ist, impliziert das Vorliegen von Gründen; die Tatsache, dass etwas diese oder jene Farbe hat, nicht. Die Abhängigkeit 35. Einen kurzen Überblick für den Bereich der Ästhetik gibt beispielsweise Reicher. Vgl. Reicher (2005), S. 63-80. Zur metaethischen Debatte vgl. Miller (2003). 36. Grundlage für das Folgende sind meine metaethischen Überlegungen, die ich insbesondere in Ernst (2007) und (2008) darlege. 37. Vgl. McDowell (2001a), (2001b).

183

etwa der Farbwahrnehmung von uns ist darum gerade keine Abhängigkeit von unserer Vernunft, sondern eine Abhängigkeit von unserer Sinnlichkeit. Obwohl also in beiden Bereichen eine Abhängigkeit von uns vorliegt, handelt es sich um grundlegend verschiedene Abhängigkeiten. Dementsprechend fällt es uns auch schwer, zu sagen, welches Vermögen uns zur Erkenntnis von Schönheit befähigt, während es uns leicht fällt, zu sagen, welches Vermögen uns die Erkenntnis etwa von Farben erlaubt: das Sehvermögen. Der Vergleich ist darum letztlich wenig erhellend. Es gibt jedoch einen besseren Vergleich. Ästhetische Erkenntnis weist nämlich (ebenso wie moralische Erkenntnis) eine grundlegende Analogie mit wissenschaftlicher Erkenntnis auf. Um das zu sehen, muss man sich klar machen, dass Wissenschaft nicht einfach im Sammeln von Faktenwissen besteht. Es geht vielmehr um die miteinander zusammenhängenden Ziele, Erklärungen zu geben und Vorhersagen zu machen.38 Um diese Ziele erreichen zu können, müssen Wissenschaftler nicht einfach nur herausfinden, was der Fall ist. Sie müssen vielmehr Gesetzmäßigkeiten erkennen beziehungsweise natürliche Arten finden. Erst der Verweis auf Gesetze und natürliche Arten macht Erklärungen und Vorhersagen möglich.39 Die Erkenntnis von Gesetzmäßigkeiten und natürlichen Arten ist jedoch eine Erkenntnis von Gründen (als solchen) und damit keine rein empirische Angelegenheit. Das lässt sich besonders gut an Goodmans bekanntem „neuen Rätsel der Induktion“ deutlich machen.40 Nehmen wir an, wir stellen fest, dass alle Smaragde, die wir bisher gesehen haben, grün sind, und wir schließen daraus (induktiv) auf die Gesetzmäßigkeit, dass alle Smaragde grün sind. Daraus ergibt sich dann insbesondere die Vorhersage, dass die Smaragde, die wir in Zukunft beobachten werden, auch grün sein werden. Betrachten wir aber jetzt folgende Definition:

38. Wie man sieht, habe ich hier vor allem die Naturwissenschaften vor Augen. Meine These ist also, dass ästhetische Erkenntnis analog zu naturwissenschaftlicher Erkenntnis zu deuten ist. Vgl. dazu auch Goodman (1978), Steinbrenner (2007). Die Besonderheit meines Ansatzes besteht, wie noch deutlich werden wird, darin, dass ich das entscheidende Vergleichsmoment in der Normativität wissenschaftlicher Urteile, also in deren Bezogenheit auf Gründe sehe. 39. Tatsächlich haben wir es hier mit einem komplizierten Netz von miteinander zusammenhängenden Begriffen zu tun: Erklärung, Vorhersage, Naturgesetz, natürliche Art, Kausalität, kontrafaktische Konditionale, Bestätigung, Induktion etc. Ich greife hier der Übersichtlichkeit wegen nur einen Aspekt, nämlich die Beziehung zwischen Vorhersage, Gesetzesartigkeit und natürlichen Arten, heraus. 40. Vgl. zum Folgenden Goodman (1983), 3. Kapitel.

184

Das Prädikat „grot“ trifft auf alle Gegenstände zu, die vor dem Zeitpunkt t beobachtet wurden und grün sind, und auf solche, die nicht vor dem Zeitpunkt t beobachtet wurden und rot sind. Nehmen wir an, wir setzen für t einen Zeitpunkt ein, der gerade knapp in der Zukunft liegt. Dann gilt Folgendes: Alle Smaragde, die wir bisher beobachtet haben, waren ja grün, also waren sie der Definition entsprechend auch grot. Würden wir daraus jedoch den (induktiven) Schluss ziehen, dass alle Smaragde grot sind, so kämen wir zu der Vorhersage, dass die Smaragde, die wir erstmals nach t beobachten werden, rot sein werden, denn ein Gegenstand ist, falls er nicht vor t beobachtet wurde, nur dann grot, wenn er rot ist. Wir sind offensichtlich der Ansicht, dass unsere bisherigen Beobachtungen die Hypothese bestätigen, dass alle Smaragde grün sind, nicht jedoch die Hypothese, dass alle Smaragde grot sind. Anders gesagt: Wir betrachten die Tatsache, dass die bisher beobachteten Smaragde grün aussahen, als Grund für die Überzeugung, dass alle Smaragde grün sind, während wir die Tatsache (sic!), dass die bisher beobachteten Smaragde grot aussahen, nicht als Grund für die Überzeugung auffassen, dass alle Smaragde grot sind. Der entscheidende Punkt ist nun: Grüne und grote Smaragde sehen vor dem Zeitpunkt t, also solange wir noch Interesse an einer Vorhersage haben, gleich aus. Das heißt aber, dass wir nicht sehen können, ob die beobachteten Smaragde tatsächlich grün oder grot sind. Dennoch glauben wir (zumindest wenn wir keine Induktionsskeptiker sind), dass wir erkennen können, dass die beobachteten Smaragde grün und nicht grot sind. Wäre das nicht möglich, so könnten wir keinen entsprechenden induktiven Schluss ziehen, und eine Vorhersage wäre nicht möglich. Wissenschaftliche Erkenntnis setzt damit die Möglichkeit nichtsinnlicher Erkenntnis voraus. Diese Form der Erkenntnis kann man auf verschiedene Weise beschreiben: Man kann sagen, dass der Wissenschaftler erkennen muss, welche Hypothesen gesetzesartig sind – die Hypothese, dass alle Smaragde grün sind, gilt dann als gesetzesartig, die Hypothese, dass alle Smaragde grot sind, nicht –, oder dass der Wissenschaftler natürliche Arten finden muss – grüne Dinge gehören dann einer (abgeleiteten) natürlichen Art an, grote Dinge nicht –, und es gibt noch einige weitere mögliche Charakterisierungen. Entscheidend ist im vorliegenden Zusammenhang: Wie auch immer wir wissenschaftliche Erkenntnis beschreiben, es handelt sich nicht um sinnliche Erkenntnis, sondern um die Erkenntnis von Gründen (als solchen). Wenn wir entscheiden, ob die Smaragde grün oder grot

185

sind, tun wir das nicht allein aufgrund der Art und Weise, wie Smaragde aussehen, denn grüne und grote Smaragde sehen gleich aus – zumindest solange wie wir ein Interesse an induktiven Schlüssen haben, also vor t. Entscheidend ist hier vielmehr, wie wir das, was wir sehen, auffassen: als grüne oder als grote Smaragde. Und je nachdem glauben wir, einen Grund zu der Erwartung zu haben, dass andere Smaragde auch grün oder auch grot (also rot) aussehen werden. Damit liegt hier aber dieselbe Konstellation vor wie im Fall von ästhetischer Erkenntnis: Wenn wir von einer Sache sagen, sie sei schön, dann erkennen wir, dass die (potenzielle) Wahrnehmung dieser Sache intrinsisch wertvoll ist. Das impliziert, dass wir einen Bezug zu Gründen und damit zu unserer Vernunft herstellen. Wir sehen aber nicht, dass die Sache schön ist. Wir sehen, wie wir gesagt hatten, etwas Schönes, aber nicht die Schönheit. Entsprechend gilt im wissenschaftlichen Fall: Wenn wir von einer Eigenschaft sagen, sie sei natürlich, dann bewerten wir diese Eigenschaft ebenfalls in gewisser Weise. Wir stellen nämlich auch hier einen Bezug zu Gründen und damit zu unserer Vernunft her. Und auch hier sehen wir nicht, dass die Eigenschaft natürlich ist. Wir sehen, wie wir sagen können, natürliche Eigenschaften. Aber die Natürlichkeit selbst sehen wir nicht! Die Natur ästhetischer Erkenntnis, und damit unser Sinn für Schönheit, lässt sich somit in Analogie zu wissenschaftlicher Erkenntnis verständlich machen.41 Fassen wir zusammen: Ich habe in diesem Aufsatz versucht, die Natur des Sinns für Schönheit zu klären. Dazu habe ich zunächst das Verhältnis zwischen diesem Sinn für Schönheit und unserer „gewöhnlichen“ Wahrnehmung untersucht. Wie sich zeigte, ist ästhetische Erkenntnis nichts anderes als die Erkenntnis des intrinsischen Wertes bestimmter Wahrnehmungen (oder wie man auch sagen kann: des intrinsischen Wertes ästhetischer Erlebnisse). Damit zeigt sich der Sinn für Schönheit aber, darin dem Sinn für Moral gleich, als ein Sinn für Gründe. Da moralische Erkenntnis nicht weniger problematisch ist als ästhetische Erkenntnis, habe ich versucht, Letztere durch einen anderen Vergleich verständlicher zu machen: durch den Vergleich mit wissenschaftlicher Erkenntnis. Vielleicht wird man auf dieser Grundlage zu dem Schluss kommen, dass auch wissenschaftliche Erkenntnis alles andere als leicht zu verstehen ist. Diese Ansicht teile ich. Immerhin wäre aber gezeigt, dass ästhetische und moralische Erkenntnis 41. Tatsächlich habe ich die Analogie hier nur skizziert. Wie die Analogie zwischen Moral und Wissenschaft genau aussieht, beschreibe ich in Ernst (2008). Den Bereich der Ästhetik deute ich prinzipiell in der gleichen Weise.

186

ihrer Natur nach nicht problematischer ist als wissenschaftliche Erkenntnis. Alle drei sind aus dem gleichen Grund schwer zu verstehen: Es handelt sich um Formen der Vernunfterkenntnis. Ein wesentlicher Unterschied zwischen wissenschaftlicher und ästhetischer Erkenntnis scheint jedoch auch vor diesem Hintergrund bestehen zu bleiben: Während in Bezug auf jene kaum jemand relativistische Intuitionen hat, ist im Bereich der Ästhetik der Relativismus geradezu die Standardposition. Die Frage nach der Relativität ästhetischer Urteile habe ich in diesem Aufsatz eingeklammert. Möchte man die Natur des Sinnes für Schönheit weiter klären, muss man sie aufgreifen.42 LITERATUR Ayer, Alfred J.: Language, Truth and Logic. London: Penguin, 1990. Bender, John W.: „Realism, Supervenience, and Irresolvable Aesthetic Disputes“, in: Journal of Aesthetics and Art Criticism 54 (1996) 4, 371–381. Budd, Malcolm: Values of Art: Pictures, Poetry, and Music. London: Penguin, 1995. Carroll, Noël: „Aesthetic Experience: A Question of Content“, in: Kieran, Matthew (Hg.): Contemporary Debates in Aesthetics and the Philosophy of Art. Oxford: Blackwell, 2006, 69–97. Davies, David: Art as Performance. Oxford: Blackwell, 2004. — „Against Enlightened Empiricism“, in: Kieran, Matthew (Hg.): Contemporary Debates in Aesthetics and the Philosophy of Art. Oxford: Blackwell, 2006, 22–34. Dickie, George: Evaluating Art. Philadelphia: Temple University Press, 1988. Ernst, Gerhard: „Das Amoralistenargument“, in: Deutsche Zeitschrift für Philosophie, Berlin 54 (2006) 2, 245–260. — „Moralische Erkenntnis“, in: Bohse, Helen; Walter, Sven (Hg.): Sektionsvorträge der GAP 6, Paderborn: Mentis, 2007, 850–862. — Die Objektivität der Moral. Paderborn: Mentis, 2008. Gibbard, Alan: „Morality and Thick Concepts“, in: Proceedings of the Aristotelian Society 66 (1992), 267–283. Goldman, Alan H.: Aesthetic Value. Boulder: Westview Press, 1995. Goodman, Nelson: Ways of Worldmaking. Indianapolis: Hackett, 1978. 42. Für hilfreiche Kommentare zu früheren Fassungen dieses Aufsatzes danke ich Erich Ammereller, Karin Ernst, Geert Keil, Catrin Misselhorn, Maria E. Reicher, Reinhold Schmücker und Jakob Steinbrenner.

187

Goodman, Nelson: Fact, Fiction, and Forecast. Cambridge/MA: Harvard University Press, 1983. Hume, David: „Of the Standard of Taste“, in: ders.: Selected Essays. Oxford: Oxford University Press, 1998, 133–153. Kieran, Matthew: Revealing Art. London, New York: Routledge, 2005. Kieran, Matthew (Hg.): Contemporary Debates in Aesthetics and the Philosophy of Art. Oxford: Blackwell, 2006. Levinson, Jerrold: „Pleasure and the Value of Works of Art“, in: ders.: The Pleasures of Aesthetics. Philosophical Essays. Ithaca, London: Cornell University Press, 1996, 11–24. Lopes, Dominic McIver: Sight and Sensibility. Evaluating Pictures. Oxford: Clarendon Press, 2005. McDowell, John: „Aesthetic Value, Objectivity, and the Fabric of the World“, in: ders.: Mind, Value, and Reality. Cambridge/MA: Harvard University Press, 2001a. — „Values and Secondary Qualities“, in: ders.: Mind, Value, and Reality. Cambridge/MA: Harvard University Press, 2001b, 131-150. Miller, Alexander: An Introduction to Contemporary Metaethics. Cambridge: Polity Press, 2003. Reicher, Maria E.: Einführung in die philosophische Ästhetik. Darmstadt: Wissenschaftliche Buchgesellschaft, 2005. Schmücker, Reinold: „Funktionen der Kunst“, in: Kleimann, Bernd; Schmücker, Reinold (Hg.): Wozu Kunst? Die Frage nach ihrer Funktion. Darmstadt: Wissenschaftliche Buchgesellschaft, 2001, 13–33. — „Kunstkritik als demokratischer Prozeß“, in: Franke, Ursula; Früchtl, Josef (Hg.): Kunst und Demokratie. Positionen zu Beginn des 21. Jahrhunderts (Zeitschrift für Ästhetik und Allgemeine Kunstwissenschaft, Sonderband). Hamburg: Meiner, 2003, 99–113. — „Kann das schönste Mädchen jemals häßlich sein? Hermeneutische Spekulationen über einen Satz Adornos in weiterführender Absicht“, in: Klemme, Heiner F.; Pauen, Michael; Raters, Marie-Luise (Hg.): Im Schatten des Schönen. Die Ästhetik des Häßlichen in historischen Ansätzen und aktuellen Debatten. Bielefeld: Aisthesis Verlag, 2006, 241–257. Sibley, Frank: „Aesthetic Concepts“, in: ders. (Hg.): Approach to Aesthetics. Oxford: Oxford University Press, 2001, 1–23. Steinbrenner, Jakob: „Lassen sich ästhetische Werte erkennen?“, in: Meggle, Georg; Jäger, Christoph (Hg.): Kunst und Erkenntnis. Paderborn: Mentis, 2005a, 136-158. — „Wertung/Wert, ästhetischer“, in: Barck, K.H. u.a. (Hg.): Historisches Wörterbuch der ästhetischen Grundbegriffe Bd. 6. Stuttgart: Metzler, 2005b, 588–617.

188

Steinbrenner, Jakob: „Art Samples and Science“, in: Ernst, Gerhard; Scholz, Oliver R; Steinbrenner, Jakob (Hg.): From Logic to Art. Heusenstamm: Ontos, 2008, in Vorbereitung. Walton, Kendall L.: „Categories of Art“, in: Philosophical Review 79 (1970), 334367. Williams, Bernard: Ethics and the Limits of Philosophy. Cambridge/MA: Harvard University Press, 1985. Zangwill, Nick: „Aesthetic/Sensory Dependence“, in: ders.: The Metaphysics of Beauty. Ithaca, London: Cornell University Press, 2001a, 127-145. — „The Beautiful, the Dainty and the Dumpy“, in: ders.: The Metaphysics of Beauty. Ithaca, London: Cornell University Press, 2001b, 9-23.

189

Grazer Philosophische Studien 76 (2008), 191–198.

AN ALTERNATIVE ACCOUNT OF EPISTEMIC REASONS FOR ACTION: IN RESPONSE TO BOOTH Darrell Patrick ROWBOTTOM University of Edinburgh & University of Bristol

Summary In a recent contribution to Grazer Philosophische Studien, Booth argues that for S to have an epistemic reason to ψ means that if S ψ’s then he will have more true beliefs and less false beliefs than if he does not ψ. After strengthening this external account in response to the objection that one can improve one’s epistemic state in other fashions, e.g. by having a gain in true beliefs which outweighs one’s gain in false beliefs, I provide a challenge to it. My main objection, which I advance with the aid of several examples, is that such epistemic reasons could not motivate any action whatsoever. I close by developing an alternative account, which avoids this problem by appeal to internal considerations.

I. Like Booth, I think that there can be epistemic reasons for (and against) actions. But unlike Booth, I do not think that one has an epistemic reason to perform any action which would result in increasing one’s true beliefs and/or decreasing one’s false beliefs (when failing to perform the action would not). So what I want to challenge, precisely, is the following definition: For S to have epistemic reason to ψ, [sic] means that ψ-ing will result in S having more true beliefs and less false ones than not ψ-ing. (Booth 2006, 134) What I do not want to challenge is the further claim that: If this view about what it is to have an epistemic reason is correct, then there can be such things as epistemic reasons for action, since actions can clearly help in the attainment of the goal of believing truths and avoiding falsehoods. (Ibid.)

So I agree with Booth (2006, 143) that ‘it is at least possible for us to have epistemic reasons for action’, according to his account, and in fact would go much further: we clearly have epistemic reasons for many of our actions. But my objection is that epistemic reasons seem to be ubiquitous from this external perspective, such that it is difficult to see how glancing around a room is any more significant than building a particle accelerator. Most notably, this flies in the face of several accounts of what constitutes good inquiry—e.g. that of Popper (1959)—although Booth (ibid.) is concerned with the ways that ‘our role as enquirers might warrant (might even oblige) certain courses of action’. In the next section I will explain my criticism in detail. I will then endeavour to tackle it by suggesting that an internal account of epistemic reasons is required in place of Booth’s, and that we can look to the idea of testing hypotheses in order to provide this.

II. Before presenting my main argument against Booth’s account of epistemic reasons, I’d first like to point out a minor problem with his definition. Specifically, he says that there are two requirements for an action to be considered ‘epistemic’: doing it (rather than not) must (A) result in more true beliefs and (B) result in less false beliefs. But consider his example of going to the library, in order to consult a book. Isn’t it possible that doing this will result in more true beliefs, but not less false ones, particularly if the book concerns some area of which you have no prior knowledge? (Imagine you are a freshman philosophy student, with no previous experience in the subject, and are going to read the first chapter of an introduction to epistemology. You may have no opinions about the nature of knowledge at all.) And if you generate one new false belief through reading the book—e.g. due to a misprint—but also generate a large number of new true beliefs, then are we to say that you had no epistemic reason to read the book? This is counterintuitive, and against the spirit of Booth’s approach. More carefully, we might therefore want to require that the gain in true beliefs should outweigh the gain in false beliefs, or that the loss in false beliefs should outweigh the loss in true beliefs; see Vahid (2003), who defends a diachronic version of the truth-goal, for instance. It’s worth adding that although Booth can help himself to such a reformulation, it raises the thorny question of how we should understand ‘outweigh’. If we take each true belief to have equal intrinsic value, then ‘Some men are mortal’ might be just as good as ‘All men are mortal’, and ‘Tony Booth is a philosopher’ might be just as good as ‘Tony Booth is a male

192

philosopher who studied for his PhD at the University of Durham’.1 But if we deny that each true belief has equal intrinsic value, or indeed that true beliefs have any intrinsic value at all, then we might reach a rather different conclusion. I shall come back to this later. This brings me to my main argument against Booth’s understanding of epistemic reasons, which I will advance by asking the reader to consider the following scenarios: (α) An eccentric gentleman knocks on your door and offers to count all the blades of grass in your garden and give you the total. He doesn’t want any payment, and it won’t be any hassle to let him do so. (You’re going to be away for the day.) What’s more, you’re absolutely convinced that he’ll give you a true answer, because he’s well-known expert at counting blades of grass! (And he will give you a true answer, irrespective of what you think.) Should you take him up on his offer? (β) You’re sat on the toilet, contemplating. It’s really no hassle for you also to have a look around the room, although you have no particular desire to do so (or not), and there is no other non-epistemic (e.g. ethical) reason to do so (or otherwise). But looking around the room will result in the formation of several new true beliefs, such as ‘There was a spider present in the room while I was going to the toilet’, and ‘There was a fly trapped in the spider’s web’. Does this therefore mean that you should look around the room? It’s worth adding that changing these examples so that they concern the requirement of (A) and (B)—just in case one insists on this—is easy. In the case of (α), we need only add that you have a prior opinion about the number of blades of grass on your lawn (e.g. that it lies the range 20,000-100,000) which you will correctly classify as false if you receive the gentleman’s report. Similarly you could correct a false belief, e.g. about the colour of the towels in the room, by looking around in (β). But why are these examples important? In order to see, consider the following element of Booth’s defence of his account: [I]n the absence of any strong argument to the contrary, we must be forced to concede it possible that epistemic reasons can at times motivate action, even though it might be the case that seemingly “epistemic actions” are most often de facto best explained by practical rationality. (Booth 2006, 143) 1. It could be argued, instead, that measurement should occur over so-called ‘atomic beliefs’, e.g.: ‘Tony Booth is a philosopher’, ‘Tony Booth is male’, ‘Tony Booth has a PhD’, and so forth.

193

The foregoing examples provide the basis for a ‘strong argument’ that epistemic reasons—understood in Booth’s external fashion—cannot motivate action. Only an epistemic reason is relevant in (α) and (β), but it seems intuitively clear that it is insufficient to motivate the agent to act in one way rather than another. What’s more, this holds even though the agent is clearly aware of the chance to increase his true beliefs and/or minimise his false ones in one of the cases, namely (α). So even though I have allowed a minimal internal consideration to enter, this still does not solve the problem for Booth. It is perfectly rational, I contend, to politely refuse the offer of the gentleman in (α). In response to this objection, one might urge that these examples are not representative. But consider an argument from analogy, concerning other types of reason. If you have an ethical reason to do X, and there are no reasons—ethical or not—to do otherwise, then surely you ought to do X. Similarly, if you have a reason of self-interest to do Y, and there are no reasons—of self-interest or not—to do otherwise, then you ought to do Y. But why should this not hold when we substitute epistemic reasons, if they are genuinely reasons for action? Why should epistemic reasons operate differently from other sorts of reasons?2 At the very least, the ball now lies in Booth’s court. My challenge to him is to provide one clear example of how epistemic reasons—as he defines them—can motivate action.

III. Compare and contrast the following examples, from Booth (2006, 133), with (α) and (β): ‘performing a control experiment; asking an eminent professor their opinion on a matter on which they have expertise; conducting a conference or research seminar in a particular way.’ What’s the difference? I contend that in each of these examples, the actions are liable to be performed because the actor has the belief that performing the action will result in gaining true beliefs and/or eliminating false ones, and has the intent to gain true beliefs and/or eliminate false ones, with respect to some peculiar subject matter. (I also think that this will, ultimately, be for a practical end; more about this shortly.) But this means that ψ should be an action intentionally performed by S, and that the epistemic reason for performing the action must be accessible to S. By way of contrast, first consider having an unintentionally acquired habit which unerringly leads one to gain a couple of new true beliefs, at no cost in increase of false 2. Dancy (2005) similarly argues that we should expect all kinds of reasons to be functionally isomorphic, and uses this as an argument against moral generalism.

194

beliefs, each day.3 Does one have an epistemic reason for ‘doing’ the relevant habitual things? The answer appears to lie in the negative because one isn’t really performing those actions, any more than one grows one’s own hair. One might respond that there may be an epistemic reason not to attempt to lose the habit (or even to attempt to preserve it) if one is aware of the habit’s beneficial effects. I agree. But if one is aware of this then that one has (psychological) access to a reason not to attempt to lose (or to attempt to preserve) the habit. One has a motivation precisely because one has such access. Similarly, Alston (1989, 83–84) suggests that ‘Epistemic evaluation is undertaken from what we might call “the epistemic point of view”… by the aim at maximizing truth and minimizing falsity in a large body of beliefs’.4 But evaluation is, of conceptual necessity, an internal matter; it is only performed by agents (or perhaps by hardware/software built/designed by agents). Consider why a scientist performs control experiments, for instance. The answer, in part if not whole, is to provide a strong test of hypotheses. As Popper (1959, 418) points out, it is ‘our sincere efforts to overthrow’ which matter here, whereas incidentally acquired information does not. With the examples in the previous section in mind, however, it might still be suggested that the reason for wanting to test any given hypothesis is practical. (Remember example (α), where the possible action is intentional, and you are aware of the potential benefit, i.e. the possibility of acquiring a new true belief.) In answer to this, I contend that the practical reason operates at a different level to the epistemic one. So I accept that there must be a practical reason to want to determine whether the hypothesis is true or false, in order for the action to be motivated. (There must be some problem that one wants to solve.) But since there are many ways in which one might try to do so, there can still be an epistemic reason for performing one type of test rather than another. Thus we do not need to rely on the notion that true belief has an intrinsic value, in the way that Booth appears to, and this is a considerable advantage.5 Imagine we can think of two ways to challenge a given hypothesis. One will look for evidence e, and the other for evidence e1. But although the first test 3. Consider also that Booth doesn’t show (or suggest) that believing is not an action in its own right, so to ψ might even be to believe some particular proposition unless we accept the prior constraint and doxastic involuntarism. 4. See the extensive discussion of David (2001, 151–152), who considers a variety of approaches to epistemology and concludes that in each: ‘Truth is either explicitly referred to as a goal or aim, or it is implicitly treated as such’. 5. Alston (2005, 31) also appeals to the intrinsic value of true belief: ‘[T]heoretical investigation is often undertaken for the sake of its bearing upon practical enterprises; and even where it isn’t, theoretical results often turn out to have unforeseen practical utility. But the attainment of knowledge and understanding are also of intrinsic value …’.

195

will prove marginally more difficult to perform, in terms of time and money, it will prove more severe. (Let P(e | h & b)>>P(e | b), whereas P(e1 | h & b)>P(e1 | b), for instance. And let the probabilities be intersubjective, rather than subjective; see Rowbottom (forthcoming) for more.) Isn’t this a classic situation where we have an epistemic reason to perform the former rather than the latter? And isn’t this explicable only by appeal to internal considerations, since those probabilities reflect group degrees of belief? I believe so. In short, our reasons for action in pursuit of truth (or specific truths) and our reasons for pursuing truth (or specific truths) can come apart from an internal perspective. When you (intentionally) listen to a weather report before going out, you do so because you are hoping to form a true belief (or beliefs) about what the weather will be like. You prefer this over (only) looking up at the sky for epistemic reasons. But your motivation for wanting to have true beliefs about the weather is to be appropriately prepared, given your desires: to have an umbrella with you if it rains while you are out, so that you don’t get wet, for example. Booth’s account doesn’t appear to have the resources to deal with this.

IV. In conclusion, I shall provide a rough formulation of my alternative account. Let S be any individual or group. For S to have an epistemic reason to ψ: (1) S must believe that S is able intentionally to ψ; (2) S must believe that doing ψ will (ultimately) result in either (A*) a gain in true beliefs that will outweigh the gain in false beliefs or (B*) a loss in false beliefs that will outweigh the loss in true beliefs; (3) S must believe that doing ψ will (ultimately) result in a positive change in beliefs that will bear on a specific hypothesis, h, which S wishes to test.6 A few further remarks are in order. First, ‘ultimately’ is included in (2) and (3) in order to cover actions which only indirectly contribute to achieving the relevant results; consider fetching a piece of apparatus from a cupboard, as part of setting up a complex experiment, for instance. Second, ‘outweigh’ in (2) should be understood in a practical (or utility-related) sense. So a loss in many true beliefs that have no expected utility is acceptable, in return for the gain of one belief that does. Third, ‘positive’ is included in (3) in order to rule out cases where the 6. A possible extension, which gives epistemic reasons greater scope, is ‘or will aid in the solution of some particular problem, p, which S wishes to solve’.

196

change in beliefs is anticipated not to be truth-conducive. Else one might have an epistemic reason to perform an action which one anticipated would result in the loss of true beliefs which were highly relevant to h, although overall this loss would be expected to be outweighed by the loss in false beliefs irrelevant to h. Fourth, (1)–(3) only state ‘S must believe’, rather than ‘S must have a justified belief that’, and this is an intentional—although potentially controversial—feature of my account. Fifth, S’s beliefs need not be true, and doing ψ could be disastrous for S (in more than one respect)! Finally, it is worth mentioning that this account might be incomplete in so far as it precludes epistemic reasons relating to syntactic elegance, explanatory power, and so forth. My own view, for what it is worth, is that these are practical virtues; that a theory with these qualities need not be more verisimilar than a competitor without, when those theories are otherwise equally as good, and that they are not epistemic goals in their own right. In saying this, I accept that the more elegant of two theories may be the easier to understand as well as to use. I suspect that there are many objections to this account which I have not anticipated, and would not be surprised to find that it is fundamentally flawed. What’s more, the whole notion of epistemic reasons for action might be thoroughly misguided. Nonetheless, I anticipate that my criticisms of Booth’s approach will remain pertinent.7

REFERENCES Alston, W. P. 1989: “Concepts of Epistemic Justification”, in Alston (ed.) Epistemic Justification: Essays in the Theory of Knowledge (Ithaca, NY: Cornell University Press). — 2005: Beyond “Justification”: Dimensions of Epistemic Evaluation (Ithaca, NY: Cornell University Press). Booth, A. R. 2006: “Can there be Epistemic Reasons for Action?”, Grazer Philosophische Studien 73 162–173. Dancy, J. 2005: “Moral Particularism”, in Zalta (ed.) The Stanford Encyclopedia of Philosophy, URL: http://plato.stanford.edu/archives/sum2005/entries/moralparticularism. David, M. 2001: “Truth as the Epistemic Goal”, in Steup (ed.) Knowledge, Truth and Duty: Essays on Epistemic Justification, Responsibility and Virtue (Oxford: Oxford University Press). 7. For helpful criticisms of earlier versions of this paper, I am grateful to Peter Baumann, Tony Booth, and an anonymous referee.

197

Popper, K. R. 1959: The Logic of Scientific Discovery (New York: Basic Books) Rowbottom, D. P. (forthcoming): “Intersubjective Corroboration”, Studies in History and Philosophy of Science. Vahid, H. 2003: “Truth and the Aim of Epistemic Justification”, Teorema XX/3, 83–91.

198

Grazer Philosophische Studien 76 (2008), 199–208.

FURTHER REMARKS ON PROPERTY DESIGNATORS AND RIGIDITY (REPLY TO LÓPEZ DE SA’S CRITICISMS) Benjamin SCHNIEDER Humboldt-Universität zu Berlin

Summary Are all canonical property designators (i.e. nominalizations of predicative phrases) rigid? Dan López de Sa recently criticized the arguments I gave for an affirmative answer to that question. The current article rebuts López de Sa’s objections. 1. Introduction Singular terms for properties can be derived from predicative phrases (i.e. general terms and/or predicates) by means of nominalization. Let me call such property designators canonical (they exhibit several forms, most notably ‘F-ness’, ‘being F ’, ‘to be F ’, and ‘the property of being F ’). In a previous article I argued that all canonical property designators are rigid in Kripke’s sense.1 Dan López de Sa recently tried to show that my arguments do not achieve their goal.2 But I am not convinced; López de Sa’s criticisms can be rebutted. For a start, let me briefly review what exactly is at issue between López de Sa and me. We agree that there are flexible property designators. Properties can be picked out by definite descriptions which describe them by some contingent features (thus, ‘the most famous virtue of Socrates’ flexibly designates wisdom). We disagree over the question whether all canonical property designators (which are nominalized expressions) are rigid. The crucial question is whether nominalization always rigidifies or not. We again agree that many examples of canonical property designators are unquestionably rigid, for instance: ‘wisdom’, ‘the property of being human’, or ‘being a cubic object’. Dissent sets in when more convoluted cases are considered. Thus, 1. See Schnieder (2005). 2. See López de Sa (2006).

CRT ‘having the colour of ripe tomatoes’ may appear to be different, because the term may appear to denote a colourproperty, viz. the property of being red. If, but only if, this assumption were correct, we should classify the designator as non-rigid: ripe tomatoes could have been blue, and if CRT designates the property of being red when used to talk about the actual world, it should designate the property of being blue when used to describe a counterfactual situation in which tomatoes are blue. So, whether CRT is rigid or not hinges on the truth-value of (1) ‘having the colour of ripe tomatoes’ designates the property of being red. If (1) is false, we lack any reason to deny that CRT is rigid; if (1) is true, however, we have all the reason in the world to regard CRT as non-rigid. (The example can be seen as representative for a number of other cases; ‘having the most common vice,’ or ‘having the shape of the sun’ may for analogous reasons appear to be flexible.) Now, while López de Sa thinks that (1) is true and CRT is non-rigid, I hold the opposite: (1) is false and CRT is rigid. It does not designate a colour-property, but rather a relational property which is possessed (in any possible world) by exactly those entities that stand to ripe tomatoes in a certain relation, namely in that of sameness of colour (i.e. of the colour that ripe tomatoes have in the respective possible world ).

2. Defence In my original article I formulated six arguments for the thesis that CRT does not designate the property of being red (but designates, rigidly, a relational property). I commenced with three arguments which, as I frankly admitted, are certainly not decisive (hence, I am sympathetic with López de Sa’s comments on these arguments).3 So let me just focus on the three arguments that I regarded as conclusive then. One of them was this: the truth of (1) is at odds with the truth of (2) Having the colour of ripe tomatoes is a property my T-shirt has; but it would not have had this property, if ripe tomatoes had been brown instead of red. 3. Cp. Schnieder (2005: 232–33).

200

The phrase ‘this property’ takes up the reference of ‘having the colour of ripe tomatoes’. If the latter term denoted the property of being red, then ‘this property’ would denote it too; if we understood the terms this way, we should dismiss (3) as false, because it would then imply the following falsity: (2*) My T-shirt would not have had the property of being red if ripe tomatoes had been brown instead of red. But—it seems to me—we accept (2) as true. That we do so is one of the data I relied upon. Take away my data and you take away my point: if someone has the intuition that (2) is false, then he need not be moved by my argument. I am not sure whether López de Sa has the intuition that (2) is false. He does not explicitly say so.4 But if he has it, I can not do anything about that; perhaps, intuitions can vary in this case. If each of my other two arguments equally depended on a premise that I accept as intuitively correct while López de Sa does not, we should acknowledge our diverging intuitions and bury the hatchet. However, in criticising my other arguments, López de Sa did not just deny my starting points; rather, he accepted them but tried to show that they do not support my position. This is a more substantial disagreement, which I will now address. a. Modal properties of properties For my next argument, I changed the example: instead of CRT, I used the designator VS being the virtue that Socrates was most famous for. López de Sa apparently is somewhat uneasy with the new example; he writes: It is not clear what one should say about the expressions ‘being (identical to) the virtue that Socrates was most famous for’ and ‘being (identical to) the Head of the Catholic Church.’ But suppose […] that one holds they flexibly signify (respectively) the property of being wise and of being Ratzinger in actuality—as Schnieder seems to think [I] should do. (López de Sa 2006: 228) 4. What he does say (2006: 227) rather sounds as if he accused me of committing a petitio. But this criticism is ungrounded; I rely on (2) as an intuitive linguistic datum that may reveal facts about the semantics of CRT. If this were not legitimate, then López de Sa’s own argument (see below) for the non-rigidity of CRT would equally be a petitio. I take it that neither is the case.

201

This is, in fact, what I think. López de Sa only agrees under reserve. But why? He correctly points out that the ‘being’ in VS is not the nominalization of the copulative ‘is’, but rather of the ‘is’ of identity—yet I cannot see why this should make any difference. The copulative ‘is’ was not mentioned when López de Sa argued for the non-rigidity of CRT. Rather, he relied on the intuition that CRT designates the property of being red, and he took the intuition to be supported by the intuitive truth of (3) Having the colour of ripe tomatoes is being red.5 However, it is equally intuitive to think that ‘being (identical to) the virtue that Socrates was most famous for’ designates the property of being (identical to) wisdom, and this intuition may seem to be supported by the intuitive truth of (4) Being the virtue that Socrates was most famous for is being wisdom.6 The copulative ‘being’ is not required to make a canonical property designator appear to be flexible. Rather, it is the presence of a flexible definite description as a part of a canonical property designator that accounts for the apparent flexibility: CRT contains the flexible ‘the colour of ripe tomatoes’ and VS contains the flexible ‘the virtue that Socrates was most famous for’. The flexibility of these expres5. I agree that (3) is intuitively acceptable; but this only proves López de Sa’s point if (3) is to be taken as an identity statement. However, I produced a number of examples in my earlier article (2005: 230f.) to show that constructions of the form ‘to ϕ is to ψ’ or ‘ϕ-ing is ψ-ing’ are not normally used to make identity statements. Whoever believes that lying is sinning or that to be late is to be impolite is not committed to the absurd beliefs that sinning is lying or that to be impolite is to be late (which he would be committed to if the statements were statements of identity). In his article (at p. 223f.), López de Sa agrees with this observation but thinks that statements such as (3) can be read as identity statements and that the intuitions supporting (3) do support it in that reading—but he does not give any argument for this claim. Now here is an argument for the contrary position: if intuitions in favour of (3) would clearly support it in the identity reading, then we should have equally strong intuitions in favour of the following explicit identity statement: (3*) The property of having the colour of ripe tomatoes is no other property than the property of being red. But while most people readily accept (3), they hesitate to accept (3*). Even López de Sa admits that he has no firm intuitions about (3*). But then he cannot use intuitions about (3) as a substitute for intuitions about (3*), and thus he loses the primary support for his view. 6. Moreover, notice that our original case of CRT does not contain the copulative ‘is’ or ‘being’ either (it does not contain any ‘is’ or ‘being’); its main verb is ‘have’. And notice finally that I can simply give López de Sa the copulative ‘being’. Just replace the ‘being’ in VS with ‘being identical to’; you get: ‘being identical to the virtue that Socrates was most famous for.’ Here, the ‘being’ is not that of identity but rather has its copulative use.

202

sions explains why we may think the whole terms flexibly denote the property of being red and the property of being (identical to) wisdom respectively. The main verb of the canonical property designator does not play any important role for these intuitions; apart from ‘being’, ‘having’, or ‘being (identical to)’, it may also be some other verb, like ‘seeing,’ ‘striving for,’ ‘lacking,’ etc. Thus, ‘seeing the colour of the sky’ may seem to denote the property of seeing the colour blue, ‘striving for the virtue that Socrates was most famous for’ may seem to denote the property of striving for wisdom, etc. I cannot see any reasons to follow the intuitions in one case but not in the other. Now, López de Sa criticizes my argument independently of any misgivings about the example. Let me start with the argument: I take sentence (6)

Being the virtue that Socrates was most famous for is only a contingent feature of wisdom.7

to be (rather obviously) true. (6) can be spelled out as follows: (6*) Being the virtue that Socrates was most famous for is a feature of wisdom, but it is not necessary that wisdom possesses this feature. This statement is backed by more specific counterfactual statements, such as: (6**) Being the virtue that Socrates was most famous for is a feature of wisdom. But wisdom would not have had this feature, if Socrates had been best known for his patience. That (6) is true is a datum I work with. Without it, I have no argument. Fortunately though, López de Sa agrees with me on the truth of (6). Now if (6) is correct, we can prove that ‘being the virtue that Socrates was most famous for’ does not designate the property of being wisdom. For if it did, we would have: (7)

Being wisdom is only a contingent feature of wisdom.

(7) can be spelled out as:

7. I use ‘contingent’ instead of my original ‘accidental’ in order to avoid debates about the proper understanding of ‘essential’ / ‘accidental’ here. The current argument needs only the modal notion of contingency.

203

(7*) Being wisdom is a feature of wisdom, but it is not necessary that wisdom possesses this feature. And this is obviously false. López de Sa thinks that I am not allowed to reason like this, because there is an allegedly parallel case in which the same reasoning would lead to an absurd conclusion. He uses the following example to show that something is at odds with my reasoning: (8)

Being the head of the Catholic Church is only an accidental feature of Ratzinger.

López de Sa points out, correctly, that (8) is true and that this does not show that (9)

Ratzinger is not the head of the Catholic Church.

So far so good, but what follows? Why should I be committed to this nonidentity? Sentences (6) and (8) are structurally analogous. In both sentences, we have a complex property designator: VS ‘being the virtue Socrates was most famous for,’ and HCC ‘being the head of the Catholic Church’. Both these designators contain as a part a flexible designator: VS contains ‘the virtue Socrates was most famous for’ and HCC contains ‘the head of the Catholic Church’. The latter expression designates Ratzinger, and the parallel assertion for (3) is: ‘the virtue Socrates was most famous for’ designates wisdom. And this is something I do not deny; instead, I presupposed this (so, I presupposed: wisdom = the virtue that Socrates was most famous for). I concluded from (6) that the property of being wisdom is not identical to the property of being Socrates’s most famous virtue. So, I am obliged to draw the parallel conclusion from (8)—which is: (10) The property of being the head of the Catholic Church is not identical to the property of being Ratzinger. That I can perfectly live with this result should be no surprise. Moreover, this entailment is perfectly compatible with the truth of

204

(11) Ratzinger is the head of the Catholic Church, just as the first entailment is compatible with the truth of (12) Wisdom is the virtue that Socrates was most famous for. Pace López de Sa, I am not in any way obliged to infer (9) from (8), just as I am not obliged to infer from (6) that wisdom is not the virtue Socrates was most famous for. Now López de Sa did not point out any specific failure in my reasoning, but merely raised the objection that if I were entitled to conclude as I do, then we would also be entitled to conclude (9) from (8), which cannot be the case. But we saw that he confuses two related statements of identity; what parallels the conclusion I drew from (6) is (10), not (9). But while (9) is obviously wrong, (10) is just what I hold. López de Sa’s objection therefore is a non-starter and my argument still stands. b. The meaning of canonical property designators My final argument has a virtue that the others lacked: while they only give a reason to think that CRT is rigid, the last argument shows the reason why it is rigid. The argument is based on more general reflections about the semantics of canonical property designators. I proposed an analysis of such terms according to which the following schema is valid:8 (Prop) Being F (to be F, F-ness) is the property which is essentially such that it is possessed by all and only Fs. This, I thought, straightforwardly explains why these designators are rigid. López de Sa does not dispute the correctness of the schema; but he denies that (Prop) conflicts with the alleged flexibility of CRT:9 Suppose then that ‘having the color of ripe tomatoes’ flexibly signifies being red in actuality. Hence having the color of ripe tomatoes is being red. Now being red is the property which is essentially such that it is possessed by all and only red things, and red things are all and only the things that have the color of ripe tomatoes. Hence, having the color of ripe tomatoes is the property which is essentially such that it is possessed by all and only the things 8. For a detailed defence of this analysis see Schnieder (2006). 9. López de Sa uses ‘signify’ where I use ‘designate’; this is but a difference in terminology.

205

that have the color of ripe tomatoes. Hence the instance of [(Prop)] is true. (2006: 228f.) López de Sa apparently succumbs to a notorious modal fallacy here. From claims of the following forms, P1 a is essentially such that all (and only) Fs are R-related to it, and P2 All and only Fs are Gs, we are not entitled to conclude that C

a is essentially such that all (and only) Gs are R-related to it.

Take the following example: P1 Having kidneys is essentially such that all and only kidney-owners have it. P2 All and only kidney-owners are heart-owners. C Therefore: Having kidneys is essentially such that all and only heart-owners have it. The conclusion is a non sequitur, and so is the conclusion in López de Sa’s case. In fact, it is provably false; notice first that (13) Redness (the property of being red) is not essentially such that all things that have the colour of ripe tomatoes have it. For, had ripe tomatoes been brown and had there furthermore been some brown things, then some things with the colour of ripe tomatoes had not possessed redness. So, on López de Sa’s assumption that (1*) Redness = the property of having the colour of ripe tomatoes, it follows that:10 (14) Having the colour of ripe tomatoes is not essentially such that all things that have the colour of ripe tomatoes have it. 10. The inference is an uncontroversial application of Leibniz’s Law; the relevant singular terms do not occur inside the scope of any non-extensional operator.

206

There is no way of reconciling the alleged flexibility of CRT with the validity of (Prop): Since the latter is an outflow of a general analysis of the semantics of canonical property designators, we know that CRT is rigid. My fifth and sixth argument therefore withstand López de Sa’s criticisms. He accepts the data the arguments work with and he fails to show that any fallacies are involved in my reasoning. Hence, I can still conclude that all canonical property designators are rigid.

3. Concession Let me finally come to an aspect of my earlier work that is in need of an amendment; it may come as a little surprise: I am now willing to admit that CRT (‘having the colour of ripe tomatoes’) denotes the property of being red. Or at least that it may be used to denote the property of being red. But I reckon that in so far as it may be so used, this does not speak against its rigidity. Take a further look at CRT: as remarked before, it contains a definite description, ‘the color of ripe tomatoes’. Now Keith Donnellan famously argued that definite descriptions have what he called a referential use, in which they do not only denote something, but in which they are used to refer (in an ambitious sense of ‘refer’) to something (which may or may not be the entity denoted by the term).11 If his idea is correct, then we should be able to use the description ‘the colour of ripe tomatoes’ in order to refer to the colour red. If the description is thus used while it occurs as a part of CRT, then the whole designator seems to denote the property of having the color red (or, equivalently, the property of being red). Does this concession force me to give up my earlier position? No. The reason is that in the described use, CRT would not be a flexible designator. It would still be rigid, but it would rigidly designate the property of having the color red. For assume we use CRT as described, i.e. as meaning something like: having this colour (where ‘this’ refers to the colour red). If we then evaluate CRT—thus used—with respect to some possible world, we must hold fixed the referent of the referentially used description; thus the expression will rigidly designate the property of being red (of having this colour). This is why I can admit that CRT may allow for a use in which it after all denotes the property of being red without retreating from the position I defended before.12 Acknowledging this possible 11. Cp. Donellan (1966). 12. Anyway, this case only works under some controversial assumptions. It only constitutes a semantic phenomenon if the referential use of definite descriptions is a semantic phenomenon itself – which I would doubt it is, following Kripke (1977).

207

use of CRT may, however, account for some of the intuitions that make López de Sa regard CRT as flexible.13

REFERENCES Donnellan, Keith (1966): ‘Reference and Definite Descriptions’, Philosophical Review 75, 281–304. Kripke, Saul (1977): ‘Speaker’s Reference and Semantic Reference’, Midwest Studies in Philosophy 2, 255–76. López de Sa, Dan (2006): ‘Flexible Property Designators’, Grazer Philosophische Studien 73, 221–30. Schnieder, Benjamin (2005): ‘Property Designators, Predicates, and Rigidity’, Philosophical Studies 122, 227–41. — (2006): ‘Canonical Property Designators’, The American Philosophical Quarterly 43, 119–32.

13. For valuable comments I would like to thank Miguel Hoeltje, Dan López de Sa, Alexander Steinberg, and two anonymous referees of this journal.

208

ESSAY-WETTBEWERB FÜR STUDIERENDE

ESSAY COMPETITION FOR STUDENTS

Seit dem Jahr 2007 vergibt die Gesellschaft für Analytische Philosophie (GAP) gemeinsam mit den Grazer Philosophischen Studien jährlich einen Essay-Preis. Die Preisfrage für das Jahr 2007 lautete:

Since 2007, the Gesellschaft für Analytische Philosophie (GAP), together with the Grazer Philosophische Studien, annually awards an essay prize. The prize question to be answered in 2007 was:

KÖNNTEN WIR VOLLSTÄNDIG DETERMINIERT UND DOCH FÜR UNSERE HANDLUNGEN VERANTWORTLICH SEIN?

COULD WE BE COMPLETELY DETERMINED AND YET RESPONSIBLE FOR OUR ACTIONS?

Das Gerechte und Ungerechte, das Lob und der Tadel, die Strafe und der Lohn finden auf nothwendige Handlungen keine Anwendung. (Leibniz, Theodizee)

Justice and injustice, praise and blame, punishment and reward cannot attach to necessary actions. (Leibniz, Theodicy)

Teilnahmeberechtigt waren Studierende der Philosophie, Wissenschaftstheorie, Logik oder Philosophie/Ethik, die mindestens vier Semester absolviert oder ihr Studium vor nicht mehr als einem Jahr durch einen B.A.- oder M.A.Abschluss oder ein anderes Examen beendet haben. Einzureichen waren Beiträge in deutscher oder englischer Sprache von maximal 3000 Wörtern.

Eligible were students of philosophy, philosophy of science, logic, or ethics who completed at least four semesters or finished their studies less than a year ago with a B.A. or an M.A. degree or an equivalent. They were invited to submit an essay, written in German or English, of no more than 3000 words.

Das Echo auf die Ausschreibung war überwältigend. Eingereicht wurden 59 Essays, wovon 57 die formalen Anforderungen erfüllten. Elf Essays zogen die Mitglieder der Jury (Johannes Brandl,

The response to the invitation was overwhelming. Of the 59 essays submitted, 57 met the formal criteria. The members of the jury (Johannes Brandl, Hans Rott, Thomas Spitzley and Ulla

Hans Rott, Thomas Spitzley und Ulla Wessels) in die engere Wahl. Bei der Entscheidung für die drei preisgekrönten Essays und deren Reihung orientierte sich die Jury an drei Kriterien: (i) der Originalität der Idee, (ii) der argumentativen Schärfe bei deren Entwicklung; und (iii) der Eleganz des Stils.

Wessels) shortlisted eleven essays. In selecting and ranking the three awardwinning essays they followed three criteria: (i) the originality of the idea; (ii) the argumentative rigour in developing it; and (iii) the elegance of style.

Die Jury und die Organisatoren gratu- The jury and the organisers congratulieren den Gewinnern. late the winners.

210

Grazer Philosophische Studien 76 (2008), 211–218.

A NEGLECTED ARGUMENT FOR COMPATIBILISM Lars DÄNZER Universität Bielefeld First prize of the essay competition for students of philosophy 2007, offered by the Gesellschaft für Analytische Philosophie and the Grazer Philosophische Studien

Is the truth of determinism compatible with our being responsible for (at least some of ) our actions? Among the positions which urge a negative answer to this question, the most common and widespread one is characterized by two theses: (1) It is a necessary condition for an agent’s being morally responsible for his actions that he performed them of his own free will; (2) Determinism is incompatible with our acting of our own free will. Together, these claims imply that if determinism is true, we are never morally responsible for our actions and thus anyone who wants to defend the compatibility of determinism and responsibility must reject at least one of these two claims. While there are philosophers today who adopt the strategy of rejecting (1), I will pursue in this essay the more standard strategy and focus on the second claim, i.e. on the issue of the compatibility of free will and determinism. A notable feature of the debate about the compatibility of free will and determinism is that it is dominated almost exclusively by the arguments for incompatibilism.1 The compatibilist appears to be in an entirely defensive position, trying to rebut the incompatibilist’s arguments without offering any positive arguments for his own position. My essay is intended as a contribution to the correction of this situation. I think that there is an argument for compatibilism that has not received the attention it deserves. The argument that I have in mind and that I will formulate and discuss below is related rather closely to a type of argument to be found in the literature (see below p. 216, fn. 5) but has never, as far as I know, been formulated in quite the way suggested here. It seems to me that the argument has everything required to make it a good “opening argument” for 1. In what follows I will use the terms “compatibilism” and “incompatibilism” to refer to positions concerning the (in)compatibility of determinism and free will (not responsibility).

211

compatibilism. That is, it is the sort of argument that a compatibilist can offer as a first move in the debate with the incompatibilist, thereby placing on his opponent the burden to rebut it in order to defend the incompatibilist position, very much in the way in which the consequence-argument is an “opening argument” for incompatibilism. A good opening argument of this sort is characterized by two things: It both seems to be valid and is based on premises that are highly appealing from a pre-theoretic point of view. I hope I will succeed in making good the claim that this is true of the argument presented below. Let me make two further remarks about the dialectical status of the argument. First, it does not rest on any particular (compatibilist) analysis of free will and thus can be employed by any compatibilist, no matter what his favoured analysis might look like.2 For the same reason the argument is not vulnerable to objections directed against any particular compatibilist analysis. Secondly, the argument does not, of course, relieve the compatibilist from developing a compatibilist analysis of free will nor from showing what is wrong with the incompatibilist’s arguments. It can only be one part of a compatiblist’s position and defending it from criticism will require him to engage with many of the well known and hotly debated issues in the field. The argument I am going to present is intimately related to a key methodological point that is stressed by most if not all compatibilists. The point is this: When we think about the issues of free will and determinism the important question is whether or not determinism is compatible with our “acting of our own free will” as we commonly understand this phrase. That is, it is the compatibility of determinism with our ordinary and everyday concept of an action performed of free will that is relevant. Therefore, a vital point is to pay close attention to what we ordinarily mean when we say of an action that it was performed of free will. As will become clear, this methodological point lies at the very heart of the argument. Here, then, is the argument: (P1) There are cases of the following type T: — A person A has just finished graduate school and is offered a job by two different companies X and Y; — A is not put under pressure or threatened by her parents nor by anyone else to decide in a particular way; — A does not consume any psychoactive drugs and does not suffer from any addiction that would influence her decision; 2. In this respect, the argument differs from most other compatibilist arguments, most notably from what is sometimes called the “Ethics argument” or the “conditional analysis argument” for compatibilism.

212

— A does not suffer from a brain damage nor is her brain affected in any other pathological way; — A is also not subject to any highly unusual condition like the following: She has not been brain washed, she is not hypnotised, not psycontrolled by martians, there is no controlling device implanted in her brain, and so on; — A engages in a decision process with the following features: A is well informed about the respective jobs; she takes herself plenty of time to consider her wishes and goals, to think about the different options and to weight reasons for and against each other; and all these have an influence on A’s decision; — this decision process results in A’s decision to accept the job offered by company X. (P2) The existence of cases of type T is compatible with determinism. (P3) Cases of type T are perfectly clear instances of actions performed of the agent’s own free will. Therefore: (C1) There are cases of actions performed of the agent’s own free will. (from P1 and P3) (C2) The existence of actions performed of the agent’s own free will is compatible with determinism. (from P2 and P3). Obviously, there are actually two different arguments here. There is firstly the argument from (P1) and (P3) to (C1) and secondly the argument from (P2) and (P3) to (C2). Since the focus of this essay is the issue of the compatibility of free will and determinism, I will restrict my attention to this second argument. However, even given this restriction there is still good reason to include (P1) into our considerations. On the one hand (P1) makes it clear that cases of type T are intended as cases of a familiar, actually occurring—as opposed to a merely possible or imagined—kind. On the other hand, (P1) can serve as a reason to accept (P2), as I will argue below. Let’s consider (P1) first. It is an existential statement which asserts that there are or occur cases of a certain specified type T. As such, it is clearly both contingent and a posteriori. T is characterized by a number of conditions which are severally necessary and jointly sufficient for instantiation of T and which I will call the elements of T. The characterization of type T is intended to capture a perfectly normal and standard kind of case which we believe to occur every day. Some of the elements in T (in particular the reference to graduate school and the job decision) are, of course, arbitrary, i.e. replacing them with something else would not make a difference to the argument. On the other

213

hand, there is a point in including some such element in T, rather than trying to make it more general and unspecific, namely to keep the cases as concrete as possible. I take it that the truth of (P1) is plain and uncontroversial—or at least it is so at first sight. I think it is a very deeply entrenched part of our everyday or folk theory about the world that there are indeed cases of type T. Of course, it might turn out that we are wrong about this—but I find it very hard to see how this might be the case.3 In the context of the present debate I can think of only one reason why an incompatibilist, more particularly a hard determinist, might want to deny (P1), namely that he believes that determinism entails that there are no cases of type T, i.e. that she denies (P2). So let’s turn to (P2). (P2) is the claim that the existence of cases of type T, i.e. the truth of (P1), is compatible with the truth of determinism. Would an incompatibilist deny (P2)? I don’t think so. There are two points worth mentioning here. First, I cannot see any reason why we should think that the truth of determinism might imply that there are no instances of type T. While there may be prima facie reasons to believe that determinism is incompatible with our acting or deciding freely, I don’t see any reason why this worry should also apply to cases of type T. Note that the characterization of T does not contain any of the (alleged) conditions for free actions that are commonly claimed by incompatibilists to be incompatible with determinism, namely (a) that the actor could have chosen or acted otherwise and (b) that the actor is the “ultimate source” of his action (see e.g. McKenna 2004: 5–8). Thus, there seems to be nothing in the incompatibilist’s position that would provide grounds for rejecting (P2). Secondly, notice that if the incompatibilist were to deny (P2), then his position would turn out to be much more radical and to have much more far reaching consequences than is generally assumed. Indeed, I think that quite a strong positive reason for a hard determinist to accept (P2) is simply that he would otherwise commit himself to the denial of (P1). As I said, I think that to deny (P1) is to deny something which seems just too obviously true and thus any such denial would seriously increase the burden on the hard determinist to justify his position. Let us turn to (P3). The crucial point about this premise is that it is intended as reporting a semantic or conceptual “intuition” or “intuitive judgment”. To bring this out more clearly, we could rephrase it as “Cases of type T are perfectly clear instances of what we (ordinarily) mean by ‘acting of one’s own free will’”. The point of saying that cases of type T are “perfectly clear” instances of free acts is simply to emphasize that our intuitions concerning these cases 3. One way in which (P1) might turn out to be false, for instance, is that eliminativism about the mental turned out to be true.

214

are very clear-cut and unambiguous; that is, we neither feel that they need to be specified in more detail in order for our semantic intuitions to be clear, nor do we feel that they are some sort of borderline cases. Furthermore, the intended sense of the phrase “perfectly clear instance” is, of course, such that to be a perfectly clear instance of some type entails being an instance of this type. I take it that this proviso should eliminate any worry that the argument is invalid. How is (P3) justified? As a semantic intuition its justification is allegedly a priori and not grounded in inference from other premises. This is not the place to argue for and justify the use and indispensability of such intuitions in philosophy (for such an argument see e.g. Jackson 1998). It seems clear to me that intuitions of this sort play a fundament role in all areas of philosophy where the analysis of an ordinary notion is involved.4 In addition, however, semantic intuitions of the kind reported in (P3) are frequently and without hesitation relied on in the free will debate as well, in order to support or defeat claims to the effect that a certain condition or conjunction of conditions is necessary or sufficient for an act’s being free. For instance, incompatibilists make essential use of such intuitions when they are trying to refute compatibilist analyses of free action, e.g. the conditional analysis. Other cases in point are the Frankfurt-cases and Dennett’s character-examples. What is true is that the incompatibilists normally use “negative” intuitions, i.e. intuitions to the effect that a certain case is not an instance of a free act. However, this is simply due to their argumentative goals and it can hardly be held that the use of “negative” intuitions is somehow more legitimate than the use of positive ones. Thus, there is surely nothing wrong in general with the use of semantic intuitions in arguments in the free will debate. Now, to say that there is nothing wrong in general with such intuitions is of course not to show that there is nothing wrong with the intuition reported in (P3) in particular. I will return to this point but first I want to dispel another worry about my argument. Almost certainly it will have occurred to some readers that the foregoing argument (at least the one from (P1) and (P3) to (C1)) is simply a version of what is known as “the paradigm case argument” (“PCA” for short), which has long been shown to be fatally defective. I must restrict myself to some short comments on this topic. First, the question of whether the argument is an instance of the PCA has no simple answer because there is no such as thing as “the” PCA. What is grouped together under this label by different people is a “family” (in the Wittgensteinian sense) of related arguments which, however, differ from each other in crucial respects (see von Savigny 1981). There are, no 4. The most famous examples in this respect are maybe the debates about the notions of knowledge (Gettier) and meaning (Grice).

215

doubt, some reasons to include the argument presented above in this family.5 At the same time, and more relevantly to its assessment, the above presented argument differs in crucial respects from the arguments most commonly associated with the label “PCA”, including the argument discussed by van Inwagen (1983), with the result that it is not open to the standard objections against the PCA (see for instance Ayer 1963, Passmore 1966, Chapter 6 ) nor to those of van Inwagen (1983: 108–112). To see this, it is actually enough to note that both the standard objections as well as van Inwagen’s objection were designed to show that the PCA is invalid by showing that if it were valid then so would be certain other arguments which are obviously not valid. Since the argument given above is no doubt valid, objections of this sort must be beside the point. The crucial difference which makes the argument immune to these objections lies in the fact that the notion of a “perfectly clear instance of F” as employed in the argument does not involve any dubious assumptions about how the meaning of “F” could be learned, defined or taught, as these are commonly taken to be involved in the notion of a “paradigm case”. Let us return to (P3). For all I have said so far it seems clear that the incompatibilist has to reject (P3) if he wants to escape the conclusion of the argument. And indeed, while semantic intuitions may have a strong prima facie justification they are not indefeasible, i.e. we can be brought to change our judgement about a given case in the light of further considerations (see e.g. Jackson 1998: 35–36). To consider in detail how an incompatibilist might challenge (P3) and what the compatibilist’s reply should look like would be to open another chapter which is beyond the scope of this essay. If I have managed to convince the reader that it is worthwhile to pursue this question further, I have achieved the aim of this essay, namely to make good the claim that the argument presented is in fact an interesting opening argument for compatibilism that deserves more attention than it has got so far.6 I want to conclude my essay by pointing out that independently of what the final verdict on the argument may be, it does have the merit to bring into sharper 5. The argument from (P1) and (P3) to (C1) fits the description that Antony Flew, who is commonly considered to have applied the PCA to the free will debate in the first place, gave of the PCA in one of his later publications on this topic (see Flew 1966: 264). Furthermore, the addition of something like premise (P2) to the core “PCA” (i.e. to (P1) and (P3)) is just what Peter van Inwagen (1983: 107–108) has suggested in his reconstruction of the PCA as an argument for compatibilism. 6. One more worry about (P3) might be that it simply begs the question against the incompatibilist. However, I cannot see why this should be so. First, notice that (P3) by itself does not entail anything about the compatibility issue. Secondly, it is hard to see how a semantic intuition could be question begging at all; for its prima facie “justification” does not depend on any argument whatsoever.

216

focus what is at issue between the compatibilist and the incompatibilist. By making clear the incompatibilist’s position with respect to cases of type T, it helps us to see more clearly what is and what is not claimed by the incompatibilist to follow from the truth of determinism. And this, in turn, might be relevant to the evaluation of several issues in the debate. Here is what the compatibilist and the incompatibilist agree upon. They agree that the truth of determinism does not imply that there are no cases of type T and, therefore, that for all conditions Ci fulfilled by cases of type T, it does not imply that if determinism is true, then Ci is never fulfilled. It seems to me that this is an important point to keep in mind when we ask ourselves how to assess the claims and arguments of the incompatibilist, for instance when thinking about what does and what does not follow from the consequence argument. And here is where they disagree. What the incompatibilist claims and the compatibilist denies is that our ordinary concept of a free act or decision involves a condition C such that C is not fulfilled by cases of type T. Note that this implies, in particular, that the sense in which the could-have-done-otherwise condition and the ultimate-source condition referred to above are to be understood according to the incompatibilist is such that being of type T is not sufficient for fulfilling these conditions. This, again, might be relevant when thinking about the could-have-done-otherwise-condition in particular. For it is strongly tempting to think that in the relevant, everyday sense of this condition, it is fulfilled by cases of type T. In fact, it seems that one could construct an argument perfectly analogous to the one discussed in this essay for the compatibility of the CDOcondition and determinism.

REFERENCES Ayer, A. J. (1963): “Philosophy and Language”, in: The Concept of a Person and other Essays, New York: St. Martin’s Press. Flew, Antony (1966) “Again the Paradigm”, in: P. K. Feyerabend & G. Maxwell (eds.): Mind, Matter, and Method, Minneapolis: University of Minnesota Press. Jackson, Frank (1998): From Metaphysics to Ethics, Oxford [et al.]: Clarendon Press. McKenna, Michael (2004): “Compatibilism”, The Stanford Encyclopedia of Philosophy (Summer 2004 Edition), Edward N. Zalta (ed.), URL = . Passmore, John (1969): Philosophical Reasoning, New York: Basic Books.

217

Savigny, Eike von (1981): „Das so genannte ,Paradigm Case Argument‘: Eine Familie von anti-skeptischen Argumentationsstrategien“, Grazer Philosophische Studien, vol. 14. van Inwagen, Peter (1983): An Essay on Free Will, Oxford [et al.]: Clarendon Press.

218

Grazer Philosophische Studien 76 (2008), 219–227.

ÜBER MORALISCHE VERANTWORTUNG UND ALTERNATIVE MÖGLICHKEITEN Anselm SPINDLER Universität Frankfurt Zweiter Preis des Essay-Wettbewerbs für Studierende der Philosophie 2007, ausgeschrieben von der Gesellschaft für Analytische Philosophie und den Grazer Philosophischen Studien

Können wir verantwortlich handeln, wenn jede unserer Handlungen determiniert ist? Ein wichtiger Angelpunkt der klassischen Diskussion um diese Frage ist das „Prinzip alternativer Möglichkeiten“ (PAM), demzufolge ein Akteur nur dann für seine Handlung verantwortlich gemacht werden kann, wenn er auch anders hätte handeln können. Inkompatibilisten argumentieren dabei, dass Determinismus und Verantwortlichkeit unvereinbar sind, weil der Determinismus keine alternativen Handlungsmöglichkeiten zulässt. Und wer zeigen will, dass Personen auch unter Bedingungen des Determinismus für ihr Tun verantwortlich sein können, scheint den Nachweis erbringen zu müssen, dass der Determinismus entgegen eines ersten Anscheins mit der Offenheit relevanter Handlungsalternativen durchaus vereinbar ist. Harry Frankfurt hat diese Auseinandersetzung indes grundlegend hinterfragt, indem er die Aufmerksamkeit auf ihren common ground gerichtet hat. Dabei versucht er zu zeigen, dass das PAM zu Unrecht als Formulierung einer notwendigen Freiheitsbedingung für Verantwortlichkeit interpretiert wird. Und wenn sein Argument überzeugt, ist es nicht mehr evident, woher die eingangs gestellte Frage ihren philosophischen Witz bezieht. In diesem Essay soll untersucht werden, wie plausibel Frankfurts These ist. Dabei wird zunächst seine Argumentationsstrategie skizziert (1), um anschließend zwei Einwände aufzunehmen, die das PAM mit dem kantischen Prinzip „sollen impliziert können“ assoziieren (2). Frankfurts Erwiderung auf diese Kritik hat ihre Wurzeln in handlungstheoretischen Grundüberzeugungen und kann die Einwände unter Zuhilfenahme einer zusätzlichen verantwortungstheoretischen Prämisse zurückweisen (3). Dies hat wichtige Konsequenzen für die klassische Fragestellung zu Determinismus und Verantwortung (4).

219

1. Aristoteles skizziert in seiner Nikomachischen Ethik den Fall eines Seefahrers, der in einen Sturm gerät und die geladenen Waren über Bord wirft, weil er nur so Aussichten hat, Schiff und Besatzung erfolgreich aus der Notlage herauszumanövrieren. Doch trotz der widrigen Umstände, die seine Handlung erzwingen, hält es Aristoteles für unplausibel, ihn schlichtweg für nicht verantwortlich zu erklären. Aus der Perspektive des PAM lässt sich diese Intuition mit dem Hinweis begründen, dass dem Schiffskapitän alternative Handlungsmöglichkeiten offen stehen. Sicherlich gibt es gute Gründe dafür, die Ladung zu opfern, denn jede alternative Handlung ist mit Konsequenzen verbunden, die ein vernünftiger Akteur zu Recht für unerträglich hält. Aber der Zwang, mit dem sich der Kapitän konfrontiert sieht, ist der Zwang des vernünftigen Grundes. Dieser ändert nichts daran, dass de facto mehrere Möglichkeiten zur Wahl stehen. Daher lässt sich sagen, dass er für seine Rettungstat verantwortlich ist, denn er hätte schließlich auch anders handeln können. Harry Frankfurt schlägt (mit Aristoteles) eine andere Strategie ein, um die Frage der Verantwortlichkeit des Kapitäns zu klären (Frankfurt 1988: 1–10). Er verortet den Kern der Zuschreibung von Verantwortung in der Bestimmung des Handlungsprinzips, und versucht dabei in einem ersten negativen Schritt zu zeigen, dass diese Zuschreibung nicht auf den Schultern alternativer Möglichkeiten ruht. Zur Plausibilisierung dieser These bedient sich Frankfurt eines Gedankenexperiments, in dem ein Akteur aufgrund seiner eigenen praktischen Überlegungen eine Handlung vollzieht, aber dies unter Umständen tut, die jede alternative Handlungsweise verschließen. Ein solcher Fall ist z.B. gegeben, wenn Jones sich aus eigenen Gründen dafür entscheidet, ein bestimmtes Stück Marmorkuchen zu verzehren. Er weiß jedoch nicht, dass Black zuvor einen Überwachungs- und Manipulationsmechanismus in sein Gehirn implantiert hat, der es ihm erlaubt, Jones’ Handlungsprozesse genau zu beobachten. Nun will auch Black, dass Jones das betreffende Stück Marmorkuchen verzehrt, und er steht bereit, Jones vermittels der implantierten Maschinerie zu dieser Handlung zu bewegen, sollte dieser Anstalten machen, irgend eine andere Handlung zu vollziehen oder gar untätig zu bleiben. Heute aber hat Black Glück: aus eigenem Antrieb und ohne Wissen um die merkwürdigen Umstände greift Jones nach dem Marmorkuchen und isst ihn auf. Die Pointe von Fallbeispielen dieser Art ist für Frankfurt, dass sie aufzeigen, dass diejenigen Faktoren, die eine Handlung unvermeidlich machen, nicht immer auch deren Ursache sind. Jones kann nichts anderes tun, als den Marmorkuchen zu verzehren – denn in jedem alternativen Szenario stünde Black bereit, genau diese Handlung von Außen zu induzieren. Für den tatsächlichen Verlauf der Dinge spielt Black jedoch keine Rolle. Statt-

220

dessen trifft Jones eine eigenständige Entscheidung, die für die Hervorbringung seiner Handlung hinreichend ist. Folglich markiert die Präsenz von Black keinen Unterschied zu Handlungen, die Jones unter alltäglicheren Bedingungen vollzieht und für die er zweifelsohne verantwortlich gemacht werden kann. Damit verliert die Offenheit alternativer Möglichkeiten aber ihre Plausibilität als notwendiges Kriterium für Verantwortlichkeit. Denn es sind Fälle der Überdetermination von Handlungen denkbar, in denen alternative Möglichkeiten effektiv verschlossen sind, ohne dass dadurch aber der „normale“ Verlauf der Dinge beeinträchtigt wird. Stattdessen vollzieht der Akteur eine eigenständige Handlung, die er auch angesichts offener Alternativen gewählt hätte. Im Gedankenexperiment zeigt sich also, dass das PAM mit einem Kriterium für die Zuschreibung von Verantwortung arbeitet, das unter den Bedingungen von Frankfurts Fallbeispielen nicht erfüllt ist, ohne dass davon jedoch die Verantwortlichkeit des autonomen Akteurs berührt wird. Frankfurt schlägt deshalb vor, die Wurzel der Zuschreibung von Verantwortung stattdessen in der Unterscheidung zwischen dem autonomem Handlungsmechanismus eines Akteurs und einem von außen aufgezwungenen oder manipulativ induzierten Mechanismus zu sehen. Wer eine Handlung nur deshalb vollzieht, weil er nicht anders kann, mag von Verantwortlichkeiten freigesprochen werden. Wer jedoch nur aufgrund seiner eigenen Motive eine selbständige Handlung vollzieht, ist dafür verantwortlich, selbst wenn im Hintergrund Faktoren lauern, die die Handlung unausweichlich machen. Für den Seefahrer des Aristoteles bedeutet dies, dass auch seine Verantwortung nur scheinbar auf der Offenheit von Handlungsalternativen ruht. Denn diese Alternativen lassen sich im Gedankenexperiment subtrahieren, ohne dass dadurch der selbstbestimmte Handlungsprozess des Akteurs in einem verantwortungsrelevanten Sinne beeinträchtigt wird.

2. An Frankfurts Zurückweisung des PAM schließt sich einerseits eine Debatte darüber an, ob „Frankfurt-style cases“ tatsächlich „Frankfurt-style cases“ sind. In diesem Sinne versuchen verschiedene Autoren etwa zu zeigen, dass man keine Szenarien skizzieren kann, in denen dem Akteur wirklich alle alternativen Möglichkeiten genommen sind. Diese Diskussion geht jedoch an der Pointe der Fallbeispiele insofern vorbei, als zu einer Erwiderung auf Frankfurt ein zusätzlicher Gedankenschritt erforderlich ist. Denn wer Frankfurt begegnen will, muss nicht nur zeigen, dass es in den Fallbeispielen alternative Möglichkeiten gibt, sondern auch, dass diese alternativen Möglichkeiten für die Zuschreibung von

221

Verantwortung entscheidend sind. Das Spielfeld dieser Diskussion ist daher weniger die Metaphysik intentionalen Handelns als unsere moralischen Intuitionen im Hinblick auf die Attribution von Verantwortlichkeit. Ihre zentrale Frage lautet, ob es entgegen Frankfurts Vermutung nicht doch in bestimmten Fällen plausibel ist, die Verantwortlichkeit eines Akteurs an die Offenheit alternativer Möglichkeiten zu binden. In genau diese Richtung argumentiert z.B. Peter van Inwagen. Zwar gesteht er zu, dass dem PAM durch „Frankfurt-style cases“ der Boden entzogen wird. Allerdings bezieht sich das PAM nur auf die tatsächlich vollzogenen Handlungen eines Akteurs. Für Handlungsunterlassungen dagegen lässt sich ein ähnlicher Grundsatz entwickeln, an dem Frankfurts Kritik spurlos vorbeigeht (van Inwagen 1978). Dieses „Prinzip möglicher Handlungen“ (PMH) besagt, dass ein Akteur nur dann für eine Unterlassung zur Rechenschaft gezogen werden kann, wenn er die ausgebliebene Handlung auch tatsächlich hätte vollziehen können. Van Inwagen konstruiert dazu einen entsprechendes Beispiel, in dem ein Akteur als Zeuge eines Verbrechens von einem Anruf bei der Polizei absieht, weil er es vorzieht, mit der Angelegenheit nichts zu tun zu haben. Er weiß jedoch nicht, dass zu relevanter Zeit das Telefonnetz zusammengebrochen ist. In dieser Situation ist es van Inwagen zufolge unplausibel, den Akteur für den ausgebliebenen Anruf bei der Polizei verantwortlich zu machen; denn wie hätte er diese Handlung vollziehen können? Der Akteur mag dafür verantwortlich sein, dass er nicht versucht hat, die Polizei anzurufen – that much he could have done; aber er muss nicht gerade stehen für die Unterlassung einer Handlung, die er nicht hätte vollziehen können. Folglich scheint es zumindest im Falle von Handlungsunterlassungen sinnvoll zu sein, Verantwortung mit der Offenheit alternativer Möglichkeiten in Verbindung zu bringen, weil man sonst Akteure dafür verantwortlich machen würde, das Unmögliche nicht getan zu haben. Einen zweiten Einwand, der in eine ähnliche Richtung geht, hat David Widerker formuliert. Auch er ist der Ansicht, dass Frankfurts Argument das PAM unterminiert, jedenfalls in Fällen, in denen die Zuschreibung von Verantwortung mit einer positiven oder neutralen moralischen Bewertung der Handlung verknüpft ist. Er will jedoch zeigen, dass Frankfurts Argument nicht greift, wenn mit der Zuschreibung von Verantwortung Tadel verbunden ist (Widerker 2003). Denn nimmt man etwa an, dass Jones’ Verzehr des Marmorkuchens eine tadelnswerte Handlung ist, dann impliziert dieses Urteil, dass Jones den Marmorkuchen nicht hätte anrühren sollen. Diese Option ist jedoch unter Frankfurt’schen Bedingungen per definitionem verschlossen, sodass die in der Zuschreibung von Schuld implizierte Forderung der Unterlassung leer ist, denn sie verlangt von Jones, das Unmögliche zu tun. Daher argumentiert Widerker, dass die Offenheit alternativer Möglichkeiten zumindest im Spezialfall

222

von tadelnswerten Handlungen erforderlich ist, um sinnvoll von moralischer Verantwortung sprechen zu können. Die beiden skizzierten Einwände gegen Frankfurt gleichen sich in doppelter Hinsicht. Zum einen machen sie gegen Frankfurt eine Asymmetriethese im Hinblick auf die Zuschreibung von Verantwortung geltend: In van Inwagens Fall ist dies eine Asymmetrie zwischen Handlungen und Unterlassungen, und in Widerkers Fall eine Asymmetrie zwischen lobenswerten und tadelnswerten Handlungen. Zum anderen beziehen beide Autoren die Plausibilität ihrer Argumente daraus, dass sie das PAM mit dem kantischen Prinzip (K) in Verbindung bringen, wonach „sollen“ stets „können“ voraussetzt. Letzteres fängt die Intuition ein, dass Pflichten als handlungsanleitende moralische Gebote nur dann sinnvoll konzipiert sind, wenn sie sich im Einzelfall nur auf solche Handlungen beziehen, die ein Akteur ausführen kann. Und wenn es evident ist, dass ein Akteur zur Ausführung einer Handlung nicht in der Lage ist, dann kann es für diesen Akteur keine entsprechende Handlungsverpflichtung geben. Van Inwagen und Widerker argumentieren nun, dass das gleiche für die Zuschreibung von Verantwortung gilt. Auch hier ist es unplausibel, einen Akteur für eine Handlung verantwortlich zu machen, die er nicht ausführen kann. Es ist jedoch unklar, in welcher Relation diese beiden Prinzipien überhaupt stehen, und Frankfurt hält sie für nicht besonders eng (Frankfurt 1988: 95f ). Denn während das PAM eine notwendige Bedingung für die Verantwortlichkeit eines Akteurs in zusätzlichen, alternativen Handlungsmöglichkeiten sieht, verknüpft K die Zuschreibung einer Pflicht mit der Voraussetzung, dass der betreffende Akteur die geforderte Handlung auch tatsächlich ausführen kann. Das postulierte Implikationsverhältnis zwischen Fähigkeit und Verpflichtung kann also auch dann greifen, wenn dem Handlenden nur eine einzige Möglichkeit offen steht, während das PAM die Zuschreibung von Verantwortung in einem solchen Fall für ungerechtfertigt hält. Aber ist diese Diagnose überzeugend? Die Pointe der Kritik von van Inwagen und Widerker liegt darin, dass beide Handlungssituationen ins Feld führen, für deren Bewertung K und das PAM scheinbar ineinander greifen. Van Inwagens Argument ist dabei, dass die Zuschreibung von Verantwortlichkeit zumindest im Spezialfall von Unterlassungen nicht nur von der Relation zwischen dem Akteur und seiner Unterlassung abhängt, sondern auch in Erwägung gezogen werden muss, ob es dem Akteur überhaupt möglich war, die unterlassene Handlung auszuführen. Im Fall von Unterlassungen besteht die Evaluationsbasis für die Zuschreibung von Verantwortung folglich nicht nur aus der tatsächlich „ausgeführten“ Handlung (nicht die Polizei rufen), sondern auch aus der Relation des Akteurs zur derjenigen Handlung, die er absichtlich unterlässt (die Polizei rufen). Gemäß K gilt nun, dass ein Akteur nur dann für die Unterlassung einer Handlung verantwortlich gemacht werden kann, wenn

223

er dieser Handlung auch hätte ausführen können. Dies bedeutet jedoch nichts anderes, als dass dem Akteur eine bestimmte alternative Handlungsoption offen stehen muss, wenn die Zuschreibung von Verantwortung gerechtfertigt sein soll. Auch Widerker macht sich diese Strategie zu Nutze, um zu zeigen, dass K und das PAM in Fällen von tadelnswerten Handlungen eine engere Beziehung eingehen, als es Frankfurt lieb sein kann. Er nimmt dabei an, dass der Tadel bereits das Urteil enthält, dass der Akteur die Handlung, für die er getadelt wird, nicht hätte ausführen sollen. Dieses Urteil ist aber gemäß K nur dann gerechtfertigt, wenn der Akteur seine tadelnswerte Handlung auch tatsächlich hätte unterlassen können. Das Argument ist also auch hier, dass es für die Zuschreibung von Verantwortlichkeit nicht (immer) nur auf den Verlauf der tatsächlich ausgeführten Handlung ankommt. Denn im Spezialfall von tadelnswerten Handlungen muss zusätzlich die Relation des Akteurs zu einer entsprechenden Unterlassung geprüft werden. Das heißt jedoch nichts anderes, als dass der getadelte Akteur mindestens eine alternative Handlungsmöglichkeit gehabt haben muss, wenn die Zuschreibung von Verantwortung gerechtfertigt sein soll.

3. Frankfurts Replik auf die skizzierten Einwände hat zunächst handlungstheoretische Wurzeln. Dabei sieht er sich insbesondere Vertretern kausaler Handlungstheorien gegenüber, deren Grundintuition in Davidsons Aufsatz Actions, Reasons, and Causes formuliert ist (Davidson 2001: 3–19). Sie besagt, dass die spezifische Differenz von Handlungsereignissen gegenüber anderen Körperbewegungen in ihren besonderen kausalen Voraussetzungen zu finden ist, und dies sind die Wünsche und Überzeugungen des Akteurs. Vertreter kausaler Handlungstheorien sehen sich allerdings mit zwei notorischen Problemen konfrontiert, dem Problem chaotischer Kausalketten einerseits, und dem Problem der Beschreibung von Handlungsunterlassungen andererseits. Zudem schaffen sie es nicht, Eigenschaften der Handlungsereignisse selbst als deren Spezifikum zu charakterisieren. Um diese Schwierigkeiten zu umgehen, weist Frankfurt den Ansatz zurück, Handlungen über ihre kausale Geschichte zu verstehen. Stattdessen versucht er, agency als eine synchrone Relation zwischen einem Akteur und seinen Körperbewegungen zu explizieren, und fasst sie entsprechend als kausale Kontrolle (guidance) auf, vermittels derer der Akteur während der Handlung in kausalem Kontakt zu seinen Körperbewegungen steht (Frankfurt 1988: 69–79). Zugleich teilt er jedoch mit Davidson die Auffassung, dass diese als Handlungen identifizierten, kontrollierten Körperbewegungen das einzige sind, was wir tun; über diesen Basishandlungen entstehen keine zusätzlichen komplexen Handlungen

224

im Danto’schen Sinne. Vermeintlich zusätzliche Handlungsereignisse sind vielmehr unter die Handlungsfolgen zu rechnen, für die der Akteur bestenfalls in einem derivativen Sinne verantwortlich ist (Frankfurt 1988: 99f ). Auf dieser Grundlage kann Frankfurt nun zunächst van Inwagens Differenzierung zwischen „failing to call the police“ und „failing to try to call the police“ begegnen. Diese Unterscheidung ist für Frankfurt lediglich eine Unterscheidung auf der Beschreibungsebene, die auf der Ebene der Ereignisse nicht reflektiert wird. Denn in beiden Fällen sind die mentalen Zustände und die Körperbewegungen des Akteurs schlicht die selben. Die Frage nach dem Zustand des Telefonsystems mag für die Wahl einer treffenden Beschreibung relevant sein und betrifft dann die möglichen Folgen einer entsprechenden positiven Handlungen. Letztere sind für Frankfurt aber (mit Davidson gesprochen) up to nature, sodass es keinen plausiblen handlungsanalytischen Unterschied zwischen der Unterlassung einer Handlung und der Unterlassung eines Versuchs der Ausführung dieser Handlung gibt, der eine Differenzierung im Hinblick auf die Verantwortung des Akteurs rechtfertigen könnte. Und dasselbe gilt für die von van Inwagen ins Feld geführte Asymmetriethese im Bezug auf Handlungen und Unterlassungen. Frankfurts gegen kausale Handlungstheorien entwickeltes Konzept der „guidance“ ist mit der Unterlassung von Körperbewegungen konsistent, solange sich diese Unterlassung in der Reichweite der kausalen Kontrolle des Akteurs befindet. Folglich gibt es auch hier keinen wesentlichen handlungsanalytischen Unterschied, sodass jeder Versuch, Handlungen und Unterlassungen im Hinblick auf die Bedingungen von Verantwortlichkeit zu differenzieren, erst einmal kontraintuitiv ist (Frankfurt 1994: 620–623). Neben diesen handlungstheoretischen Vorgaben hat Frankfurts Replik jedoch noch eine zusätzliche verantwortungstheoretische Prämisse, die besagt, dass zur Klärung von Verantwortungsfragen nur solche Faktoren herangezogen werden können, die auch für eine Erklärung der betreffenden Handlung relevant sind (Frankfurt 1988: 8). Solche Handlungserklärungen verweisen für Frankfurt auf die Motive und Überzeugungen des Akteurs, die verständlich machen, warum er auf eine bestimmte Art und Weise gehandelt hat.1 Diese zusätzliche Prämisse erlaubt es Frankfurt, eine wichtige Unterscheidung zu treffen, die Van Inwagen verschlossen bleibt. Denn dessen PMH lässt nicht zu, zwischen einem Akteur zu unterscheiden, der auf den Anruf bei der Polizei nur wegen des defekten Telefons verzichtet, und einem Akteur, der (wie in van Inwagens Beispiel) den Anruf bei der Polizei nur deshalb unterlässt, weil er in die Angelegenheit nicht hineingezogen werden möchte, ohne dabei jedoch zu wissen, dass er nicht telefonieren kann. Während in beiden Fällen die selben äußeren Handlungsbe1. Sie werden dadurch jedoch nicht zu Kausalerklärungen (Frankfurt 1988: 75).

225

schränkungen gegeben sind, unterscheiden sich die beiden Akteure in sofern, als der erste in Gegenwart eines funktionierenden Telefons die Polizei gerufen hätte, wohingegen der zweite auch dann von einem Telefonat abgesehen hätte; sein Verhalten ist gegenüber dem Zustand des Telefonsystems indifferent. Folglich leistet der Verweis auf das zusammengebrochene Telefonsystem nichts für die Erklärung seiner Handlung, während sie im ersten Fall der wichtigste Schlüssel zu einem besseren Verständnis ist. Und da die Differenzierung dieser Fälle für die Zuschreibung von Verantwortlichkeit von größter Bedeutung ist, kann die bloße Offenheit von Handlungsalternativen nicht als notwendige Bedingung für Verantwortlichkeit angesehen werden. Stattdessen ist entscheidend, welche Rolle die Verschlossenheit von Alternativen spielt. Ist sie für die Sequenz der Ereignisse unerheblich, weil der Akteur eine selbständige Handlung aus eigener Entscheidung vollzieht, kann sie nicht sinnvoll als Grund für dessen Dispens angeführt werden. Dies gilt auch für Widerkers Verteidigung des Marmorkuchensünders Jones. Es ist unplausibel, die Verantwortlichkeit einer tadelnswerten Person an Faktoren zu binden, die für die entsprechende Handlung möglicherweise irrelevant sind (Frankfurt 2003: 343ff). Auch das hieße nämlich, den Unterschied zwischen Fällen zu übersehen, in denen ein Akteur sich nur deshalb am Marmorkuchen vergreift, weil er nicht anders kann, und Fällen, in denen ein Akteur auf der Grundlage seiner eigenen Entscheidung zum Frevel den Kuchen verspeist, nicht ahnend, dass ein verborgener Manipulator jede andere Handlungsweise unterbinden würde. Die bloße Anwesenheit von Black kann nicht implizieren, dass Jones von seinen moralischen Verpflichtungen befreit ist, auch wenn die bloße Anwesenheit von Black zuverlässig dafür sorgt, dass Jones keine alternativen Handlungsmöglichkeiten hat. Damit kann das entscheidende Kriterium für die Verantwortlichkeit eines Akteurs aber auch hier nicht in der schlichten Offenheit von Handlungsalternativen liegen, sondern es hängt alles davon ab, ob der Akteur der autonome Urheber der Handlung ist, die er vollzieht. Folglich kann es keine verantwortungstheoretische Asymmetrie zwischen lobeswerten und tadelnswerten Handlungen geben, und das PAM erweist sich erneut als von K unabhängiges Prinzip.

4. Weder van Inwagen noch Widerker können also gegen Frankfurt die Relevanz alternativer Möglichkeiten plausibel machen – und damit auch nicht die Bedeutung der klassischen Diskussion zwischen Kompatibilisten und Inkompatibilisten. Wer folglich die Frage nach der Vereinbarkeit von moralischer Verantwortung und Determinismus stellt, muss entweder zeigen, dass der Determinismus auch jenseits

226

der Frage nach alternativen Möglichkeiten ein Problem für moralische Verantwortung bereithält, oder eingestehen, dass die Kompatibilitätsfrage für moralische Verantwortlichkeit of no particular significance ist (Frankfurt 1988: 95).

LITERATUR Davidson, Donald (2001): Essays on Actions and Events. Frankfurt, Harry (1988): The Importance of What We Care About. — (1994): „An alleged asymmetry between actions and omissions“, in: Ethics, Vol. 104, No. 3. — (2003): „Some thoughts concerning PAP“, in: Widerker, David/McKenna, Michael (Hg.): Moral Responsibility and Alternative Possibilities. Van Inwagen, Peter (1978): „Ability and responsibility“, in: The Philosophical Review, Nol. 87, No. 2. Werderker, David (2003): „Blameworthiness and Frankfurt’s argument against the principle of alternative possibilities“, in: Widerker, David/McKenna, Michael (Hg.): Moral Responsibility and Alternative Possibilities.

227

Grazer Philosophische Studien 76 (2008), 228–235.

WEEDING IN THE GARDEN OF FORKING PATHS – YET ANOTHER LOOK AT ALTERNATE POSSIBILITIES Andreas MAIER Universität Zürich First prize of the essay competition for students of philosophy 2007, offered by the Gesellschaft für Analytische Philosophie and the Grazer Philosophische Studien

1. Introduction It is a common assumption that in order to be morally responsible for her actions a person must be free to do otherwise. This idea can be captured in the so-called principle of alternate possibilities (PAP) as it was stated by Harry Frankfurt (cf. Frankfurt 1969: 167): PAP A person can be morally responsible for what she has done only if she could have done otherwise. Since determinism is the thesis that there is for any instant only one physically possible universe, and thus precludes any alternative courses of action, it seems quite natural to think that if determinism is true no one would be morally responsible for what they did. Call this the incompatibilist thesis. Proponents of compatibilism who in contrast hold that moral responsibility is possible even if determinism obtains have to show that having alternative possibilities is not a necessary condition for moral responsibility, and hence that PAP is false.1 One way of doing this was proposed by Harry Frankfurt who sought to refute PAP by developing a scenario (’Frankfurt-style example’, FSE) in which an agent has no alternative possibilities but is responsible anyway. In this essay I want to show that although Frankfurt’s argument has a strong intuitive appeal, there are good reasons to believe that his scenario alone does not suffice to rebut PAP. To do this I will give a short account of Frankfurt’s original example (2), and then discuss some important incompatibilist objections 1. Another compatibilist option, namely to give a conditional analysis of ‘could’ in such a way that PAP can be satisfied even under conditions of determinism will not be considered in what follows. For a thorough discussion see Kane 1996: 52–58.

228

against it (3). This discussion will show that proponents of FSE have to make substantial assumptions about the nature of excuses which incompatibilists do not share. If this is correct the quarrel between the two parties can be understood as a disagreement about the very conditions under which an agent can be excused for her morally wrong action. Although I think that there are good reasons to believe that the compatibilist’s account of excuses is more plausible than its incompatibilist counterpart I will not take sides for either here. The more limited aim of this paper is to show which accounts of excuses compatibilists and incompatibilists implicitly presuppose in their assessments of Frankfurt’s example and thus to make clear where their respective positions are in need of additional argumentative support.

2. Frankfurt-style examples In his article ‘Alternate Possibilities and Moral Responsibility’, Frankfurt attacks PAP by assuming that [a] person may do something in circumstances that leave him no alternative to doing it, without these circumstances actually moving him or leading him to do it - without them playing any role, indeed, in bringing it about that he does what he does[.] (Frankfurt 1969: 168) Frankfurt holds that in such circumstances the agent is morally responsible for her action although she could not have done otherwise as she actually did; if this is true, then PAP is falsified. In order to support these assumptions, Frankfurt gives the following example: FSE “Suppose someone, Black, let us say—wants Jones to perform a certain action. Black is prepared to go to considerable lengths to get his way, but he prefers to avoid showing his hand unnecessarily. So he waits until Jones is about to make up his mind what to do, and he does nothing unless it is clear to him […] that Jones is going to decide to do something other than what he wants him to do. If it becomes clear that Jones is going to decide something else, Black takes effective steps to ensure that Jones decides to do, and that he does do, what he wants him to do. […] Now suppose that Black never has to show his hand because Jones, for reasons of his own, decides to perform and does perform the very action Black wants him to perform. In that case, it seems clear, Jones will bear precisely the same moral responsibility for what he does as he would have borne if

229

Black had not been ready to take steps to ensure that he do it.” (Frankfurt 1969: 172f.) The following features of this scenario are important for Frankfurt’s argument: FSE1 At time t0 there is a sign prior to Jones’s decision that ‘makes it clear’ what decision he will take. At t1 Jones decides what to do, and at t2 acts according to his decision. FSE2 If the prior sign at t0 shows that Jones will at t1 make a decision D1 which is not in accordance with Black’s wishes, Black will force him to make a decision D2 which is in accordance with Black’s wishes. So, at t1 Jones can only make decision D2, and thus lacks alternate possibilities. FSE3 In the actual sequence of events the prior sign at t0 shows that Jones will make decision D2, so Black does not interfere. At t1 Jones makes decision D2, acts in accordance with it and so does as Black wanted him to do. FSE4 Since Black did not interfere in the actual sequence of events, Jones bears the same responsibility for his action as he would have if Black were not present. Since Black did not intervene in the actual course of events (FSE3), but Jones could not decide otherwise than he did (FSE2), Jones is responsible for his action (FSE4) in spite of having no alternate possibilities; hence, PAP is false.

3. Incompatibilist worries and compatibilist responses Frankfurt’s assumption that FSE is sufficient to refute PAP has been challenged by most incompatibilists.2 Their objection usually exploits the fact that the relation between the prior sign and Jones’s decision in FSE2 must be either causally deterministic or causally indeterministic, and contend that in either case the compatibilist is unable to establish the truth of FSE4 without begging the ques2. A notable exception are so-called ‘source-incompatibilists’ who hold that being the ultimate source of one’s action is more important than having alternate possibilities; cf. McKenna 2000.

230

tion against the incompatibilist (cf. Widerker 1995: 250–2).3 In the following sections both horns of the dilemma will be closely examined. The assumption of determinism If the relation between the occurrence of the prior sign and Jones’s decision is deterministic, then the incompatibilist does not have to accept that Jones is morally responsible for his action and hence can deny FSE4: since incompatibilists hold that under conditions of determinism moral responsibility is impossible, stating that Jones is responsible in Frankfurt’s scenario despite the assumption of causal determinism would beg the question against the incompatibilist.4 Compatibilists like Fischer have pointed out that this is a misrepresentation of their position: to show that the ability to do otherwise is not a necessary condition for moral responsibility proponents of FSE do not have to claim that Jones is morally responsible in this scenario; they only maintain that if Jones is not responsible for what he did, it is not simply because he lacked alternate possibilities (cf. Fischer 2002b: 297). While Fischer holds that this can be seen ”simply by reflecting on the examples” (cf. Fischer 2002b: 292) incompatibilists allude to the fact that even such a hypothetical claim does not follow from FSE1-FSE4 alone (cf. Ekstrom 2002: 311). The problem seems to lie with FSE4 which states that Black’s inactivity in the actual sequence of events gives us good reason to think that Jones is as responsible as if Black had not been there at all. The compatibilist here simply takes for granted that we are justified in passing from Black’s non-intervening in the actual sequence to Jones’s bearing as much responsibility as if Black had not been present at the scene. But since this inference is not as self-evident as compatibilists might think it is this move is in need of further justification. To see what exactly is missing in the compatibilist’s story it is important to note that there is no argument given for the contention that the possibility of Black’s intervention is no reason to withhold a judgement of Jones’s responsibility. And this would only be true if any conditions which could give us reason to excuse Jones for his action had to be looked for in the actual sequence of events only. So, in order to infer that Jones bears the same responsibility as if Black had not been present from Black’s non-intervening in the actual sequence, the 3. Some compatibilists have tried to evade this dilemma by proposing revised versions of FSE which do not rely on prior signs since alternate possibilities are blocked at any rate (cf. Mele/Robb 1998 and Hunt 2000). The grounds on which the incompatibilist can reject these examples will be examined in the next paragraphs. 4. Cf. amongst others Ekstrom 2002, Goetz 2005.

231

compatibilist has to make substantial assumptions about the nature of excuses: only if it is true that the only place to look for excusing conditions is the actual sequence of events, it follows that the fact that Jones is lacking alternate possibilities is no reason to think that he is not morally responsible for what he did. So, the compatibilist seems to be committed to an excusing principle like the following: ExcC An agent A is excused for performing a morally wrong action ϕ iff any excusing conditions obtained in the actual sequence of events. If this is the additional assumption the compatibilist needs to make the Frankfurt scenario work, then what are the reasons for the incompatibilist to reject it? A preliminary answer would be that for the incompatibilist the absence of alternate possibilities is in itself an excusing condition; so, if Black’s presence in the Frankfurt-scenario precludes any possibility of doing otherwise (see FSE2), the incompatibilist can always deny Jones’s responsibility on these grounds. And this is why the compatibilist just cannot convince his opponent that FSE can establish that PAP is false. In the following section this preliminary answer will be made more precise by looking at the second horn of the dilemma. The assumption of indeterminism If the relation between the occurrence of the prior sign and Jones’s decision is indeterministic, it is not true that Jones did not have alternate possibilities, since then the occurrence of the prior sign can only be a (more or less) reliable indicator for Jones’s decision and it is plainly false that his decision was unavoidable (cf. Widerker 1995: 179f.). The following example should illustrate this contention: suppose that at t0 a prior sign occurs, showing that Jones will decide at t1 in a way D2, which is in accordance with Black’s wishes. Since there is no deterministic connection between the prior sign and Jones’s decision, it might be that Jones decides at t1 in a way D1. If Black wants to intervene, he has to interfere in Jones’s already started decision-making process (cf. Fischer 2002a: 194). So, since it is open at t0 how Jones will (begin to) decide at t1, it seems that Jones has alternate possibilities and this fact explains why he is morally responsible. In reply to this challenge compatibilists have pointed out that the alternate possibilities in such a scenario are not sufficiently robust to hold Jones responsible for his action, or as Fischer puts it: they seem to be mere ‘flickers of freedom’ and the possibility of exhibiting a prior sign which is not in accordance with

232

Jones’s decision seems to be too flimsy to make a morally relevant difference (cf. Fischer 1994: 134–5). To understand Fischer’s point it is important to note that these flickers would occur in the alternative sequence only in those cases where Jones does not decide in accordance with Black’s wishes. So, the only difference between the scenario which presupposes determinism and the scenario which presupposes indeterminism is that in the latter a flicker of freedom could have occurred (but in fact has not). And it seems implausible to think that Jones cannot be held accountable for his action in the first case but in the second, since there does not seem to be a morally relevant difference between these cases. This argument has been widely accepted and it seems very plausible to think that flickers of freedom cannot be the decisive element when we assess Jones’s responsibility.5 But should the incompatibilist really worry about this argument? I do not think so because with such a line of reasoning the compatibilist implicitly acknowledges that Frankfurt’s original formulation of PAP failed to capture the incompatibilist’s position: it would seem strange for the compatibilist to claim on the one hand that mere flickers of freedom are not robust enough to ground moral responsibility, but to depict the incompatibilist on the other hand as endorsing a principle, PAP, that allows of exactly that. A reasonable objection here might be that it was the incompatibilist herself who held that if Jones is responsible under conditions of indeterminism it is because he had alternate possibilities. But it is important to note here that PAP is a principle the compatibilist ascribes to the incompatibilist; and the fact that Frankfurt’s formulation of PAP leaves room for a sort of alternate possibilities which is even a prima facie unplausible candidate as a condition for responsibility constitutes a good reason for the incompatibilist to reject commitment to PAP in the first place. So, if the incompatibilist is not committed to PAP, what would make an alternate possibility robust enough to count as morally significant? Widerker states that for the incompatibilist holding an agent responsible for a morally wrong action requires that we can tell her what she should have done instead (cf. Widerker 2000: 191). And the answer to this question is clear: she should have omitted the morally wrong action.6 With this in mind, we can make the preliminary account of the incompatibilist conception of excuses more precise: 5. For a criticism of this view cf. Della Rocca 1998. 6. It is worth noting here that since this is the answer to the question what the agent should not have done, ‘omitting’ here means ‘intentionally omitting.’ This rules out so-called ‘limited blockage counter examples’ where the agent can avoid a wrong action at t1 by doing something morally insignificant at t0, e.g. by drinking a sleeping potion without knowing about its effects (cf. McKenna 2003/Pereboom 2003). The incompatibilist who does not subscribe to Frankfurt’s formulation of PAP and only endorses morally significant alternatives to be necessary for responsibility does not have to worry about these examples.

233

ExcI A is excused for performing a morally wrong action ϕ iff A could not have omitted the performance of ϕ. With this account of excuses it becomes clear on what grounds the incompatibilist can reject FSE: since FSE2 states that Jones can by no means avoid acting as he does and according to ExcI an actor who cannot avoid acting as he does cannot legitimately be held responsible for his action it follows that Jones cannot bear any responsibility for what he did. So, under the assumption that ExcI is correct FSE4 would be false and Frankfurt’s argument as a whole would be unable to establish the falsity of PAP. As in the case of her opponent the incompatibilist is in need of further arguments to back up her excusing principle in order to support her position.

4. Conclusion If the account of the debate given in the last sections is correct one important issue underlying the debate between proponents of FSE and their incompatibilist opponents seems to be the question which conception of excuses is the correct one. And the reason why the dispute about PAP seems to go round in circles could be that in the highly artificial Frankfurt scenario there is no hint which account of excuses is more plausible—and so everyone is left with their (compatibilist or incompatibilist) intuitions: if one puts in the intuition that the place to look for excusing conditions is the actual sequence only (ExcC) then one is indeed justified in believing that Jones can be held accountable for his action in the Frankfurt-scenario. If, on the other hand, one finds more plausible the idea that the impossibility to omit a wrong action is an excusing condition per se (ExcI) then there seems to be no way of establishing Jones’s responsibility in Frankfurt’s example. In conclusion it seems to be a promising strategy for both compatibilists and incompatibilists to back up their arguments with a substantial account of excuses and so give good reasons for the correctness of their positions instead of adhering to mere intuitions their opponents just do not and cannot share.7

7. I would like to thank Holger Baumann, Marcus Willaschek, and the participants of Peter Schaber’s colloquium at the University of Zürich for valuable comments.

234

REFERENCES Della Rocca, Michael (1998): “Frankfurt, Fischer and Flickers”, in: Noûs, 32(1), 99–105. Ekstrom, Laura Wadell (2002): “Libertarianism and Frankfurt-Style Cases”, in Kane (2002), 309–321. Fischer, John Martin (1994): The Metaphysics of Free Will. An Essay on Control, Blackwell. — (2002a): “Frankfurt-Style Compatibilism”, in Watson (2003), pp. 190–211. — (2002b): “Frankfurt-Style Examples and SemiCompatibilism”, In Kane (2002), 281–308. Frankfurt, Harry G. (1969): “Alternate Possibilities and Moral Responsibility”, in Watson (2003), 167–177. Goetz, Stewart (2005): “Frankfurt-Style Counterexamples and Begging the Question”, in: Midwest Studies in Philosophy, XXIX, 83–105. Hunt, David (2000): “Moral Responsibility and Unavoidable Action”, in: Philosophical Studies, 97(2), 195–227. Kane, Robert (1996): The Significance of Free Will, Oxford University Press. Kane, Robert, ed. (2002): The Oxford Handbook of Free Will, Oxford University Press. McKenna, Michael (2000): “Source Incompatibilism, Ultimacy, and the Transfer of Non-Responsibility”, in: American Philosophical Quarterly, 38(1), 37–51. — (2003): “Robustness, Control, and the Demand for Morally Significant Alternatives: Frankfurt Examples with Oodles and Oodles of Alternatives”, in Widerker and McKenna (2003), 201–217. Mele, Alfred and Robb, David (1998): “Rescuing Frankfurt-Style Cases”, in: The Philosophical Review, 107(1), 97–112. Pereboom, Derk (2003): “Source Incompatibilism and Alternative Possibilitities”, in Widerker and McKenna (2003), 185–199. Watson, Gary, ed. (2003): Free Will. 2nd edition, Oxford University Press. Widerker, David (1995): “Libertarianism and Frankfurt’s Attack on the Principle of Alternative Possibilities”, In Watson (2003), 177–211. — (2000): “Frankfurt’s Attack on the Principle of Alternative Possibilities: A Further Look”, in: Philosophical Perspectives, 14, 181–201. Widerker, David and McKenna, Michael, eds. (2003): Moral Responsibility and Alternative Possibilities, Ashgate.

235

Grazer Philosophische Studien 76 (2008), 237–246.

SUBSTANZEN – NEUE PERSPEKTIVEN AUF EIN ALTES THEMA1 Christian KANZIAN Universität Innsbruck In ihrer Einleitung formuliert die Herausgeberin, Käthe Trettin, Idee sowie Ziel der Edition, und skizziert die Struktur des Bandes. Ziel ist es, „die Leistungsfähigkeit der Substanzkategorie aus der Perspektive einer metaphysica generalis oder allgemeinen Ontologie zu überprüfen“ (8). Was sind Substanzen? Kann ihre Annahme ontologisch verteidigt werden? Lassen sich Substanzen auf basalere Entitäten zurückführen? Das sind die Fragen, die den Leitfaden der gesammelten Beiträge ausmachen. Die Idee ist, diese Fragen vor allem systematisch anzugehen. Das schließt nicht aus, dass auch Platz ist für Beiträge zur Geschichte der Substanz-Debatte (Teil II: Aristoteles, Locke und Leibniz). Und zwar als „Pufferzone“ (9) zwischen Proponenten der Substanzen (Teil I: Kriterien des primär Seienden) und ihren Opponenten (Teil III: Revisionen). Teil I umfasst Beiträge, die versuchen, den Begriff einer Substanz zu klären sowie inhaltlich zu bestimmen, und so die Kategorie der Substanzen zu verteidigen. Einen nach Eigendefinition „aristotelischen“ Standpunkt (33) nimmt Jonathan Lowe in seinem Beitrag „Substance and Identity“ ein. Sein vorrangiges Ziel ist eine Klärung des Substanzbegriffs über seinen Zusammenhang mit „Identität“: Substanzen unterliegen bestimmten Identitätsbedingungen. Sie sind genau zu identifizieren und somit zählbar. Diese Identitätsbedingungen hängen von ihrer Art („kind“, 45) ab. Substanzen sind Träger von Eigenschaften. Substanzen haben keine zeitliche Ausdehnung, folglich auch keine zeitlichen Teile. Sie sind „endurer“, was besagt, dass sie in einem strikten Sinn durch die Zeit mit sich identisch sind. Substanzen sind grundlegende Partikularien, im Unterschied zu den Universalien; aber auch relativ zu Partikularien wie Ereignissen und Zuständen, welche ebenfalls von Substanzen abhängen. Die Verteidigung des Substanzbegriffs im ersten Teil ist nicht nur im Sinne einer traditionellen Interpretation zu verstehen, sondern durchaus auch im Sinne einer Re-Konstruktion. In diesem Sinne ist der Beitrag von Frank Hofmann zu sehen: „Substrate, Substanzen und Individualität“. Seine Lösung besteht in der ontologischen Interpretation „individueller Substanzen“ als Instanziierungen von Eigenschaften 1. Besprechungsaufsatz zu Käthe Trettin (Hg.), Substanz. Neue Überlegungen zu einer klassischen Kategorie des Seienden. Seminar Klostermann. Frankfurt am Main: Vittorio Klostermann GmbH 2005, 280 Seiten, ISBN 3-465-03441-4.

an Raumzeit-Punkten, verstanden als unteilbare Substrate. Hofmanns Ontologie umfasst neben den Substraten (Raumzeit-Punkte) und Eigenschaften, die durchaus als Universalien interpretiert werden können, auch noch Tatsachen, das „Instanziieren von Eigenschaften durch Substrate“ (96). Demgegenüber kann der Versuch Benjamin Schnieders, Substanzen als unabhängige Entitäten einzuführen „Substanz und Unabhängigkeit“ als Reformulierung klassischer Einsichten bzgl. dieser Kategorie verstanden werden. In Anlehnung an sein Buch Substanzen und (ihre) Eigenschaften diskutiert Schnieder verschiedene Ansätze, „Unabhängigkeit“ im Hinblick auf Substanzen zu definieren, um seine Idee einer „explanatorischen Unabhängigkeit“ (70ff) auszufalten. Im historischen Teil II legt Christoph Rapp in seinem Artikel „Aristoteles und aristotelische Substanzen“ Aristoteles’ Substanzbegriff dar. Nach einer Hinführung über die vor-aristotelische Begriffsgeschichte von „ousia“ unterscheidet Rapp zwischen zwei verschiedenen Forschungsprojekten bzgl. Substanz bei Aristoteles: Substanzen als die grundlegenden Entitäten; bzw. das Was dieses Grundlegende von anderen Entitäten unterscheidet. Rapp differenziert zwischen den Theorien der ousia in der Kategorienschrift und in der Metaphysik. Hier sind die konkreten Einzeldinge (Mensch, Pferd) selbst die ousia; dort ist es jene individuelle artspezifische Form, welche das Einzelding dazu macht, was es ist. Rapp legt dar, warum diese unterschiedlichen Begriffsbestimmungen von ousia nicht als Gegensatz interpretiert werden müssen und referiert über den aktuellen Stand der Forschung in der Sache. David Wiggins legt in seinem Beitrag „Substance“ ebenfalls großen Wert auf die aristotelische Entwicklung des Substanzbegriffs von Kategorienschrift zu Metaphysik. Wiggins verteidigt beide Konzeptionen. Die Substanzen der Kategorienschrift sind „continuants“, die durch sortale Determination bestimmt sind (110). Die sortale Determination gibt u.a. diachrone Kontinuitätsbedingungen an und bestimmt die Aktivitäten der Substanzen. Gegen empiristische Kritiker in der Geschichte (Locke) und in der aktuellen Debatte betont Wiggins, dass die Substanz der Kategorienschrift nicht auf die Idee von eigenschaftslosen Substrata („bare subjects“) verpflichte; stärker: „no reader of Aristotle’s Categories will ever agree to make sense of it“ (121). Die Auffassung der Metaphysik, Substanzen seien individuelle Formen, rekonstruiert Wiggins als Ergebnis der aristotelischen Versuche, weder der Materie noch dem Kompositum aus Materie und Form Substanzstatus zusprechen zu müssen (132). Die Perspektive Leibniz’ bringt Michael-Thomas Liske („Inhärenz und Tätigkeitsprinzip: Leibniz’ Substanz als rationale Konstruktion und mentales Phänomen“) in den historisch orientierten Teil ein. Leibniz versucht, den „rationalen“ Zugang Spinozas, samt dem bekannt starken Unabhängigkeitskriterium für Substanzen, mit der „empirisch-phänomenologischen“ Substanz-Theorie der aristotelischen Tradition zu vereinen. Für erstere gilt nur das Gesamtsystem

238

der Wirklichkeit als Substanz, letztere kann, v.a. aufgrund der Relativierung des Unabhängigkeitskriteriums, eine Vielheit von Substanzen annehmen. Leibniz’ Lösung ist „spiritueller Monismus“ (178): Einzelsubstanzen, die das Ganze der Wirklichkeit auf verschiedene Weisen bewusstseinsmäßig repräsentieren. Dem entspricht auch Leibniz’ logischer Zugang zur Substanzthematik: Einzelsubstanzen sind durch einen vollständigen Individualbegriff bestimmt, der auch ihre relationalen Bezüge zur Gesamtwirklichkeit umfasst. Der dritte Teil ist der „Revision“ des Substanzbegriffs gewidmet. Johanna Seibt („Der Mythos der Substanz“) spricht von Substanzen eben als „mythologischen Instanzen“, die eigentlich, aufgrund des methodologischen und inhaltlichen Fortschritts der Ontologie, bereits überwunden hätten sein müssen. Gegen Mythen ist freilich schwer zu argumentieren (198). So ist es die Aufgabe zu zeigen, wie hinderlich die mangelnde Überwindung der Substanzen für ontologische Problemlösungen sei. Ein Punkt ist die Vermischung von Problemen im Kontext von Identität (numerische Einheit) und Individuation (Individualität, Selbigkeit). Sie ist typisch für das Substanzdenken und auch in Entwürfen zu finden (z.B. in Quines „physical-object“ Ontologie, Bündeltheorien nach Russell und Campbell), die sich eigentlich als Alternativen zu Substanz-Ontologien verstehen – weil diese „Alternativen“ das Substanzdenken ja doch nicht gründlich genug überwunden hätten. Ein weiterer Hinweis sind Probleme in der ontologischen Deutung der Persistenz. „Endurance“-Theorien, offensichtlich im Substanz-Denken verhaftet, scheitern u.a. an Leibniz’ Gesetz (davon wird unten noch die Rede sein), „perdurance“-Deutungen daran, dass sie die Persistenz als Abfolge zeitlicher Teile deuten, die selbst dinghaft – im Sinne charakteristischer Merkmale von Substanzen – zu verstehen sind. Es bleibt der Ausweg einer radikalen Revision, wie sie nur in einer Prozessontologie, etwa nach Sellars, zu finden ist. Auch Peter Simons strebt in seinem Beitrag „The Ties that Bind …“ eine Revision der Substanz-Ontologie an. Was kann prädikative Propositionen wie „Platon ist weiß“ (233) wahr machen? Es sind, wenn man wie Simons nicht auf Universalien rekurrieren möchte, individuelle Substanzen und Eigenschaften oder Tropen. Wie aber hängen diese zusammen? Nachdem es keine plausiblen Theorien des Zusammenhalts von Substanzen und Tropen gibt, plädiert Simons für die Lösung, dass die Tropen selbst es sind, die voneinander abhängend, Dinge des Alltags konstituieren (vgl. 234). Simons unterscheidet dabei verschiedene Typen von Abhängigkeitsrelationen zwischen Tropen. An der Basis steht die Fundierungsrelation (nach Husserl), die darauf beruht, dass es Tropen gibt, in deren Natur es liegt, nicht ohne bestimmte andere vorzukommen. Derart zusammenhängende Tropenverbände machen den Kern („Individualessenz“) von alltäglichen Dingen aus. Von Tropen an der „Peripherie“ hängt der Kern generisch ab; d.h. nicht von bestimmten, aber von „Familien“

239

oder Genera von Tropen, etwa Größe, Geschwindigkeit etc. Am äußersten Rand („halo“, 238) finden sich Tropen, von denen der Kern weder individuell noch generisch abhängt. Die dritte Revision der Substanz-Ontologie kommt von Donald Mertz. In seinem Beitrag „Ontic Predicates as Substance“ zeichnet er folgende Alternativen: Die erste ist die „Substratumtheorie“ von Substanzen. Grundlegend seien eigenschaftslose Träger (Substrata), von denen Eigenschaften abhingen. Damit geht einher eine Reduktion mehr-stelliger Eigenschaften auf einstellige, weil nur letztere von einzelnen Trägern abhängen können. Diese Theorie leidet unter den logischen und ontologischen Problemen von Substrata-Theorien, wie die Problematik der Annahme von „bare“ particulars, die ja aufgrund ihrer Eigenschaftslosigkeit keine ontologischen Funktionen, etwa die der Individuation, erfüllen könnten. Mertz plädiert für die zweite Alternative: Grundlegend seien komplexe Strukturen selbst, bestehend aus einem Einheitsprinzip: „ontic predicates“, die verschiedene Bestandteile in die Ganzheit komplexer Strukturen einbinden, und so auch die Individuation dieser komplexen Strukturen gewährleisten. Innerhalb der Strukturen gibt es keine unabhängigen Teile. Die Struktur als Ganzes ist in gewissem Sinne unabhängig und kann so, auf höherer Ebene, wieder Element umfassenderer Strukturen werden. Die Edition hat einige kleinere Schönheitsfehler und eine etwas gröbere Lücke. Zu den kleinen Mängeln gehört, dass nicht alle Beiträge einheitlich gestaltet sind. So fällt der Beitrag von Seibt deutlich aus dem Rahmen (v.a. das Literaturverzeichnis), während die Gestaltung der Beiträge von Hofman und Mertz an Deutlichkeit zu wünschen lässt (z.B. die Überschriftenformate). Dazu kommen einige Unklarheiten betreffend die Frage, ob die von der Herausgeberin angekündigte schematische Gliederung in „Proponenten“ (Teil I), „Opponenten“ (Teil III), und dazwischen (Teil II) die historischen Beiträge, wirklich durchgehalten wird. Ob z.B. Hofmanns „wissenschaftlicher Atomismus“ (96f ) als SubstanzProponenten-Standpunkt zu verstehen ist, kann in Frage gestellt werden (was der Qualität seines Beitrags keinen Abbruch tut). Auch bilden die Überschriften zu den Hauptteilen nicht wirklich, wie man heute zu sagen pflegt, die oben genannten Intentionen und Inhalte dieser Abschnitte ab: Ob sich etwa die systematische Befürwortung und Verteidigung (Proponenten) von Substanzen in der Angabe von „Kriterien des primär Seienden“ erschöpfen, kann ebenso bezweifelt werden, wie eine Präsenz von Locke im zweiten Teil, die es rechtfertigen würde, ihn in der Überschrift auf gleicher Ebene mit Aristoteles und Leibniz zu nennen, (die freilich ganz hervorragend behandelt werden). Trotz dieser kleinen Mängel gibt die Edition einen guten Überblick über einige Konfrontationslinien in der Auseinandersetzung um das Thema „Substanz“, und vermittelt eine differenzierende

240

Betrachtung, v.a. der aristotelischen Geschichte dieser Kategorie. Dies macht zweifellos eine der inhaltlichen Stärken der Edition aus. Als gröbere Lücke betrachte ich das Fehlen eines Abschnitts mit meta-ontologischen Reflexionen. Um diesen Mangel deutlich zu machen, muss ich etwas ausholen und werde zunächst drei der inhaltlichen Konfrontationslinien in der Auseinandersetzung um Substanzen näher unter die Lupe nehmen, die in den Beiträgen des dritten Teils im Mittelpunkt stehen. 1. V.a. in D. Mertz’ Beitrag ist die Verpflichtung auf „Substrata“ oder „Träger“ von Eigenschaften durch die Substanz-Ontologie entscheidendes Motiv zu einer Revision derselben. Substrata seien, weil selbst Träger von Eigenschaften, eigenschafts- folglich bestimmungslos. Sie tragen m.a.W. Eigenschaften, haben jedoch keine. Ihre Akzeptanz hat nicht nur empirische Probleme, sondern auch (onto-)logische. Worauf gründet sich die Annahme solcher „reinen“ oder „bare“Substrata, wenn sie empirisch vollkommen unzugänglich sind? Und selbst wenn es sie gäbe: Können sie die von ihnen erwarteten ontologischen Funktionen erfüllen, z.B. wie Mertz herausstreicht, die Individuation von Entitäten? – Ich möchte hier keine Lanze brechen für eine „bare-substrata“ Theorie, jedoch darauf hinweisen, dass es sich bei derselben um das Resultat einer historischen (Fehl-) Entwicklung in der Substanz-Ontologie handelt, auf die man sich allein mit der Akzeptanz von Substanzen nicht verpflichtet. (Das ist auch Wiggins’ Anliegen, siehe oben.) Dahingestellt sei, inwiefern sich bare-substrata-Theorien auf Aristoteles beziehen können. In der Neuzeit ist es v.a. John Locke, der unter Substanzen nicht ganze Dinge bzw. ganze Lebewesen versteht, sondern eben bare-substrata, welche als selbst eigenschaftslose Träger von Eigenschaften der Dinge bzw. Lebewesen angenommen werden. Diese Träger sind somit, wie Locke eingesteht, ein „something we know not what“2. Wie aber kann man als Substanz-Ontologe die Verpflichtung auf Lockes „something we know not what“ vermeiden? Was ist die Sackgasse? Wo ist der Ausweg? Zuerst zur Sackgasse. Die „bare-substrata“ Theorie hängt m.E. an einer Interpretation eines Alltagsdinges bzw. eines Lebewesens als Zusammenfügung, bestehend aus Eigenschaften, dem Träger dieser Eigenschaften, sowie Relationen („Inhärenzen“), welche Träger und Eigenschaften verbinden. Das Bild einer Traube, an der man nur die Beeren, aus prinzipiellen Gründen aber weder deren Stiel noch die Verbindungsstücke zwischen Beeren und Stiel sehen kann, mag hier (cum grano salis) zur Illustration dienen. Eigenschaften, egal ob als reine Individuen (Tropen) oder als Instanzen von Universalien interpretiert, sind in diesem Modell „Quasi-Dinge“, die sich von ihrem Träger nur durch ihr „in“ sein 2. John Locke, Essays, chapt. 23, § 2.

241

müssen unterscheiden. Die Inhärenz ist eine zweistellige Eigenschaft, ein QuasiDing der besonderen Art. Dem Träger sind alle Eigenschaften äußerlich, (wie dem Stiel die Beeren). Er ist die Substanz, die selbst nicht „in“ sein kann, und (deshalb) als selbständig aufgefasst wird. Dieses Modell führt in besagte Sackgasse. Was aber ist der Ausweg? Zwei Stellen bieten sich dabei an: 1) Eigenschaften sind keine „Quasi-Dinge“, die 2) durch irgendwelche „dicken“ Relationen (im Sinne Mulligans3) an Substanzen, verstanden als Substrata, gebunden werden müssten. Und ein drittes blockiert das Modell von vornherein: Das ganze Ding bzw. das ganze Lebewesen ist die Substanz. Wie könnte eine solche alternative Substanz-Ontologie aussehen? Ein paar dünne Hinweise, (die in den folgenden Abschnitten 2–4 etwas eingedickt werden): Substanzen sind die ganzen Dinge bzw. Lebewesen. Ein jedes Lebewesen, um bei diesen zu bleiben, besteht aus einem Material- und einem Formaspekt; dem „Woraus“ es ist und der Struktur dieses „Woraus“. Eigenschaften sind Modifikationen eines der beiden Aspekte, gemeinsam mit dem sie jeweils einen Zustand bilden. Eigenschaften sind also Elemente von Zuständen oder „gutted states“ (im Sinne Armstrongs4). Die Verbindung von Eigenschaft und Materialbzw. Formaspekt ist keine dicke Relation, sondern eine durchaus dünne, weil „formale“ Relation. Zustände selbst konstituieren nämlich diese für ihre Einheit maßgebliche Verbindung ihrer Elemente (z.B. im Sinne Tegtmeiers5). Das Ziel kann hier nicht sein, diese Alternative auszufalten. Für den relevanten Punkt ist das auch nicht notwendig. Es reicht der Hinweis, dass wir uns als SubstanzOntologen nur über die Festlegung auf ein sehr problematisches Modell von Substanzen auf eine Substrata-Theorie verpflichten. Diese Festlegung können wir vermeiden. Also können wir als Substanz-Ontologen auch die SubstrataTheorie vermeiden. 2. Das zweite Thema, das im Streit zwischen Befürwortern und Gegnern von Substanzen eine wichtige Rolle spielt, ist die Persistenz, das Fortdauern durch die Zeit. Johanna Seibt hat in ihrem Beitrag ein Dilemma der Persistenz gezeichnet, in dem sich traditionelle Ontologien, nicht nur mit Substanzen, aufspießen 3. Kevin Mulligan, „Relations – through Thick and Thin“. In: Erkenntnis 48, 2–3 (1998), 325–353. 4. U.a. David Armstrong, A World of States of Affairs, Cambridge: CUP 1997, z.B. 29. Armstrong spricht hier zwar von „gutted states of affairs“. Der Sache nach aber sind seine Sachverhalte das, was wir hier als Zustände auffassen. 5. Erwin Tegtmeier, Grundzüge einer kategorialen Ontologie, Freiburg: Alber 1998, 174: „Sicher sind die Bestandteile eines Komplexes irgendwie verbunden. Sonst bildeten sie keinen Komplex. Aber dies beruht, meiner Ansicht nach, nicht auf einem besonderen Bestandteil, der die übrigen miteinander verbindet, sondern auf dem Komplex selbst“. Ich möchte diese (begründete) Einsicht übernehmen und auf Zustände als Komplexe anwenden, um den Verbinder Inhärenzrelation einzusparen.

242

würden. Ein Horn dieses Dilemmas ist allerdings für Substanz-Ontologen reserviert: die „endurance“-Interpretation der Persistenz. Diese Interpretation besagt, dass das Fortdauern diachrone Identität in einem strikten Sinn ist. Eine Substanz ist in einem strikten Sinn durch die Zeit dieselbe. (Das andere Horn des Dilemmas ist die „perdurance“-Deutung der Persistenz, siehe oben). Die Problematik der endurance besteht v.a. im Phänomen der Änderung. Zum einen führt das in einen handfesten Konflikt mit Leibniz’ Gesetz, das für Identität Ununterscheidbarkeit in Eigenschaften fordert. Bleiben Substanzen, wie es das „endurance“Modell vorsieht, trotz Änderungen von Eigenschaften strikt dieselben, steht das Leibniz’ Gesetz entgegen. Zum anderen, und das kann man auch unabhängig von Leibniz’ verstehen, läuft man in die Gefahr eines Widerspruchs, wenn man behauptet, dass ein und dieselbe Substanz eine Eigenschaft hat (vor einer Änderung) und eine Eigenschaft nicht hat (nach einer Änderung). Die Erwiderung von Substanz-Ontologen zur Rettung von „endurance“ und Änderung ist die zeitliche Indexikalisierung des Zukommens von Eigenschaften. Ich möchte hier eine Version dieser zeitlichen Indexikalisierung schildern und zeigen, dass man so auf der Seite der „endurance“ dem Horn des Persistenz-Dilemmas entkommen kann.6 Beginnen möchte ich mit Überlegungen bezüglich des Verhältnisses von Substanzen zur Zeit. Was heißt es überhaupt, dass eine Substanz x eben zu einer Zeit t vorkommt und in der Folge zu t F, sprich Eigenschaften hat, wie es bei der zeitlichen Indexikalisierung des Zukommens von Eigenschaften behauptet wird? M.E. haben Substanzen zur Zeit ein rein akzidentelles Verhältnis. Zeit gehört nicht zu den Konstituenten von Substanzen. Substanzen sind dreidimensional. Den akzidentellen Bezug aber der Substanzen zur Zeit machen andere als substanzielle Partikularien aus. Der Bezug zur Zeit wird für Substanzen dadurch hergestellt, dass Substanzen in andere Partikularien, etwa Zustände, eintreten. (In Anwendung des oben Gesagten wäre zu sagen, dass Substanzen genau dann in einen Zustand eintreten, wenn einer ihrer Aspekte, Material oder Form, modifiziert wird. Im Folgenden ist das gemeint, wenn davon die Rede ist, dass eine Substanz in einen Zustand eintritt.) Chisholm spricht davon, dass Zustände die Geschichte von Substanzen ausmachen7, und es ist nichts anderes als die Geschichte von Substanzen, die ihre Zeitlichkeit bestimmt. Dass eine Substanz zu einer Zeit vorkommt, heißt demnach, dass sie in einen Zustand eintritt, welcher ihm zeitliche Verhältnisse vermittelt. 6. Seibt (218f ) listet insgesamt sechs Möglichkeiten auf, die zeitliche Indexikalisierung des Zukommens zu verstehen. Ich muss es mir hier versagen, diese aufzulisten, zu diskutieren, insbesondere gesondert aufzuweisen, dass mein Vorschlag nicht von den Zurückweisungen Seibts betroffen ist. 7. … wobei er Ereignisse unter die Kategorie der Zustände subsumiert.

243

Der Verweis insbesondere auf Zustände ist deshalb von Bedeutung, weil er auch das Verständnis dessen erleichtert, was es bedeutet, dass ein x eben zu t F ist. Ein Zustand ist ja nichts anderes als das F-sein von x (und das wie gesagt nichts anderes als eine Modifikation eines seiner Aspekte). „x ist zu t F“ besagt demnach, dass eine Substanz x in einen Zustand <x F> eintritt, und dass dadurch sein Eintritt in zeitliche Verhältnisse konstituiert wird. Jedes Eintreten eines x in ein <x F> hat ja zur Folge, dass es ein Eintreten in ein <x F> zu t ist. Ich spreche (vom Zustand aus gesehen) auch davon, dass der Zustand die Substanz in zeitliche Verhältnisse bringt. Entscheidend ist, dass wir uns durch diese Deutung bei der Beschreibung von Änderungen nicht in Widersprüche verwickeln. Nach dieser Deutung tritt ein x, dem eine Eigenschaft F zukommt, in einen Zustand <x F> ein, und folglich in zeitliche Verhältnisse t. Widersprüchlich wäre es lediglich anzunehmen, dass dasselbe x in einen Zustand <x G> eintreten könnte (wobei G und F inkompatible Eigenschaften sind), und in Folge des Eintritts in <x G> in zeitliche Verhältnisse t´, wobei gilt t = t´. Nicht widersprüchlich ist aber die Annahme, dass ein x in einen Zustand <x F> eintritt, und folglich in die zeitlichen Verhältnisse t, und dasselbe x in einen Zustand <x G> eintritt, und folglich in die durch von t verschiedenen zeitlichen Verhältnisse t´. Und genau darin bestehen m.E. Änderungen. Es ist nicht widersprüchlich zu behaupten, dass ein und dieselbe Substanz hintereinander in inkompatible Zustände eintritt, sich eben ändert. Ebenso kann diese Analyse Leibniz’ Gesetz Genüge tun. Man müsste es nur adaptieren: Wenn für ein x und für ein y gilt, dass sie in der gleichen Abfolge in die gleichen Zustände eintreten und durch das Eintreten in gleiche Zustände stets in die gleichen zeitlichen Verhältnisse gebracht werden, dann folgt daraus, dass x und y identisch sind. Im Sinne einer notwendigen Bedingung könnte man sagen, dass Substanzen nur dann identisch sind, wenn sie in der gleichen Abfolge in die gleichen Zustände eintreten und durch das Eintreten in gleiche Zustände stets in die gleichen zeitlichen Verhältnisse gebracht werden. Durch die zeitliche Indexikalisierung des Zukommens von Eigenschaften zu Substanzen kann der Widerspruchsverdacht und der Verdacht der Verletzung des Leibnizschen Gesetzes durch „endurance“-Deutungen der Persistenz zurückgewiesen werden. Dadurch entkommt die Substanzontologin dem von Seibt aufgezeigten Dilemma. 3. Das dritte Thema mit hohem Konfliktpotenzial zwischen Befürwortern und Gegnern von Substanzen ist die (vermeintliche) Unvereinbarkeit einer Substanz-Ontologie mit den Ergebnissen der Quantenphysik. Diese Unvereinbarkeit steht im Hintergrund eines „revisionären“ Beitrags zur Edition, nämlich jenes Peter Simons. Simons Abkehr von einer Substanz-Ontologie hin zu einer Tropen-

244

Ontologie, wie sie hier in “The Ties that Bind … “ propagiert wird8, ist wesentlich geprägt von folgender Einsicht: Substanzen sind unverzichtbar bei der Beschreibung der Grundstrukturen unserer alltäglichen (Lebens-)Welt, aber, und das ist der springende Punkt, eben nur für diese. An der physikalischen Basis der Welt, sprich auf der Ebene der Quantenphysik, gibt es nichts, das den ontologischen Merkmalen der Substanzen entspräche, etwa diachrone Identität etc. Ontologie ist aber eine Theorie der (physikalischen) Basis der Welt. Also habe sie sich von Substanzen zu verabschieden, und tropistisch zu werden. – Ich denke, dass wir diese Schlüsse nicht ziehen müssen. Es kann durchaus eine Versöhnung geben zwischen Substanz-Ontologie und Quantenphysik. Diese setzt freilich einige, wenn man so will Ontologie-theoretische Überlegungen voraus, die sowohl ihren Status als philosophische Disziplin als auch ihr Verhältnis zu den Naturwissenschaften, etwa zur Physik betreffen. Nehmen wir (gegen Simons) an, die Ebene der Alltagswelt sei die „letzte“ oder grundlegende für ontologische Überlegungen. Ontologie ist eine Theorie der Alltagswelt. Nehmen wir weiterhin hin, Substanzen seien (wie auch Simons einräumt) irreduzible Bestandteile der Alltagswelt. Dann sind Substanzen tatsächlich so etwas wie Bestandteile des grundlegenden Inventars der Ontologie. Wie aber kommen wir von hier zur Berücksichtigung der Ergebnisse der Quantenphysik? Substanzen, das habe ich oben angedeutet, bestehen aus einem Material- und aus einem Formaspekt. Es ist durchaus legitim, beide Aspekte weiter zu analysieren. Eine Analyse dessen, was ontologisch gesehen der Materialaspekt von Substanzen ist, kann unter verschiedener Rücksicht geschehen, auch unter naturwissenschaftlicher. In Anwendung der methodisch eingeschränkten Zugangsweise z.B. der Physik kann es zu interessanten Ergebnissen einer solchen Analyse kommen. Die Quantenphysik hat hier ihren Ort. Die Ergebnisse der Quantenphysik sind also relevant als solche, sprich als Ergebnisse einer nach Maßgabe einer Naturwissenschaft methodisch eingeschränkten Forschungsperspektive. Es gibt aber keine Ontologie der Quantenphysik, die darüber hinausginge, dass die Quantenphysik eine Physik des Materialaspekts von Substanzen ist. (Darüber weiter zu reflektieren wäre Aufgabe von Naturphilosophie bzw. von Wissenschaftstheorie; erstere mit Hauptaugenmerk auf das Objekt, letztere auf die Zugangsweise zu demselben.) Manche werden hier sicher einwenden, dass dies falsch sei, weil es den methodischen Unterschied zwischen Ontologie und Naturwissenschaft zu hoch ansetze. Die Ontologie sei eine Disziplin, die sich den Forschungsergebnissen der Naturwissenschaft, in unserem Fall der Quantenphysik, zuwenden müsse, um daraus ihre Theorien 8. Siehe dazu v.a. auch Simons’ Artikel „Farewell to Substances …“ in: Ratio (new series) 11 (1998), 235–252, sowie “Particulars in Particular Clothing …” in: Philosophy and Phenomenological Research 54 (1994), 553–576.

245

abzuleiten. Sie sei eine induktive oder a posteriorische Disziplin. Ontologie sei Naturwissenschaft, mit etwas allgemeinerer Begrifflichkeit. (Naturphilosophie brauche es nicht mehr, die wissenschaftstheoretische Fragestellung sei von vornherein entschieden.) 4. Hier zeigt sich m.E. der eigentliche Ort der Auseinandersetzung um Substanzen. Damit bin ich bei der oben angekündigten grundlegenden Bemerkung. Ist die Ontologie tatsächlich induktiv oder a posteriorisch, wird sie sich in ihrem Aufbau an den Ergebnissen der (Quanten-)Physik orientieren müssen. Dann aber wird sie, wie Simons zu Recht meint, Substanzen „Lebewohl“ sagen. Tropen sind dazu weit besser geeignet. Dieser Auffassung einer induktiven Ontologie entspricht das Postulat der Abwendung von unserer alltäglichen Lebenswelt als eigentlichem Objekt der Ontologie und einer Hinwendung zu ihrer physikalischen Basis. Manche haben dies die Abkehr von „deskriptiver“ bzw. die Hinwendung zu einer „revisionären“ Ontologie genannt. Ist man freilich der Auffassung, dass ein solches Verständnis von Ontologie falsch und die Ontologie gegenüber jeder Naturwissenschaft methodisch und in ihrer Erklärungsweise unabhängig ist, dann kann man diese Konsequenz blockieren. Dann steht es einem auch offen, die Analyse der Grundstrukturen der alltäglichen Lebenswelt als genuin ontologisches Unterfangen zu verstehen. Bei einer solchen Analyse haben Substanzen aber eine irruduzible Funktion. Und damit möchte ich auch schließen: Ich denke, dass es gut und sinnvoll ist, die traditionellen Merkmale der Substanzen, z.B. ihre Unabhängigkeit, ihre Funktion als „Träger“ von Eigenschaften, aber auch ihre Persistenz ontologisch zu analysieren. Selbstverständlich ist es wichtig, hier Fehlentwicklungen aufzuzeigen, zu kritisieren, u.U. sogar zum Schluss zu kommen, dass Substanzen hoch problematische Gebilde sind. Ich meine jedoch, dass die Substanzdebatte zu kurz greift, wenn sie nicht die entsprechenden Implikationen bzgl. eines grundlegenden Verständnisses von Ontologie und ihrem Verhältnis zu den Naturwissenschaften mitberücksichtigt. Zumal sowohl Befürworter als auch Gegner von Substanzen ihre Argumentation stets vor dem Hintergrund eines solchen grundlegenden Verständnisses von Ontologie entwickeln, oft leider, ohne darüber zu reflektieren. Meta-Ontologie ist angesagt, wenn wir uns nicht auf Nebenschauplätzen in endlose Scharmützel ohne Erfolgsaussichten ergehen wollen.

246

Grazer Philosophische Studien 76 (2008), 247–249.

Wayne M. MARTIN: Theories of Judgment: Psychology, Logic, Phenomenology. Cambridge: Cambridge University Press, 2006. xiv + 188 pp. ISBN 0-521-84043-0. $ 75.00. Though theories of judgment are not among the most hotly disputed items in contemporary philosophy, they most certainly were so for long stretches of the philosophy’s past. This book is of considerable value insofar as it provides us with discussions of such theories and may be of great service in drawing attention to the unsettled issues under consideration as ones worthy of attention in future philosophical investigations. In the introduction the author maintains that the range of phenomena to be treated in theories of judgment may be identified by means of “some tautologies and examples”, namely “Judgment is what judges do”, “A judge is a figure of authority and responsibility”, “Some judgments are snap judgments”, and “Judgment occupies a place in both theory and practice” (1 f.). The phenomena that are thereby identified, according to Martin, have been examined from three points of view, called “faces of judgment”. The first of these faces is psychology, which is concerned with “the explanation of the behavior of intelligent organisms” (3). The second one is logic, which is concerned with inferential structure. In this context Martin significantly remarks, “The centrality of the theory of judgment has been somewhat submerged in modern mathematical logic, but we shall see that debates in the logical theory of judgment were at the heart of the revolution that gave rise to the modern logical tradition” (4). The third face of judgment is phenomenology, which Martin characterizes as “the study of the structure of experience, particularly of the way in which things (entities, objects) manifest themselves in experience” (4). In this review I will restrict myself to comments on Martin’s considerations regarding the history of philosophy and will not concern myself with the fifth and final chapter that deals with the depiction of judgment in

art, which is a topic beyond the scope of my expertise. The first chapter is concerned with theories of judgment from a psychological point view and is specifically focused on three experimental approaches. Two of these are contemporary, whereas the other one is found in David Hume’s Treatise of Human Nature. Martin is here particularly concerned with Hume’s attempt to identify the characteristic of belief which distinguishes it from mere conception. While Hume’s result, namely that belief is characterized in terms of “force and vivacity” and by “feeling”, is found wanting, the author nonetheless finds Hume’s experiment instructive insofar as both phenomenological and logical considerations come into play therein. According to Martin, Hume is engaged in phenomenology insofar as he inspects his consciousness in order to isolate the distinguishing characteristic of belief. The logical aspect of Hume’s approach is found in his assertion that we do not add the idea of existence to another idea whenever we think that something exists, e.g. that we do not add the idea of existence to the idea of God when we think that God exists. Martin also discerns phenomenological and logical factors at work with regard to the contemporary experimental approaches which he discusses. The second chapter is concerned with Kant’s theory of judgment as synthesis and is especially focused on the logical aspects thereof. In this regard Kant’s distinction between analytic and synthetic judgments and his table of judgments are discussed. The great challenge for a defender of Kant is to determine where existential judgments are to be identified by means of applying the former distinction and the latter table (thus to classify existential judgments in terms of quality, quantity, relation, and modality). In his endeavor to elaborate on this challenge he draws from both the Kritik der reinen Vernunft and Kant’s lectures on logic, though he finds the possible solutions unsatisfactory for various reasons. It is a great credit to Martin that, in view of this shortcoming on Kant’s part, he devotes some discussion of the logic that arose

in the Herbartian school, specifically that of Moritz Drobisch. While Herbart and his students made a sharp distinction between existential (or “thetic”) judgments and categorical (or “synthetic”) ones, Martin still finds in the logical work of Drobisch a strong tendency to characterize judgments in terms of the subject-copula-predicate model, which is only appropriate for categorical judgments. At the end of the nineteenth century Franz Brentano stands out as a towering figure who attempted to solve the problem of assimilation of existential judgments into logic by rejecting the subject-copula-model altogether and thus by characterizing all judgments as existential. The second chapter accordingly ends with a discussion of this strategy in the theory of judgment. Unfortunately, however, Martin’s chronology is inaccurate, for he says, “Brentano’s calculus main logical doctrines were first set out in 1874, and his calculus was elaborated in detail by 1877” (65). In Psychologie vom empirischen Standpunkt (the work of 1874 to which Martin refers) Brentano explicitly says that his logical doctrines and particularly the relevant theory of judgment were already expounded upon in his Würzburg lecture course given in the winter semester of 1870/71.1 Here mention should of course be made of the fact that logic for Brentano is a practical discipline which draws upon various theoretical ones, especially upon „descriptive psychology“, also called „descriptive phenomenology“ (beschreibende Phänomenologie).2 The three faces of judgment accordingly make up a unity in Brentano’s theory. While Martin shows some appreciation of the connection between logic and phenomenology for Brentano, insofar as he is well aware of the fact that for Brentano both affirmative and negative judgments are to be discerned and distinguished from other acts of consciousness simply on the basis of inner perception, Martin is nonetheless off the mark when he characterizes such perception as „attentive introspection“ (73), for Brentano in fact takes great pains to distinguish inner perception from any sort of attentive directedness to mental states, so-called „inner observation“ (innere Beobachtung).3 If indeed phenomenology would require such observation, as

Martin seems to think it does, it would be an impossible endeavor from Brentano’s perspective. Certainly Martin is to be commended for giving Brentano his due as one of the most outstanding judgment-theorists. However, his account of Brentano’s contribution in this regard is not free of inaccuracy. In the third chapter Martin attempts to make good his promise to show the relevance of the theory of judgment to the origins of modern mathematical logic by discussing Frege’s introduction of the judgment stroke. Though contemporary logicians work with propositions, formalized by single letters such as p and q, in Frege’s Begriffsschrift we find these prefaced by a turnstile (A), which symbolizes judgments rather than mere propositions. Consideration is given by Martin to various criticisms of this aspect of Fregean logic and also to various defenses thereof. While he does not recommend assimilating the judgment stroke into the logical calculus, he emphasizes that Frege’s logic was indeed meant to be a logic of judgment and not merely a logic of judgment-contents (“propositions”) and points out that this is particularly a problem for Frege, who wanted to dispel all psychological considerations from logic. Since judgments are presumably mental acts, they should have no place in logic as Frege conceives of it. At the same time Martin thinks that Frege insisted upon using the judgment stroke in logic, because logic was, according to Frege, first and foremost concerned with truth. This is to be seen in Frege’s suggestion that the judgment stroke can be seen as equivalent to such predications as “The violent death of Archimedes at Syracuse is a fact” instead of “Archimedes died a violent death at Syracuse”. In the fourth chapter Martin comes to consider the phenomenological face of judgment as this is elaborated on in Martin Heidegger’s dissertation. In view of the failures of Hume, Kant, and presumably Brentano as well, to develop a phenomenology of judgment, Martin finds in this early and usually neglected work of Heidegger the key to such an endeavor. According to Martin, Heidegger establishes that there are logical objects, namely judgments as not identical with the mental states in which one judges,

248

by considering variations within consciousness. While Martin fully appreciates that Heidegger owes his conception of logical objects and his method of discovering them to Husserl (though not without criticism) and also to Herman Lotze (from whose logical work Husserl, incidentally, also had drawn), it is rather baffling why Martin does not rather give consideration to Husserl as the outstanding representative of phenomenology. It would have been interesting also if Martin had given consideration to twentieth century developments such as Meinong’s theory of objects and assumptions, for in this endeavor on Meinong’s part problems in the theory of judgment which were dealt with by Brentano and his other affiliates received extensive treatment. Though it is unreasonable to expect a volume of this size or even a much larger one to be an exhaustive discussion of theories of judgment since Hume and Kant, it nonetheless seems that the chapter on Heidegger could have well been replaced by a chapter on more interesting and innovative theories. It is moreover hardly of interest with respect to theories of judgment that Heidegger’s dissertation, as Martin asserts, can somehow be read from the standpoint of the later Heidegger, armed with his cryptic talk about an ontological difference. This being said, Martin’s work should be welcome as a discussion of theories which have long been in danger of being forgotten and still have great bearing on problems

that are very much alive not only in psychology, logic, and phenomenology, but also in epistemology and philosophy of language. It is to my mind especially worthwhile to read the second chapter of this volume (in spite of the inaccuracies in connection with Brentano), though the third also merits attention from the many who consider Frege to be the great innovator in contemporary philosophy. Whether the judgment stroke be acceptable or not, it remains a noteworthy historical fact that theories of judgment were at the center of attention during the early days of the analytic tradition as well as the phenomenological one. Robin D. ROLLINGER Universität Salzburg

1. Franz Brentano, (ed.) Oskar Kraus, Psychologie vom empirischen Standpunkt. Zweiter Band: Von der Klassifikation der psychischen Phänomene (Leipzig: Felix Meiner, 1925), 77 f., n. 2. Franz Brentano, (eds.) Wilhelm Baumgartner and Roderick M. Chisholm, Deskriptive Psychologie (Hamburg: Felix Meiner, 1982), 129. 3. See Franz Brentano, (ed.) Oskar Kraus, Psychologie vom empirischen Standpunkt. Erster Band (Leibzig: Felix Meiner, 1924), 40–48.

249

Grazer Philosophische Studien 76 (2008), 250–254.

Anna SIERSZULSKA: Meinong on Meaning and Truth. Frankfurt et al.: Ontos Verlag, 2005. 262 pp. ISBN 3-937202-94-3. € 78,00. The publication of the volume Untersuchungen zur Gegenstandstheorie und Psychologie (1904), containing Meinong’s essay Über Gegenstandstheorie, aroused considerable reaction in Europe. The greater part of the criticisms were directed against nonexistent objects, particularly against the socalled impossible ones, as Meinong himself noticed: “During the last few years there has perhaps not been a single instance in which the first impact with the works of object theory then available has not immediately led to a vigorous, often impassioned attack against ‘impossible objects’” (GA V, 220 fn. 1).1 In more recent debates on Meinong’s philosophy, scholars’ opinions are still divided regarding the assumption that there are non-existent objects. The overwhelming attention given to round squares and gold mountains had as its consequence that other Meinongian theories of ontological, semantic and epistemological nature, which are also of outstanding interest, were often neglected. Anna Sierszulska takes a position precisely on this issue: if we accept “that the theory of objects, as conceived by Alexius Meinong, is not only an inquiry into ontology but also an inquiry into semantic issues”—she writes at the beginning of her interesting book—, “the assumption of controversial objects like nonexistent and contradictory ones immediately becomes understandable” (9). Sierszulska’s study, while leaving aside the development and stratification of object theory over time, is mainly a careful analysis of the theory of the so-called objectives which are related to the concepts of meaning and truth. These are examined not only in their manifold aspects and implications in connection with the theories of Bolzano, Husserl, Frege and Russell, but also with reference to contemporary interpretations of Meinong. In general, Sierszulska interprets Meinong as nominalist in ontology and realist in epistemology, as we shall see;

in particular her new interpretation of objectives diverges significantly from the standard one, according to which objectives are like states of affairs. The volume consists of a brief introduction, four parts spread over nine chapters, and a conclusion, followed by a bibliography and an index of names. In the first part, entitled “Objectives and other objects of intentional reference”, some basic concepts of object theory are introduced. Sierszulska starts by claiming that the relation of metaphysics, ontology and object theory is one of progressive inclusion. Metaphysics is the narrowest of these three disciplines, since it deals only with what there is; ontology involves also the possible (what could be); object theory, finally, deals with whatever we can think or speak about (including impossible objects, like the round square). She then delineates a wellknown Meinongian classification of objects in the following way: real (i.e., material or mental) objects exist, ideal objects (like mathematical ones) subsist, non-ideal objects, that is, impossible or fictional objects but also the theoretical entities of science, neither exist nor even subsist. According to Sierszulska, the primary mission of Meinong’s object theory is to account for all non-real objects, that is, for all entities which do not have actual existence, like mathematical, logical and linguistic objects (10–11, 14–15). Since Meinong shares Brentano’s thesis of the intentional directedness of all mental acts, it is necessary to explain what such acts are directed to. In order to give a non-subjective account of all objects of intentional and linguistic reference, Meinong introduces the idea of Aussersein, which is, according to Sierszulska, fundamentally “an idea of semantic character” (22). More precisely, the “Aussersein is a domain of objects constructed as all possible arbitrary combinations of properties, and only some of these objects coincide with real objects” (9). “Possible arbitrary combinations of properties” are meant to include not only combinations of compatible properties, but also combinations which cannot be exemplified by any possible objects (like “round

square”). Objects belonging only to Aussersein do not possess any being, neither existence nor subsistence, but they are still something (instead of nothing), since one can think or speak about them (while it is impossible to think or speak about nothing). This implies a weaker notion of an object, insofar as it also applies to non-subsistent objects. According to Sierszulska, these latter, together with ideal subsistent objects, are not objects in the traditional sense, but meaning-objects, which are correlated to linguistic expressions but differ from their subjective mental contents; meaning-objects have intersubjective character and hence they permit linguistic communication (22–24). Sierszulska’s next step is to expose the objects of higher order as different from simple objecta. The former are objects which are dependent on other objects as their inferiora, in the sense that they presuppose other objects on which they are built, the existence or subsistence of the inferiora is however not required (27). According to Meinong’s essay Über Gegenstände höherer Ordnung und deren Verhältnis zur inneren Wahrnehmung (1899), objects of higher order are relations (like the similarity between two objects) and complexes (for example a melody, a red square). A complex consists of its component parts and the relation holding between them. The distinction made above between real and ideal objects applies also to objects of higher order. The main difference is that ideal objects are founded with necessity but real ones are not: for instance, the ideal relation of difference is necessarily founded if two different objects (like two colours) are given, but the real relation between a colour and its (subjective) location, in which we represent it, does not hold with necessity (because I can think of the colour which I think in this place as being in another place, or I can think of another colour as being in the same place) (GA II, 398). According to Sierszulska, but not to Meinong—as we shall see—, all objects of higher order, whether they are real or ideal, cannot exist but at most subsist (33). Sierszulska’s Meinong is a moderate nominalist, for whom reality “consists solely of ordinary objects as concrete pieces of substances” (32), qualities are particular moments

of substantial bearers, and there are no universals; ideal objects of higher order (like the difference between two colours A and B) are also particulars, because they are built on particular objects (A and B). More precisely, she maintains that Meinong is a moderate nominalist, insofar as he accepts general objects, some of which (like mathematical ones) subsist, while incomplete objects (the triangle, the ball, the red) do not even subsist. According to her, incomplete objects are not universals, even though they are general (35-36, 65, 157). This distinction between general objects and universals—as we shall see—is important for the concept of the objective proposed by Sierszulska. Unfortunately, she only affirms it, without giving a clear explanation. This is regrettable, not only considering that some philosophers have interpreted Meinong as a realist with respect to universals, but also because Sierszulska’s interpretation seems to be at odds with Meinong’s own remarks. Consider, for instance, what he says in the Selbstdarstellung: “The relationship between such incomplete objects on the one hand and Platonic ideas and universals on the other is unmistakable” (GA VII, 19; cf. also VI, 208, 740). I think that Sierszulska’s interpretation of Aussersein is correct, but her identification of real objects with existent objects is not. According to Meinong, an object is real if it can exist, that is, if its nature allows it to exist, independently of whether it actually exists or not (GA II, 394). The notion of possibility plays an important role in the case of ideal objects too, insofar as Meinong calls those objects ‘ideal’ which by nature cannot exist but only subsist. When he writes that the being of these objects “can be none other than subsistence, if indeed they have any being at all [falls ihnen überhaupt eines zukommt]” (GA IV, 74, my italics; cf. also II, 395; IV, 63–64), he means—as I understand it—that there are ideal objects which actually do not even subsist. Of course, we can disagree with Meinong by defining the notions of real and ideal on the basis of the possibility of existence and subsistence, respectively; but arguing with Meinong is a different thing from attributing our thoughts to him tout court. The notion of reality raises problems also as

251

regards complexes. We have seen that, according to Sierszulska, real objects exist, while real complexes cannot exist but at most subsist (27, 33). Now these propositions cannot both be true: either the word ‘real’ in the latter proposition means something different than in the former, or the word ‘objects’ in the first proposition does not concern objects of higher order, but only simple objecta. If real objects of higher order cannot exist, then only simple objecta—which are, according to Meinong, a limiting case (GA II, 422)— can exist. But what are these simple objecta? In the 1899 essay Meinong questions the very notion of a simple object: he considers as a real object of higher order not only the relation between two representations, but also a coloured surface (GA II, 395, 398; cf. also VII, 18), that is a primum of the perception; therefore it becomes difficult to find perceivable objects that are not objects of higher order and it even seems that their inferiora are achieved through a process of analysis by thought (cf. M. Manotta, La fondazione dell’oggettività. Saggio su Meinong, Macerata, Quodlibet, 2005, 75–76). Let us now turn to Sierszulska’s treatment of objectives (i.e. the objects of judgments and assumptions). She says that these are objects of higher order because they are dependent on other objects, which can be either objecta or other objectives, but they differ from complexes and relations: unlike all real objects of higher order, objectives cannot be real and are timeless; unlike complexes, they do not possess components; moreover, unlike ideal objects of higher order, objectives are not founded with necessity. Their dependence consists in presupposing other objects, but such a presupposition should be understood only in the sense of a logical priority. This means that the subsistence of an objective does not presuppose the being (existence or subsistence) of its inferiora; actually, there are true (and hence subsistent) objectives about non-subsistent objects like ‘that Pegasus has wings’ (44–45, 65). According to Sierszulska, “objectives are not, strictly speaking, objects of any kind”, they are rather means to apprehend complexes together with relations. An objective is the “integrating factor [integrierendes Moment]” of a complex,

that is, a “coordination that obtains between the components of a complex” (46). In short, an objective is an abstract entity, a meaningobject, “belonging to Meinong’s spectrum of pure objects” (40). In the second part of the book, entitled “Meaning and truth”, Sierszulska continues to expound her interpretation of the objective, which is radically different from the standard one. She interprets Meinong’s semantics as a three-levelled theory: sentences express objectives which are the means for apprehending complexes. Against the opinion of many scholars, Sierszulska claims that objectives are not states of affairs—and in fact she criticizes R. M. Chisholm and P. Simons on this point (155–162)—, because states of affairs consist of real objects and universal properties, while for Meinong objectives do not consist of real objects and there are no universal properties, but only general incomplete objects (65). Meinong’s objectives “are simple entities and do not possess components” (69). An objective that obtains in reality is not a fragment of reality, and hence cannot be identified with a state of affairs. If objectives are not states of affairs, what are they then? Sierszulska’s answer is: objectives are function-like entities. Here is a brief summary of her argument. Objectives are meaning entities, related to language and very similar to Frege’s thoughts—and to Bolzano’s propositions in themselves (71–72, 79– 83)—, from which they are distinguished by the fact that, while all thoughts belong to the ideal sphere of the third world, all objectives belong to Aussersein and only those which are true are ideal (72). (Here Sierszulska’s identification of ‘ideal’ with ‘subsistent’ is conspicuous, otherwise the analogy between Meinong’s objectives and Frege’s thoughts would work better). An objective is the immediate object presented by the content of a judgment or an assumption, it is also the truth bearer, while the truth maker is the factual objective. A factual objective is for Meinong a true objective, which subsists. Since—as aforesaid—an objective is the “integrating factor” of a complex through which the latter is apprehended, Sierszulska assumes that it possesses the character of a logical structure, and therefore it may be understood as a function-like entity

252

(73–75). Also the fact that Meinong accepts negative objectives justifies a propositional interpretation of objectives (76). Moreover, objectives are incomplete structures which can take different arguments. And in fact all objects of higher order are incomplete without the objects upon which they depend (77, 83). Finally, just as for Frege whatever is not an object is a function, so for Meinong whatever is not an objectum is an objective (85). To summarise, according to Sierszulska objectives are neither complexes, nor states of affairs, nor fragments of reality, but abstract entities, that is, function-like propositions; “objectives are functions taking arguments from the Meinongian domain of Aussersein and yielding either the value true or the value false” (86). The last sentence directly introduces the issue of truth in Meinong, which is by itself very complex. Actually this part of Sierszulska’s book is crucially important. According to her, the essence of Meinong’s conception of meaning and truth is logical realism (93), where by ‘logical realism’ she means the doctrine that the logical structure grasped in a judgment is identical to the logical structure which obtains in reality, or, in Meinong’s words, that a true objective, grasped in a mental act, is identical to the objective which obtains in reality (96 ff., 102 f.). Now, an objective cannot exist—only a real object can; a true objective subsists and a true judgment, which apprehends a subsistent objective, “should intend something that is correct about reality” (98). There is “a kind of agreement between the objective intended and what is in the world” (99). The thesis just exposed seems to lead to a substantial weakening of object theory, as shall become clear from what follows. What does ‘reality’ mean? Sierszulska writes that an objective is not a fragment of reality “because it includes nothing of the substance of reality, nothing of the world’s matter, nothing of its physical energy” (104). Considering also what she said at the beginning, we can deduce that by ‘reality’ she means all that exists in space-time. But then I do not understand how she can state both that a true objective is one which obtains in reality, and that, for instance, the objective ‘that Pegasus

has wings’ is true (45), since it is clear that the latter does not obtain in reality. Furthermore, ‘that 3 is greater than 2’ is a true objective, and it is so—according to Meinong—independently from reality. True objectives can concern both the sphere of the real and of the ideal, both objects that can exist and those that can subsist or even not subsist; in these last cases, we are not dealing with reality. If indeed Meinong maintains the logical realism that Sierszulska ascribes to him, then he has shot himself in the foot, because he has fallen into that prejudice in favour of the real he has claimed he wants to fight. When he writes that “the totality of what exists, including what has existed and will exist, is infinitely small in comparison to the totality of the objects of knowledge” (GA II, 486), he intends to say that the world in its totality— here ‘world’ is not synonymous with ‘reality’—is ampler than reality. Hence, attributing to Meinong the thesis that “a proposition is true only if it robustly obtains, i.e. if it is ‘correct about reality’” (101), implies questioning the fundamental aim of object theory to be a daseinsfreie Wissenschaft, that is, an a priori science, independent from experience (GA V, 239, 256–257). For correctness’ sake it must be said that later on Sierszulska relativizes her thesis: “Objectives are meaning entities, possibly correlated with certain complexes of objects in the world. In the case of negative objectives or those concerning non-existent objects, of course there is nothing to correlate them with” (182, my italics). But then, in these cases, what is a true objective? This question has to be answered, if object theory deals with non-real objects. And why continue to repeat till the end that “[t]rue objectives are factual due to reflecting the structure of reality”? (251). The third part (“Meinong’s truth in the eyes of his critics”) begins by discussing Russell’s criticism of Meinong’s conceptions on truth, and Meinong’s replies to Russell. Afterwards, Sierszulska leaves off the textual analysis and discusses selected contemporary scholars’ interpretations of objectives, in answer to which she reaffirms her own. Her choice is based on the analysis of specific matters such as the notion of a true judgment, timeless truth and others. The fourth

253

and last part (“Meinong’s theory in the perspective of philosophical semantics”) shows how the concepts previously examined allow us to treat the matter of reference from the point of view of a Meinongian semantics. The main features of some Meinongian semantics (by T. Parsons, E. Zalta, D. Jacquette and J. Paśniczek) are then examined. Thus it is clear that Sierszulska’s book is interesting in many respects and offers several occasions for discussion and exchange of ideas. I consider praiseworthy her attempt to get away from Meinong’s terminology and to translate his concepts into terms that can be accepted more readily by the research community. On the other hand, I do not agree with her tendency to neglect, rather than to discuss, those elements of Meinong’s thought which do not easily fit in with her understanding of object theory. Such elements cannot be simply ignored but must be taken seriously, if for no other reason than to

explain why we believe that certain theses are not convincing, or are by now outdated, or even are not consistent with other theses maintained by the author. It is precisely this tension between an author’s thought and our own which makes reading the author admittedly difficult and complex, but also interesting and enriching. Venanzio RASPA University of Urbino “Carlo Bo”

1. Page references of Meinong’s works are preceded by the abbreviation GA (Alexius Meinong Gesamtausgabe, hrsg. von R. Haller und R. Kindinger gemeinsam mit R. M. Chisholm, Graz, Akademische Druckund Verlagsanstalt, 1968–1978); the other page references refer to the book here examined.

254

Grazer Philosophische Studien 76 (2008), 255–258.

Graham PRIEST: Towards Non-Being. The Logic and Metaphysics of Intentionality. Oxford: Clarendon Press, 2005. xv + 190 pp. ISBN: 978-0-19-926254-0. $ 46,80. (Hardback). Paperback: 2007. ISBN: 978-0-19-923055-6. $ 34,00. In this book, which already attracted a great deal of attention, Graham Priest develops and defends a new version of Richard Routley’s socalled “noneism”, which is itself a version of Meinongianism. According to noneism, “concrete objects exist; everything else (abstract objects, worlds, merely possible objects, impossible objects) simply do not exist” (vii) but they nevertheless have intrinsic properties and stand in relations to other objects (nonexistent as well as existent ones). The main difference to “classical” Meinongianism consists in the fact that within noneism no distinction is made between different kinds of being (existence and subsistence or the like). The book consists of two parts. In Part I, Priest discusses the paradox of the intentional directedness to nonexistent objects (thereby providing one of the main motivations for noneism), and he develops the formal basis of his theory, i.e., a semantics of intentionality. The paradox of the intentional directedness to nonexistents may be stated as follows. It is prima facie true that we may think about, imagine or fear things that do exist as well as things that do not exist. This alleged “intentional directedness” to nonexistent objects seems to be “paradoxical” for the following reason: to think about, imagine or fear an object seems to amount to standing in a particular relation to the object in question. But how can one stand in a relation to something if the second term of the relation does not exist? The noneist solution to this paradox is simply to deny that there is an indissoluble connection between existence (or being in a wider sense) and being something and “simply to accept that an agent can have a relationship with a non-existent object” (57). Priest’s noneist semantics rests on the assumption that there are worlds over and

above the actual world (possible as well as impossible ones). All worlds are supposed to share the same domain. But not all objects of the domain are supposed to exist in all possible worlds. (13, 15–19) The essential innovation in Priest’s formal language is the introduction of “intentional operators”, which stand for locutions such as “believes/knows/fears that” etc. Furthermore, he introduces a special kind of quantifiers, which are supposed to have no ontological import whatsoever. Existence is expressed by means of a predicate. No distinction between “there is” and “exists” is made. (13) Nonexistent objects are not granted any kind of being. (14) In Part II, Priest fends off Russell’s and Quine’s criticism of Meinong (Chapter 5), develops a noneist theory of fictitious objects (Chapter 6) and of mathematical objects and worlds (Chapter 7) and discusses several objections against noneism (Chapters 7 and 8). In Chapter 4, Priest answers the question of what properties nonexistent objects have. A “naive” answer to this question is provided by the “characterization principle” (“CP”): “if A(x) is any property, or conjunction of properties, we can characterize an object cA, and be guaranteed that A(cA).” (83) In other words, according to this principle, a (nonexistent) object has all those properties which we use in order to characterize it. As Russell had noted already, this principle has an unacceptable consequence: if existence is considered to be a property, we may characterize arbitrary objects (winged horses, round squares etc.) as existent, and the principle seems to entail that these objects exist. Thus, according to the characterization principle, an object that is characterized as an existent round square is existent, which seems to contradict the assumption (shared by Meinongians and Anti-Meinongians) that no such thing as a round square exists. As Priest observes, [t]he standard response, from Meinong onwards, has been to accept [CP] only

if the properties deployed in CP are of a certain kind: assumptible, characterizing, nuclear, the names vary. And existence (among others) is not such a predicate. The problem for this line is to give a principled characterization of what constitutes a characterizing predicate and why. No one, as far as I am aware, has been able to do this. Certain classes of predicates can be circumscribed and deemed safe. But without an appropriate rationale, it is difficult to avoid the feeling that the class has been gerrymandered simply to avoid problems. (83) Priest rejects the option to distinguish different kinds of properties (nuclear existence and extranuclear existence or the like) and offers the following as his own solution to this problem: I suggest, the object characterized by a representation has the characterizing properties, not necessarily in the actual world, but in the worlds (partially) described by the relevant representation. Thus, Holmes has the properties he is characterized as having not at this world, but at those worlds that realize the way I represent the world to be when I read the Holmes stories. And Vulcan has the properties it is characterized as having at those worlds that realize the theory of the nineteenthcentury scientists who postulated its existence. (84) I would like to raise the following objection to this proposal. When I talk about Sherlock Holmes, then I talk about a (fictitious) object of the actual world. If there are (merely) possible worlds, then there are surely possible worlds in which the character to which I refer with the name “Sherlock Holmes” does not exist. Even worse, particularly in those worlds in which the Holmes-stories are “realized”, the fictional character Sherlock Holmes does not exist (since in the world of Conan Doyle’s stories, there is no Conan Doyle who writes detective stories). According to Priest’s theory, a name like “Sherlock Holmes” is used to refer to a “real person” who exists in other (non-actual)

worlds. But this, I would like to claim, contradicts “phenomenological experience” (to use one of Priest’s favourite phrases). If I talk about Sherlock Holmes, I do not talk about a human being in another possible world but about a character in the actual world. The question to be answered is: what are the properties of this character in the actual world? The Vulcan case is a bit different, but it raises analogous objections: it may be that some philosophers today use the name “Vulcan” to refer to a planet in other possible worlds. But the astronomers of the 19th century surely have not used the name in this sense. And it seems to me that a theory of nonexistent objects should primarily provide an account of this use of an empty name. At the end of section 4.4, Priest addresses the problem of trans-world identity: “What makes an object, x, at one world, the same object as one, y, at another?” (90) Priest’s answer is: [I]n the semantics we have, objects are just objects; they are not ‘at one world’ or ‘at another’. They have various properties at different worlds, but they are just themselves. (Thus, technically, they are not world-bound entities, but functions from worlds to identities.) But, it may be replied, they may have different identities […] at different worlds. What makes the different identities pertain to the same object? One can, in fact, ask exactly the same question about any aspect of an object. A (concrete) object has different colours (or heights, or weights) at different worlds. What makes the different colours (or heights, or weights) pertain to the same object? Well, that’s just the way things are at that world: at that world, that object has that colour (or height, or weight). Same for identity. That an object has a certain identity is just how things are at that world. (90) I must confess that I do not understand how these remarks answer the above question of transworld-identity. It seems as if Priest considers the question not as meaningful at all. But I can’t see why that should be the case—at least within a theory that rests on the assump-

256

tions that there are many distinct worlds and that there are in all these worlds the same objects. Given these assumptions, it strikes me as legitimate (and even strongly advisable!) to raise the question under what conditions an object a in a world w1 is identical with an object b in a world w2. Priest’s further remarks on this issue are not particularly helpful either. He notes that “[t]he identity of an object is not determined by its properties at any one world”. (90)— But then by what else? He follows Kripke in claiming that “the problem seems to arise because, when one asks a question of this kind, one is thinking of oneself as viewing w2 (through a sort of trans-world telescope) and having to figure out which object at that world is x1 (and which is x2)”. (Ibid.)—But what is wrong with this—given that one is ready to commit oneself to a multitude of worlds? “But identity is not determined by features intrinsic to a world, and so cannot be ‘figured out’ in that way.” (Ibid.)—But how else can it be done? Section 7.1 (entitled “Kinds of non-existent object”) may be taken to shed some light on these issues, for here Priest addresses (finally) the question of the ontological status of (non-actual) possible worlds. There, the reader learns, quite surprisingly, that non-actual worlds themselves are nonexistent objects. (134f.) But this raises another problem: according to Priest’s noneism, Vulcan, for instance, has the property of being a planet not in the actual world but in other possible worlds, which, however, themselves do not exist. But being a planet is for Priest an existence-entailing property. Thus, one has to conclude that Vulcan exists in some non-actual possible worlds. Thus, it seems that nonexistent worlds contain existent objects as constituents. How can that be?—Priest does not address this question. Chapter 6 is dedicated to fictitious objects, which are, according to Priest, paradigm examples of nonexistent objects. Priest rejects the view that fictitious objects are created by the authors of the respective fictional stories. This is simply a consequence of the premises of his theory: Doyle cannot create Holmes in the sense that he brings Holmes into exis-

tence, for Holmes doesn’t exist. But neither can Doyle create Holmes in the sense that he becomes a member of the domain; for Holmes is in any case a member of the domain, independently of Doyle’s deeds. (118–121) “So if Doyle’s activities did not determine Holmes’s status, what was it that Doyle did to Holmes? Simply, Doyle was the first to imagine Holmes, and indeed, to give the character imagined that name, which we now use to refer to him.” (120) Another kind of nonexistent objects (besides fictitious objects and non-actual worlds) are, according to Priest, all kinds of abstract objects (i.e., for instance, properties, relations, propositions and mathematical objects). (135) Priest argues: Abstract objects are just another kind of non-existent object. This at least accounts for the fact that there seems to be a very great difference in kind between ordinary concrete objects and abstract objects. The difference between existence and non-existence is about as great as can be! (135) To put it a little bit maliciously: abstract objects and concrete objects differ strongly from each other. Existent and nonexistent objects too differ strongly from each other. Therefore, abstract objects must be nonexistent objects. Towards Non-Being is a thought-provoking and well-written book that addresses a lot of important issues and contains a number of good points, and the theory that is developed in it is to be taken seriously. Surely, everyone interested in questions of intentionality and ontological commitment will benefit from it. There are, however, also two general (and mutually related) weaknesses (besides some specific internal problems, some of which I have addressed above). First, Priest nicely develops his theory but he is not particularly strong in providing positive arguments for it; and second, important rival theories are not even mentioned, let alone discussed. There are basically three arguments put forward in favour of the theory: Priest claims that it is supported by “phenomenological experience”, that it is “natural” and that it

257

provides a simple solution to a number of paradoxes. A little closer inspection, however, reveals that the first two claims are at least doubtful. Some philosophers find the assumptions that there are nonexistent objects and non-actual worlds extremely “unnatural”. (Besides, some—including the author of this review—find it unnatural to deny that fictitious objects are created by their authors.) Obviously, intuitions strongly diverge in this field, and probably they are themselves partly informed by prior theoretical assumptions (sometimes perhaps by those very assumptions which they are supposed to support). Similar considerations hold for the argument of “phenomenological experience”. Consider how Priest argues for the noneist solution to the paradox of the intentional directedness to nonexistent objects: The noneist strategy is a very natural one. Thus, for example, when one fears something, one has a direct phenomenological experience of a relation to the object of fear. And the phenomenology is quite independent of whether or not the object actually exists. What more appropriate, then, to suppose that objects may exist or not, and that their existential status is irrelevant to whether or not they can be the target of an intentional state? (57f.) I would like to object that to say that in the case of fear we have a “direct phenomenological experience of a relation to the object” is a strong and surely controversial claim. It is true that the phenomenological experience of fear does not depend on whether the “intentional object” exists or not. But to say that this experience is the experience of a relation to the intentional object is far from providing a neutral description of a given experience. Rather, it is already an interpretation—an interpretation that is probably informed by the very metaphysical theory which it is supposed to support. This is not to say that it is absurd or necessarily wrong to assume that we may stand in intentional relations to nonexistent objects. It is just to say that reference to the phenomenology of our experiences is not a strong argument in favour of this assumption. To say the

least, it is far from being as natural and selfevident as Priest suggests. The third of the three arguments (noneism provides a simple solution to some paradoxes) is the strongest one, but it is, of course, far from being cogent. For perhaps there are other simple solutions available, and if so, why should one accept noneism instead of some rival theory? Thus, a really comprehensive defence of noneism would have to include a serious discussion of at least the most important rival theories, which shows that noneism, all relevant aspects considered, is superior to the rivals. But such a discussion is not provided. Actually, Priest discusses only one alternative solution to the paradox of the intentional directedness to nonexistents, namely the assumption that the intentional object is a “mental representation”. According to this assumption, if someone feared the devil, the intentional object of this fear would be a mental representation of the devil, not the (nonexistent) devil himself. Priest—rightly—rejects this assumption. (58f.) However, Priest surely did not pick out the strongest opponent (to put it carefully). In particular, he does not even mention the adverbial theory of intentionality, i.e., the assumption that intentionality is not a real relation at all. According to the adverbial theory, a sentence like “a fears b” does not express a relation between a and b but rather a particular mental state of a, and “b” specifies a’s fear (just as “quickly” in “John runs quickly” specifies John’s running). Priest acknowledges that “there is much more to be said about all this”, but he doesn’t go into it, confining himself deliberately to exploring and developing “the simple and natural noneist strategy” (59). This is, of course, legitimate. But it leaves the task of weighing up the pros and cons of noneism in comparison to other solutions to others.* Maria Elisabeth REICHER University of Berne / University of Graz * The work on this review was supported by the Austrian Fonds zur Förderung der wissenschaftlichen Forschung (FWF), project P19471-G15.

258

Grazer Philosophische Studien 76 (2008), 259–262.

Marie McGINN: Elucidating the Tractatus: Wittgenstein’s Early Philosophy of Language and Logic. Oxford: Oxford University Press, 2006. xiv + 316 pp. ISBN 9780199244447. £40.00/$74.00/€61.50. Marie McGinn’s new book offers an interpretation of some central themes in Wittgenstein’s Tractatus and other early works, an interpretation that she sees as representing an alternative to those that have dominated recent discussion, the so-called ‘metaphysical’ interpretation—associated with Pears, Hacker, Malcolm and others—and the ‘resolute’ interpretation—associated primarily with Diamond, Conant and others. McGinn claims that one can be ‘anti-metaphysical’ without ‘resolutely’ refusing to see in early Wittgenstein ‘genuine philosophical insights into the nature of a proposition and the nature and status of logic’ (ix–x). A feature that she presents as crucial and crucially distinctive of her reading is that Wittgenstein achieves these insights ‘by means of a method that can plausibly be held to be merely clarificatory’ (x), his project being one of ‘clarifying, rather than explaining, the workings of our language’ (9). She does not claim to have invented this ‘third way’, tracing its roots in the work of Ishiguro, McGuinness, Rhees and Winch; but—as she says (x)—she has explored in much greater detail and depth possibilities that these predecessors raise. McGinn’s book is excellent; it’s admirably thorough, works extremely hard at many of the most difficult and demanding of passages in Wittgenstein’s early work and, in doing so, puts together a reading of the Tractatus that many will find persuasive. It is also a complex book, dense and sometimes difficult. It calls for careful study, which I think it will get over the next few years, and what I can offer here are really not much more than first impressions. According to McGinn, Wittgenstein’s ‘central aim’ is to make clear the essence of representation, ‘what is essentially shared by all representations of possible states of affairs’ (99). Her starting point is the claim that the

connection between language and world is ‘internal’ in being one that is not to be ‘discovered’ but is instead ‘grounded in a rule’ (80), in ‘rules of projection in virtue of which we use propositional signs to say how things are in reality’ (81). Wittgenstein’s task is to make these rules of projection ‘perspicuous’; he is not – McGinn insists—in the business of explaining ‘how language’s ability to represent the world came about’ (82, cf. also 122). It is an aspect of the internal relation mentioned above that language is ‘autonomous’ (13): rules of projection, in making possible the comparison of pictures with reality, ‘cannot themselves be represented in a picture that can be compared with reality’ (94) and the possibility of such a comparison does not depend on the existence of any particular state of affairs but solely on ‘the existence of the rules of projection’ (96). A crucial step in rendering the rules in question perspicuous is made by adopting the notion of ‘logical portrayal’. McGinn argues that, for Wittgenstein, ‘the rules of projection that lay down what counts as [a] picture’s being true or false … include the correlation of the pictorial elements’ that make up the picture ‘with objects that are the constituents of the states of affairs that are depicted’ (96); in doing so, she integrates perhaps the best known idea proposed in the tradition in which she sees herself as standing: an opposition to the idea that the correlation between names and objects arises independently—and might provide the basis of an explanation—of the use of names in propositions that are used to make true and false statements (88). Factoring in the further requirement that sense be determinate, the ‘existence’ of the rules of projection comes to be seen as dependent upon ‘the existence of primitive expressions, which stand for logically simple objects that are the simple constituents of states of affairs’ (110); and here we reach one of the crucial points in McGinn’s reading: her claim that the author of the Tractatus did not regard such requirements as carrying metaphysical import. McGinn insists that ‘[i]t is the essential structure of the [representational] sys-

tem itself, and not of what lies outside it, that [Wittgenstein] is investigating’ and that, although the ‘dogmatic’ commitment to simple objects ‘emerges in the course of ’ that investigation, that commitment ought not to be seen—and was not seen by the early Wittgenstein—as ‘a speculative claim about the essential structure of the world conceived of independently of its representation in propositions’ (116, 117, cf. also 120 and 133). Through Wittgenstein’s reflections on variables, showing, formal concepts, generality, the general form of the proposition, and ‘the great achievement of Wittgenstein’s early work’, his ‘mak[ing] clear that logic does not belong to the level of facts’ (173), McGinn argues one can follow her guiding notion, that Wittgenstein’s attempt to clarify what representation as such involves does not aspire to ‘ground’ the possibility of representation in a metaphysics but does yield ‘positive philosophical insights into how language works’ (p. ix). So, for example, once the internal relation between language and world has been made ‘perspicuous’, we will come to recognize that what Wittgenstein is doing in the opening remarks of the Tractatus, which have so often been taken to articulate a metaphysics of some sort, ‘is nothing more than tracing the logical order that is essential to language’s ability to express propositions that can be compared with reality for truth or falsity’ (137). In a review of this length, I cannot hope to do justice to this rich piece of work or present a proper evaluation of it. I suspect that most critical attention will focus on McGinn’s claims that the early Wittgenstein could have seen the ‘clarificatory’ project that she ascribes to him as lacking metaphysical implications and that ‘[w]hat at first appears to be a metaphysics is … nothing but the shadow of the logic of the language in which we represent the world’ (151). McGinn regards the requirement for simple signs as ultimately ‘dogmatic’ (109–10) and acknowledges that a certain profile does seem to be prescribed for objects: the ‘most plausible candidates’ are ‘spatial or material points, colours, temporal points, and so on’ (115). So one question will be whether this apparently dogmatic prescription regarding the character of what we ultimately think and talk about could at the time of writing

the Tractatus either have struck Wittgenstein as metaphysically innocuous or not struck him at all. Since I am broadly sympathetic to the kind of view McGinn is trying to defend here—indeed I made a similar use of the ‘shadow’ metaphor in my own recent book on the Tractatus—I won’t pursue this question; I am sure others will. Instead I wish to focus on another aspect of the ‘clarificatory’ project. Something that seems to be missing in McGinn’s book is a thorough engagement with a question which she herself posed for resolute readers in the 1999 paper in which she first set out the basic ideas of her reading: ‘if the ladder … turns out to be an illusion, how have we got anywhere by climbing it?’ (‘Between Metaphysics and Nonsense: Elucidation in Wittgenstein’s Tractatus’, Philosophical Quarterly 49: 491–513, 496) This question concerning the status of the propositions that make up the Tractatus seems a problem for resolute readers in particular because of their commitment to an ‘austere’ conception of nonsense, according to which ‘[a]nything that is nonsense is so merely because some determination of meaning has not been made’ (Diamond, quoted in 243 n. 4); nonsense is the absence of sense rather than the presence of, as it were, the wrong kind of sense. But McGinn herself endorses this conception (243 n. 4 and also 18, 19, 100, 246 and 270) and, in the discussions that come closest to her 1999 question (158–59, 252– 54), she also seems to endorse the notion that Wittgenstein’s own propositions are nonsensical: she states that the ‘work’ that these ‘perform does not depend upon their possessing a sense, but upon their enabling the reader to see clearly what the use of language makes clear’; ‘[t]his nonsense has indeed been useful’ in ‘serv[ing] to bring about a clarified vision of the logical order that—Wittgenstein believes —is there in language insofar as it represents states of affairs’ (253). But the 1999 question is: how can climbing a ‘ladder’ made up of strings of signs in which ‘nothing has been expressed’ (18) be ‘useful’, ‘bring about’ any such vision or ‘enable’ one to see anything? (This issue is related to the question of the status of the medium in which one produces ‘a description of what is essential to a system

260

for representing possible states of affairs in propositions’ (p. 159), a description which— it seems—is not itself supposed to be a ‘representation’ of a ‘state of affairs’.) More broadly, I wonder whether McGinn gives us a clear picture of the ‘clarificatory’/ ’desciptive’ method that she sees as distinctive of her early Wittgenstein. He may not be ‘attempt[ing] to take up a perspective outside language’ from which one might discern a grounding of the possibility of representation in ‘a transcendent reality’ (147, cf. also 17 and 78–79); but we need more than a specification of what he is not doing. To use the method in question is to engage in a certain kind of ‘description’ (17, 20, 26), one which ‘allow[s] language itself to reveal how it functions’ (252, cf. also 20, 21, 26, 31, 76 and 77); but what does that mean? A reason why I ask that is that it is not clear to me where we actually see the method in question at work. Wittgenstein’s ‘investigation … is entirely internal to language’ (x, 13), this being an appropriate approach because language is ‘autonomous’: ‘Everything essential to language is internal to it and can be made clear by means of description alone’ (27, cf. also 13). But accepting this ‘autonomy’ claim— which is neither transparent nor self-evidently true—is just the kind of thing that one might suppose ought to result from using one’s chosen philosophical method. So which of the central notions of the Tractatus actually can be seen as emerging through the use of what McGinn claims is its distinctive method? McGinn is clear that many cannot; she plausibly proposes that the early Wittgenstein is in the grip of a problematic and idealised conception of language (78, 120); it is at this that he directs his ‘clarificatory’/ ’descriptive’ method and, consequently, his ‘results’ are informed by a variety of distorting ‘dogmatic’ commitments, including the determinacy of sense (12), the notion that all representation has a common essence (12), his conception of facts as complexes (107) and what she calls the ‘framework intuition’ (54) (which, among other things, drives his belief in the logical independence of elementary propositions (142)). Wittgenstein arrives at these notions not through the use of the ‘clarificatory’/’descriptive’ method but—in as

much as they are ‘arrived at’ at all—through the time-honoured philosophical ‘method’ of taking what seems obvious to one as if it was obvious full-stop. But then when do we see Wittgenstein ‘simply … examining language and making clear what language itself reveals about its workings’ (17)? To take two further central examples, it is not clear to me where—according to McGinn—the notions of bipolarity or logical portrayal ‘come from’. It may be that Wittgenstein actually has no particular reason for embracing these ideas or that they represent further ‘dogmatic’ commitments. But if so, this is problematic not simply—to take the case of bipolarity—because McGinn presents this idea as driving some of Wittgenstein’s central criticisms of Frege and Russell (33– 50), or because it is a notion which—since it ‘entails contingency’ (45, cf. also 102)— would all by itself render logical and metaphysical ‘propositions’ impossible, but also —for McGinn’s metaphilosophical picture in particular—because it can’t then be seen as revealing itself through one’s ‘allow[ing] language itself to reveal how it functions’ (252). Similarly, the idea of logical portrayal might be seen as making it possible to understand how ‘[t]he rules of projection that make the comparison between a picture and reality possible do not require that the situation depicted in the picture exists’ (87). But there seems a rather loose fit between the claim that a particular commitment is adopted because it frees one from an unacceptable consequence and the claim that it reveals itself when one ‘simply … examin[es] language’ (17). McGinn’s claim is only that Wittgenstein ‘takes his approach to his task to be one of clarification’ (p. 76, italics added), and his ‘purely descriptive intentions’ are ultimately ‘frustrated’ (14, cf. also 119). But do we have a clear sense of what he is supposedly striving—but failing—to do, of what content ought to be ascribed to his frustrated ‘intentions’, to the ‘descriptive conception of philosophy’ that he ‘betrays’ (119)? What might be expected to put flesh on these bones is McGinn’s insistence on a continuity in method between the early and later work (cf., e.g., 33). But to show that that continuity is substantial—rather than merely rhetorical—one

261

would already need to have a substantial specification of the early method or to see it actually at work in McGinn’s early Wittgenstein; it’s not clear to me where one does see that or that she gives us the necessary specification. One is struck at many points in McGinn’s book by how neatly her story fits problematic passages in Wittgenstein’s early writings: I was particularly struck by her readings of, for example, TLP 3.332 (169) and 4.4661 (203), and her discussion of ‘Notes on Logic’ in Chs. 2 and 3 is the best that I am aware of currently available in print. But much of the debate about the Tractatus in recent years has centred

around the difficulty of squaring its ‘framing’ metaphilosophical remarks with the nuts and bolts of its ‘content’ and my worries—undeveloped as they are—concern how compelling the connection is between the detail of McGinn’s reading and the metaphilosophical story that is meant to subsume that detail. But those worries must not disguise the fact that her book is a very rich piece of work which has much to teach us.

262

Denis McMANUS University of Southampton

Grazer Philosophische Studien 76 (2008), 263–267.

John GIBSON & Wolfgang HUEMER (Hrsg.): Wittgenstein und die Literatur. Frankfurt am Main: Suhrkamp, 2006, STW 1782. 518 S. ISBN 3-518-29382-6. € 17.00. Seit Platon hat sich die Philosophie nicht zuletzt auch darüber definiert, nicht Literatur zu sein, auch wenn sie sich häufig literarischer Formen, wie Dialogen, Aphorismen oder Fragmenten bediente. Dabei lassen sich, vereinfacht gesagt, zwei Strategien der philosophischen Abgrenzung zur Literatur unterscheiden. Zum einen wird die Literatur in den Dienst der Philosophie genommen: Ist letztere an Ideen, der Wahrheit, den fundamentalen Strukturen von Welt und Erkenntnis interessiert, oder an der Weise, wie Sprache unser Denken bestimmt, wurde und wird von der Literatur erwartet, jene von den Philosophen entdeckten Wahrheiten über die Wirklichkeit oder Sprache eine ästhetischanschauliche Form zu geben. In der Folge wird Literaturtheorie bzw. das Studium der Literatur mit Rhetorik gleichgesetzt, also der Analyse, wie sich die literarische Sprache am besten einsetzen läßt, um philosophische Ideen möglichst adäquat darzustellen. Die zweite Abgrenzungsstrategie sprach der Literatur aufgrund ihrer Fiktionalität die Fähigkeit zur Darstellung der Wahrheit ab oder unterstellte ihr sogar, ihre Rezepienten zu täuschen und deshalb eine moralische und politische Gefahr darzustellen. Man denke an Platons Abwertung der Dichtung, weil sie Trugbilder produziere oder David Humes Bemerkung, Dichter seien „liars by profession“ (A Treatise On Human Nature, B.1.3.10) eine Einschätzung, die noch, wenn auch aus sprachtheoretischen Gründen, in Gottlob Freges These, literarische Sätze drückten nur „Scheingedanken“ aus, oder in der Bemerkung des Literaturnobelpreisträgers Bertrand Russell über Hamlet: „the propositions in the play are false because there was no such man“ (zit. in Wittenstein und die Literatur, 14) widerhallen. Beide erwähnten Abgrenzungsstrategien beruhen auf einer Erklärung der Sprache auf der Basis von Referenz und Wahrheit, was im 20. Jahrhundert eine Reihe von Ansät-

zen motiviert hat, die das Fiktionale des Literarischen referenziell bzw. als Grenzfall oder Abart der Referenzialität deuten, z.B. als wahrheitsgetreue Darstellung bloß möglicher Welten oder als Bezug auf Meinongsche Gegenstände, die es geben soll, ohne daß sie existieren, oder als die Prätention, Sprache in ihrer gewöhnlichen, weltbezogenen Form zu gebrauchen. Die literarische Sprache wird damit zu einer von der Alltagssprache deutlich getrennten Sprache – Worte in einem Roman z.B. würden etwas anderes bedeuten und bezeichnen als die gleichen Worte der Alltagssprache. Eine klare Trennung der literarischen Sprache von der Alltagssprache läßt sich jedoch nicht nur kaum begründen; sie wird auch der Relevanz nicht gerecht, die die Literatur für unser Verständnis des Alltäglichen und der Alltagssprache offensichtlich hat. Zudem wurde das referenzielle Sprachmodell von philosophisch-sprachwissenschaftlicher Seite - Wittgenstein, Austin, Quine, Derrida, Saussure, Jakobson u. a. – einer umfassenden Kritik unterzogen. Die Literaturtheorie des 20. Jahrhunderts hat sich in ihren wichtigsten Strömungen der philosophischen Abwertung und Vereinnahmung der Literatur und der Allgemeingültigkeit des referenziellen Modells widersetzt. Dabei sind im vorliegenden Zusammenhang vor allem zwei Bestimmungen der Literatur oder des Literarischen relevant: Zum einen wird Literatur als ein eigener Bereich behandelt, dessen Eigenständigkeit in seiner Selbstbezüglichkeit, und seiner Autonomie gegenüber Autorenintention oder sonstiger dem Text äußerlichen Faktoren und Beziehungen besteht (russischer Formalismus, New Criticism, Strukturalismus). Zum anderen wird die Grenze zwischen dem Literarischen und dem Philosophischen suspendiert oder verschoben und die Wirkung des Literarisch-Fiktionalen im Philosophischen und umgekehrt betont, wie etwa im Dekonstruktivismus, aber auch bei Ludwig Wittgenstein. In den Philosophischen Untersuchungen finden sich zahlreiche absurde, paradoxe und sonstige nicht-alltägliche und fiktive Beispiele zur Erläuterung unserer alltäglichen Praxis

und philosophischen Begriffe. In den dreißiger Jahren notiert Wittgenstein „Ich glaube meine Stellung zur Philosophie dadurch zusammengefaßt zu haben, indem ich sage: Philosophie dürfte man eigentlich nur dichten.“ (Vermischte Bemerkungen (VB), 483) In einem Brief an seinen Verleger Ludwig von Ficker schreibt er über den Tractatus, er sei „streng philosophisch und zugleich literarisch“ (Ludwig Wittgenstein, Briefe an Ludwig von Ficker, Salzburg 1969, 33) Später betont er „die seltsame Ähnlichkeit einer philosophischen Untersuchung … mit einer ästhetischen“ (VB, 485) und verdeutlicht ein philosophisches Problem, nämlich die Frage der Relevanz von Bedeutungserlebnissen für das Verständnis von Sätzen, mit einer Theaterszene: „Die Zusammenhänge, in denen ein Satz steht, sind am besten in einem Drama dargestellt, daher das beste Beispiel für einen Satz in einer bestimmten Bedeutung ein Zitat aus einem Drama ist. Und wer fragt die Person im Drama, was sie während des Sprechens erlebt?“ (Letzte Schriften über die Philosophie der Psychologie, Bd. 1, § 38) In dieser Suspendierung der Entgegensetzung von Philosophie und dem Literarisch-Ästhetischen liegt vielleicht einer der Gründe für die Faszination, die Wittgenstein wie kaum ein zweiter Philosoph bis heute auf bildende Künstler, Dichter, Schriftsteller, Komponisten und Filmemacher ausübt. Wittgenstein rückt außerdem die Sprache ins Zentrum, eine Sprache, die nur im und durch ihren sozialen Gebrauch, durch ihre Verkörperung in Handlungen und nicht durch ihre Abstraktionsleistung oder ihre Referenzialität Bedeutung hat. Ein weiterer Grund für Wittgensteins Einfluß auf die Kunst mag die literarische Form seines Schreibens, seine Vorliebe für das Skizzenhafte und Fragmentarische sein. Auch für die Literaturtheorie ist Wittgenstein aus mehreren Gründen interessant. Zunächst machte er sich offensichtlich Gedanken über den Status von Literatur, wenn er z.B. schreibt, daß Dichter und Musiker „etwas zu lehren haben (VB, 501), oder „daß ein Gedicht, wenn auch in der Sprache der Mitteilung abgefaßt, nicht im Sprachspiel der Mitteilung verwendet wird“ (Zettel, § 160). Aber auch Wittgensteins Kritik bedeutungsstiftender Intentionen, sein Pri-

vatsprachenargument, seine Überlegungen zur Rolle des Interpretierens beim Verstehen von Regeln und Sprache, sein Begriff der Familienähnlichkeit, seine Weigerung, die Diskussion der Sprache von einer Diskussion der Welt und umgekehrt zu trennen, und die enge Bindung des literarischen Stils seiner Texte an ihren Inhalt, können einen Beitrag zur Theorie der Literatur leisten. Der Band Wittgenstein und die Literatur ist in diesem Problemfeld angesiedelt. Seine 19 Beiträge diskutieren das Verhältnis von Philosophie und Literatur bei Wittgenstein, die Frage, welche Aspekte seiner Philosophie für literaturtheoretische Probleme relevant sind – Autorenintention, Textinterpretation, Form-Inhalt, Fakten und Fiktion, Autobiographie, das Unsagbare – und wie ein Lesen literarischer Texte mit Wittgenstein aussehen könnte. Bis auf die Beiträge von Stanley Cavell, Cora Diamond (jeweils mit neuer Einleitung) und Bernard Harrison, handelt es sich um Originalarbeiten. Wittgenstein und die Literatur umfaßt neben einer ausführlichen Einleitung von Wolfgang Huemer 5 Teile: 1. „Philosophie als Literatur, Literatur als Philosophie“, 2. „Lesen mit Wittgenstein“, 3. „Literatur und die Grenzen von Selbst und Sinn“, 4. „Fiktion und der „TRACTATUS“ und 5. „Eine weitere Perspektive“. Die 6 Beiträge des 1. Teils gehen vor allem dem Literarischen in Wittgensteins Philosophie nach. Stanley Cavell liest in „die Alltagsästhetik der Philosophischen Untersuchungen“ Wittgenstein als einen Autor der Moderne, da es sich bei den Philosophischen Untersuchungen um ein Werk handele, das „ständig sein Medium und sein Gefühl eines Bruchs mit der Vergangenheit befragt“ (37), weshalb Wittgensteins philosophische Methode der „übersichtlichen Darstellung“ auch seine Schreibweise, den Text selbst, kennzeichne. Ein weiteres modernistisches Element sieht Cavell in Wittgensteins Auffassung des modernen Subjekts als für sich selbst undurchsichtig und unübersichtlich. Auch David Schalkwyk situiert in seinem Beitrag „Wittgensteins „unvollkommener Garten“. Die Leitern und Labyrinthe von Philosophie als Dichtung“, Wittgenstein im Kontext des Modernismus. Schalkwyk setzt Wittgensteins unstetes Leben in Beziehung zu seinem skizzenhaften und unru-

264

higen, immer wieder auf dieselben Fragen zurückkommenden Schreib- und Denkstil und arbeitet die räumlichen Metaphern (Leitern, Gärten, Labyrinthe, Wildnis, verwinkelte Städte usw.) heraus, die das Unheimliche und Fremde im Alltäglichen und die „endlose Wiederkehr der unvertrauten Aspekte, in dem was Wittgenstein „Sprache“ nennt“ (97) ausdrücken, der ja ein Weltbezug immanent ist. Zugleich versteht Schalkwyk Wittgensteins Suche nach Übersichtlichkeit und dem „erlösenden Wort“ als den immer wieder scheiternden Versuch, endlich zur Ruhe zu kommen. Schalkwyk sieht in Freuds Arbeit „Das Unheimliche“ (1919) eine treffende Erläuterung der Situation Wittgensteins und der Doppelnatur des Alltäglichen und Selbstverständlichen. Freud zeigt, wie das Unheimliche und Unverstandene des Alltäglich-Vertrauten aus der rätselhaften Wiederkehr „derselben Dinge, Vorgänge und Situationen“ (Freud) entsteht. Das Nicht-Alltägliche ist jene andere Seite des Alltäglichen und Selbstverständlichen, die Wittgenstein durch das Literarische, d. h. die Erfindung fiktiver Situationen und Begriffe aufzuklären sucht, was er z.B. wie folgt ausdrückt: „Nichts ist doch wichtiger, als die Bildung von fiktiven Begriffen, die uns die unseren erst verstehen lehren“ (VB, 555). Marjorie Perloff interpretiert Wittgensteins Bemerkung, Philosophie könne man nur dichten, mit Hilfe der Frage der Übersetzbarkeit von Dichtung. Sie zeigt, daß sich Wittgensteins Texte bedeutend leichter in andere Sprachen übersetzen lassen als etwa die Gedichte Rilkes, was gegen ihren literarischen Charakter zu sprechen scheint. Ähnliches gilt aber auch für die Prosa Becketts, die experimentelle Poesie und Spielarten der konzeptuellen Kunst, weil sie, so Perloff, ähnlich wie Wittgenstein die Sprache als solche und nicht einzelne Sprachen thematisieren, wobei das Dichterische bei Wittgenstein und den erwähnten Kunstformen im Element des Erfinderischen liege. Während sich Timothy Gould der therapeutischen Seite der Philosophischen Untersuchungen und ihrer Suche nach der Ruhe vor philosophischen Fragen widmet, untersuchen Bernard Harrison und John Gibson den Weltbezug von Literatur bzw. die Frage, ob und

wie Literatur unser Wissen über die Welt und uns selbst erweitern kann. Beide Autoren zeigen in ihren Beiträgen, daß die Ablehnung des mimetisch-referenziellen Sprachmodells für die Literatur durchaus damit vereinbar ist, daß Literatur wichtiges über die Welt zu sagen hat, obwohl wir uns als Leserinnen und Leser der Fiktionalität des Gelesenen bewußt sind. Literatur leistet dies, da sie, so Harrison, unsere sprachliche Praktiken, denen ein Weltbezug immanent ist, thematisiert bzw., so Gibson, uns, ähnlich wie das von Wittgenstein diskutierte Pariser Urmeter, Standards und Kriterien zur Beurteilung unseres Lebens liefert. Die Beiträge des 2. Teils von Wittgenstein und die Literatur befassen sich mit Fragen des Lesens von Literatur und der Natur der Textinterpretation. Ausgehend von Wittgensteins Unterscheidung im Tractatus zwischen Sagen und Zeigen greift Cora Diamond eine Idee Wolfgang Isers auf, der das (philosophische) Interesse eines literarischen Textes weniger in den Ideen sieht, die er darstellt, als in dem, was der Text nicht thematisiert. In diesem Sinne interpretiert Diamond Wittgensteins bekannte Stelle aus einem Brief an Ludwig von Ficker, der Tractatus bestehe aus zwei Teilen, „aus dem, der hier vorliegt, und aus alledem, was ich nicht geschrieben habe“ (189), eine Deutung, die Diamond auch auf die Philosophischen Untersuchungen ausdehnt, deren literarischer Charakter sie demnach darin sieht, ein Text zu sein, „dessen Lektüre durch Aufmerksamkeit auf das, was ihm fehlt“, geleitet wird, etwa gewöhnliche philosophische Argumente oder allgemeine Antworten auf die im Text aufgeworfenen Fragen. Der Rezeptionsästhetiker Iser und ihm folgend Diamond sehen also die Lücken eines Textes und das Fehlen definitiver Antworten als Aufforderung oder Chance, diese Lücken im Leseprozeß selber auszufüllen. Das im Text ausgesparte wäre somit immer noch sagbar. Fraglich scheint mir jedoch, ob sich darin Wittgensteins Behandlung des „Unaussprechbaren“ als Bedingung oder „Hintergrund“ (VB, 472) des Sagbaren erschöpft. Das Unaussprechbare des Textes ist bei Wittgenstein nicht nur das in ihm Nicht-Thematisierte, das „durch Abwesenheit glänzen“ würde, aber dennoch, je nach Rezeptionskon-

265

text, immer wieder neu identifiziert werden könnte, sondern das, von dessen Nicht-Thematisierbarkeit die Bedeutung des Textes abhängt und das sich im wesentlichen entzieht, wie im Tractatus die logische Form der Wirklichkeit oder die komplexen Kontexte der Lebensformen im späteren Werk. Joachim Schulte geht in seinem Beitrag „Das Leben der Zeichen“ der Frage nach, was es im Rahmen der Philosophie Wittgensteins heißen könnte, ein Gedicht zu verstehen. Schulte stellt eine erhellende Verbindung zwischen Wittgensteins Entgegensetzung „toter“ und „lebendiger“ Zeichen und dem Sehen von Aspekten und Aspektwechseln her. Was Zeichen „beseelt“ und „belebt“, sind keine geistigen Akte des Meinens oder Verstehens, sondern ihr Gebrauch. Texte werden als Gedichte gebraucht, wenn sie nicht als Informationen oder Mitteilungen gelesen werden, sondern dem Leser in einer Art Aspektwechsel neue Bedeutungsebenen oder auch die Materialität des Gedichts – sein Klang, Rhythmus, Reim, seine „Melodie“ – erschließen. Schulte stellt auch eine interessante Verbindung zwischen dem Aspektsehen und der Unterscheidung zwischen dem, was sich sagen läßt, und dem Unaussprechlichen her, das nach Wittgenstein in einem Gedicht enthalten ist. Nach Schulte wäre es „sinnlos“, von einem Unaussprechlichen, das nicht „irgendwie“ zu erfassen ist, zu sprechen, weshalb er es mit einem Aspekt gleichsetzt, der beim Lesen je nach Sensibilität erlebt werden kann. Die radikalere Möglichkeit jedoch, daß das Unaussprechliche sich bei Wittgenstein jeder Form des Verstehens und Erlebens entzieht, weil es die Widerständigkeit des Gedichtes gegen ein endgültiges Verstehen meint, weil es also nichts Abschließendes zu verstehen oder zu erleben gibt, zieht Schulte ähnlich wie Diamond nicht in Betracht. Ausgehend von Wittgensteins Bemerkungen zum Regelfolgen, kritisieren Sonia Sedivy und Martin Stone die in der Literaturtheorie weit verbreitete Auffassung, jede Lektüre sei eine Interpretation, und die damit oft einhergehenden These, daß dem Spiel alternativer Deutungen eines Textes keine Grenze gesetzt ist. Beide Autoren wählen als Ausgangspunkt einen der einflußreichsten zeitgenössischen Literaturtheoretiker, Stanley Fish, der in sei-

nem Klassiker „Is There a Text in This Class?“ (1980) unter Berufung auf Wittgensteins Begriff der Lebensformen argumentiert, Texte seien an sich bedeutungslos und würden eine bestimmte Bedeutung erst im Akt des Interpretierens erhalten, und zwar in Abhängigkeit von den kontextabhängigen Voraussetzungen und „Situationserwartungen“ des Lesers, wobei der prinzipiell unbegrenzte Spielraum von Deutungsmöglichkeiten durch die jeweiligen Lebensformen oder „Interpretationsgemeinschaften“ eingegrenzt wird. Sedivy und Stone halten Fish, aber auch den Dekonstruktivisten Paul de Man und Hillis Miller, Wittgensteins These entgegen, eine Regel oder ein Text sei nur auf der Basis einer Bedeutung verschieden interpretierbar, die der Text auch ohne Interpretation schon hat, und die uns unmittelbar gegenwärtig ist. Allerdings setzt Wittgenstein, im Unterschied zu Sedivy und Stone die „Auffassung einer Regel, die nicht eine Deutung ist (PU, § 201) gerade nicht mit dem Erfassen einer einfachen Bedeutung gleich, sondern mit dem erlernten „Handeln nach der Regel“, so daß es nicht verwundert, daß Stone einige Mühe hat, der einfachen, vor-interpretiven Bedeutung eine nicht-metaphysische Interpretation zu geben. Die Beiträge des dritten Teils diskutieren Wittgensteinsche Themen im Zusammenhang mit der Fragen des Selbst, des autobiographischen Bewußtseins, des Sprachskeptizismus und der Psychopathologie. Richard Eldridge setzt die Poetologie Friedrich Hölderlins, Kants und Fichtes mit den Philosophischen Untersuchungen in Verbindung, während Garry L. Hagberg Wittgensteins Stellung zum Solipsismus für das Genre der Autobiographie fruchtbar macht. James Guetti vergleicht Joseph Conrads Sprachauffassung in Heart of Darkness mit Wittgensteins Bild einer leerlaufenden Sprache und versteht den zunehmenden Wahnsinn von Kurtz als ein Abtriften in den Solipsismus. Rupert Reads Beitrag knüpft an Guetti an, allerdings an dessen aus einer anderen Arbeit stammende Interpretation der sprachlosen Figur des Benjy in William Faulkners The Sound and the Fury. Read vertritt eine Art wittgensteinsche Theorie der Psychopathologie, wonach wir schwere Fälle z.B. von Schizophrenie nicht verstehen können, da die Sprache der Betroffenen solipsistisch ist, d. h.,

266

„sich selbst genügt“ und von jedem nachvollziehbaren Handlungskontext und Weltbezug abgetrennt ist. Alex Burri und Dale Jacquette behandeln im 4. Teil fiktionale Ausdrücke im Rahmen der von Wittgenstein später verworfenen referenziellen Sprachtheorie des Tractatus und die Frage, auf welche Weise Literatur in diesem Rahmen „kognitiven Wert“ (Burri) für unser Leben haben kann. Beide Beiträge zeigen gewollt oder ungewollt, wie unbefriedigend eine Literaturanalyse ist, wenn sie das referenzielle Sprachmodell als Ausgangspunkt wählt. So wird in diesem Modell etwa die Unterbestimmtheit fiktionaler Ausdrücke, bzw. der von ihnen bezeichneten Gegenstände zu einem Problem, also etwa Details, über die uns ein Text – Burri wählt Conan Doyles Sherlock Holmes Geschichten – im Unklaren läßt, z.B. die Frage, welche Farbe Holmes’ Haus in der Baker Street, oder welche Schuhgröße Watson hat. Im letzten Beitrag von Wittgenstein und die Literatur bestreitet Joseph Margolis, daß Wittgenstein eine allgemeine Methode hat,

die man auf Fragen der Ästhetik und Literatur anwenden könnte, wie es etwa Garry Hagberg in Art as Language, Margolis’ Hauptangriffsziel, versucht hat. Wittgenstein könne höchstens Inspirationen zu einer eigenen ästhetischen Theorie liefern, gerade auch dadurch, daß sein Denken zu einer Position anregt, die sich ausdrücklich in Gegensatz zu ihm setzt, wie z.B. im Fall der Kunsttheorie von Arthur C. Danto, der Sprache und Denken bzw. Sprache und Wahrnehmung voneinander trennt, gegen Wittgensteins Behauptung ihrer Einheit. Wittgenstein und die Literatur demonstriert eindrucksvoll das Potential, das in Wittgensteins Philosophie immer noch steckt, auch wenn es um Fragen geht, die Wittgenstein eher beiläufig behandelt hat. Die Beiträge zeigen aber auch, daß das Literarisch-Fiktive in Wittgensteins Philosophie eine wichtigere Rolle spielt als die dominierende angelsächsische Wittgensteininterpretation nahelegt.

267

Klaus PUHL Universität Graz

Grazer Philosophische Studien 76 (2008), 268–274.

Michael STÖLTZNER & Thomas UEBEL (Hrsg.): Wiener Kreis. Texte zur wissenschaftlichen Weltauffassung von Rudolf Carnap, Otto Neurath, Moritz Schlick, Philipp Frank, Hans Hahn, Karl Menger, Edgar Zilsel und Gustav Bergmann. Hamburg: Meiner 2006, Philosophische Bibliothek Bd 577. civ + 699 pp. ISBN 978-3-78731811-7. € 78.00. 1. Der Band Wiener Kreis. Texte zur wissenschaftlichen Weltauffassung (im Folgenden WK) ist für die „Philosophische Bibliothek“ des Meiner Verlages ein Novum: WK ist der erste unter den etwa 350 Bänden dieser Reihe, der nicht das Werk eines einzelnen Autors ist, sondern eine Sammlung von achtundzwanzig Artikeln, die von nicht weniger als acht Autoren stammen. Mit anderen Worten, es ist ein kollektives Werk, durchaus im Geiste des Wiener Kreises, der im so genannten Manifest und auch anderswo, etwa im Vorwort zu Carnaps Aufbau, die philosophische Arbeit seiner Mitglieder als ein kollektives Unternehmen zu beschreiben pflegte. Es wäre ein Mißverständnis, philosophische Kollektivarbeit im Sinne des Kreises als Ausarbeitung einer monolithischen Schuldoktrin zu verstehen. Der Wiener Kreis war eine keineswegs homogene Bewegung eigenständiger Denker, die sich zum gemeinsamen Projekt einer wissenschaftlichen Weltauffassung zusammengefunden hatten, ohne dafür ihre philosophische Eigenart aufzugeben (vgl. WK, ix). Die Nachwelt hat diesen kollektiven Charakter der Philosophie des Wiener Kreises lange Zeit kaum zur Kenntnis genommen. Stattdessen wurde, salopp ausgedrückt, der logische Empirismus des Wiener Kreises rezipiert als „Carnaps Wiener Philosophie nebst einigen Fußnoten anderer Autoren“. Die philosophiehistorische Forschung der letzten beiden Jahrzehnte hat gezeigt, daß dieses Bild der Philosophie des Wiener Kreises unzutreffend ist. WK kann als ein Resultat dieses philosophiehistorischen Fortschritts betrachtet werden. Die hier versammelten Arbeiten geben ein gutes, wenn auch nicht vollständiges Bild von der Vielfalt und dem Reichtum der

philosophischen Arbeit, die im Wiener Kreis geleistet worden ist. Zunächst eine kurze quantitative Beschreibung des Inhaltes von WK, um den potentiellen Leser darüber zu informieren, was ihn in diesem opulenten Buch von etwa 650 Seiten Text, 100 Seiten Einleitung und 50 Seiten Anmerkungen erwartet. WK versammelt insgesamt 28 Texte von Carnap, Neurath, Schlick, Frank, Hahn, Menger, Zilsel und Bergmann, die zwischen 1913 und 1938/1939 entstanden sind. Hier ist anzumerken, daß Bergmanns Beitrag Rückblick aus der Emigration nicht, wie von den Herausgebern in der Einleitung behauptet, aus dem Jahre 1936 stammt, sondern erst Ende 1938 oder Anfang 1939 entstanden ist. Die genannten acht Autoren sind in WK mit der folgenden Anzahl von Beiträgen vertreten: Schlick (7), Neurath (7), Carnap (5), Frank (3), Hahn (2), Zilsel (1), Menger (1), und Bergmann (1). Dazu kommt als wirklich kollektives Werk noch das so genannte Manifest des Wiener Kreises, das Neurath, Carnap und Hahn, wohl unter der Federführung Neuraths, gemeinsam verfaßt haben. Die 28 Originaltexte werden in sieben Kapiteln präsentiert: I. II. III. IV. V. VI. VII.

Programmschriften (2) Frühe philosophische Arbeiten der Gründer (3) Allgemeine Erkenntnislehre und Wissenschaftslehre (5) Zu den Programmen des Physikalismus und der Einheitswissenschaft (5) Protokollsatzdebatte (5) Zu Spezialproblemen einzelner Wissenschaften (7) Rückblick aus der Emigration (1)

Die meisten der in WK versammelten Arbeiten sind für jeden, der sich ernsthaft mit der Philosophie des Wiener Kreises befassen will, Pflichtlektüre: Das Manifest, Pseudorationalismus der Falsifikation, Über das Fundament der Erkenntnis, oder Die physikalische Sprache als Universalsprache der Wissenschaft, um nur einige Texte zu nennen, gehören zu den

zentralen Dokumenten des logischen Empirismus des Wiener Kreises. Bei einigen anderen Beiträgen bin ich mir nicht ganz sicher, ob sie dazu zählen, etwa bei Zilsels P. Jordans Versuch, den Vitalismus quantenmechanisch zu retten oder Schlicks Über den Begriff der Ganzheit, auch wenn die Herausgeber sich bemühen, auch für die Aufnahme dieser Texte eine schlüssige Begründung zu geben. Eine Rezension ist nicht der Ort, jeden dieser Texte im Einzelnen zu kommentieren. Das leistet bereits in exzellenter Weise die ausführliche Einleitung der Herausgeber, die hier nicht noch einmal in verkürzter Fassung referiert werden soll. Schon aus Platzgründen kann ich im Folgenden nur auf einige der in WK präsentierten Texte eingehen, ohne daß damit eine Abwertung der nicht erwähnten Arbeiten impliziert sein soll. Die in WK versammelten Texte sind in einem philosophischen und politischen Kontext geschrieben, zu dem wir heute keinen unmittelbaren Zugang mehr haben. „Metaphysikfreie Wissenschaft“, „wissenschaftliche Weltauffassung“, oder „Protokollsätze“ sind Themen, die mit heutigen wissenschaftsphilosophischen und wissenschaftspolitischen Debatten nur noch indirekt verbunden sind. Um dem Leser hier über eventuelle Verständnisschwierigkeiten hinwegzuhelfen, ist die von den Herausgebern Michael Stöltzner und Thomas Uebel verfaßte Einleitung von WK ein sehr nützliches Hilfsmittel. Sie kann als eine hervorragende, für sich lesbare Einführung in die Geschichte und Entwicklung des logischen Empirismus des Wiener Kreises angesehen werden. Als Programmschrift des Wiener Kreises gilt gemeinhin die von Neurath, Carnap und Hahn gemeinsam verfaßte Schrift Wissenschaftliche Weltauffassung. Der Wiener Kreis, das so genannte „Manifest des Wiener Kreises“, welches 1929 anläßlich einer in Prag veranstalteten Tagung für „Erkenntnislehre der exakten Wissenschaften“ vom Vorstand des „Vereins Ernst Mach“ herausgegeben wurde. Moritz Schlick, dem diese Schrift gewidmet war, zeigte sich von ihr wenig angetan: Sie war ihm zu „links“, zu propagandistisch, und sie trug allzu deutlich die Züge Neuraths, seines Antipoden im Kreis. Für Schlick als Repräsentanten des rechten Flügels des Kreises

war es deshalb wohl naheliegend wenig später (1930/31) in der gerade gegründeten Zeitschrift Erkenntnis seine eigenen, vom Manifest durchaus verschiedenen programmatischen Vorstellungen darüber zu präsentieren, was unter wissenschaftlicher Philosophie zu verstehen sei. Folgt man Schlicks, von Wittgensteins Philosophieverständnis inspirierter Wende der Philosophie, war die neue Philosophie des Wiener Kreises wesentlich eine Tätigkeit, durch welche der Sinn von Sätzen geklärt werden sollte (WK, 34). Hingegen war bei ihm nicht die Rede von einer Durchdringung der „Formen des öffentlichen und persönlichen Lebens, des Unterrichts, der Erziehung, der Baukunst durch den Geist der wissenschaftlichen Weltauffassung“, wovon das Manifest kündete (vgl. WK, 27). Schon durch die Gegenüberstellung dieser beiden Programmschriften gewinnt man einen ersten Eindruck von der Vielfalt der im Kreis koexistierenden philosophischen und politischen Einstellungen. Die Herausgeber folgen in ihrer Darstellung der Entwicklung des Wiener Kreises einer Linie, die wohl zuerst Neurath und Frank propagierten und die später von Haller aufgegriffen wurde. Danach war das, was man gemeinhin als Wiener Kreis bezeichnet, bereits präfiguriert im so genannten „ersten Wiener Kreis“, der schon vor dem ersten Weltkrieg existierte und dessen Protagonisten Neurath, Frank, Hahn und von Mises waren. Der spätere „Wiener Kreis“ um Schlick war eine Fortsetzung dieses ersten Kreises, oder, wie es die Herausgeber ein wenig umständlich formulieren: „Der 1924 etablierte donnerstägliche Kreis um den 1922 nach Wien berufenen, promovierten Physiker und nunmehrigen Professor für Philosophie der induktiven Wissenschaften Moritz Schlick ging … auf intellektuelle Netzwerke zurück, die bis in die Jahre1907-1912 zurückreichten“ (WK, xi). Diese These von der Existenz eines ersten Wiener Kreises, dessen Struktur und Entwicklung Thomas Uebel vor einiger Zeit in Vernunftkritik und Wissenschaft. Otto Neurath und der erste Wiener Kreis (Uebel 2000) im Detail untersucht hat, schlägt sich auch in der Auswahl der in WK enthaltenen Beiträge nieder. Kapitel II ist nämlich den „frühen philosophischen Arbeiten der Gründer“ gewidmet.

269

Es enthält Franks Die Bedeutung der physikalischen Erkenntnistheorie Machs für das Geistesleben der Gegenwart (1917), Neuraths Die Verirrten des Cartesius und das Auxiliarmotiv (1913), und außerdem, wohl als Hommage an den einzigen nichtösterreichischen „Gründer“ Schlick, dessen Arbeit Die philosophische Bedeutung der Relativitätstheorie (1915), die bis heute als eine hervorragende philosophische Einführung in die Grundgedanken der Relativitätstheorie gelten kann. Franks Die Bedeutung der physikalischen Erkenntnistheorie Machs halte ich für eine seiner besten philosophischen Arbeiten. Im Gegensatz zu manchen späteren verliert sie sich nicht in Invektiven gegen die „Schulphilosophie“ und andere philosophische Gegner. Frank selbst scheint von der Bedeutung dieser Arbeit überzeugt gewesen zu sein, hat er sie doch in den Jahrzehnten nach 1917 in leicht veränderter Form noch mindestens dreimal veröffentlicht. Sein Hauptanliegen war, die empiristische Wissenschaftsphilosophie Machs als Paradigma einer reflexiven Aufklärungsphilosophie zu interpretieren: Danach sind die theoretischen Begriffe der Physik (ebenso wie die der anderen Wissenschaften) immer nur Hilfsbegriffe, die zwar zunächst der Wissenschaft nützliche Dienste leisten, die aber, wenn man sie als endgültige Beschreibung einer feststehenden Wirklichkeit nimmt, den weiteren Fortschritt hemmen und deshalb aufgegeben und durch neue ersetzt werden müssen, die wieder das nämliche Schicksal erwartet. Das führt zu einer jeder Aufklärungsphilosophie inhärenten „tragischen“ Dialektik: „Sie zertrümmert die alten Begriffsgebäude, aber indem sie ein neues errichtet, legt sie schon den Grund zu einem neuen Mißbrauch. Denn es gibt keine Theorie ohne Hilfsbegriffe und jeder Hilfsbegriff wird notwendig mit der Zeit mißbraucht.“ (WK, 112). Auch Neuraths Die Verirrten des Cartesius und das Auxiliarmotiv ist eine sehr originelle Arbeit. Hier findet sich zum ersten Mal der Begriff des „Pseudorationalismus“, der für Neuraths spätere enzyklopädistische Konzeption wichtig werden sollte. Ähnlich wie bereits Peirce in seinem grundlegenden Aufsatz Die Festlegung einer Überzeugung (1877) unterschied Neurath vier Methoden der Ent-

scheidungsfindung bei unvollständiger Information: Die Methode des Instinktes, der Autorität, des Pseudorationalismus, und des Auxiliarmotivs (cf. WK, 128). Ein Anhänger des Pseudorationalismus wird charakterisiert als jemand, der auch im Falle unvollständiger Information so tut, als könnte er eindeutige und rational vollständig gerechtfertige Entscheidungen fällen. Neurath hingegen plädierte für einen aufgeklärten Rationalismus („mit Auxiliarmotiv“), der sich seiner Grenzen bewußt ist und deshalb gegebenfalls, d.h. wenn eine eindeutige wohlbegründete Entscheidung mit rationalen Gründen nicht zu haben ist, auf Auxiliarmotive, z.B. auf das Losen oder andere Zufallsmethoden zurückgreift, um eine Entscheidung herbeizuführen. Diese drei frühen Texte von Schlick, Frank und Neurath gehören meiner Meinung nach zu den bedeutendsten wissenschaftsphilosophischen Arbeiten, die im Umkreis des Logischen Empirismus in den ersten Jahrzehnten des 20. Jahrhunderts enstanden sind. Sie verdienen es ohne Zweifel, wieder allgemein zugänglich gemacht zu werden. Ihre Aufnahme in eine Anthologie von Arbeiten der wissenschaftlichen Weltauffassung des Wiener Kreises verweist aber zugleich darauf, wie schwierig es ist, die Philosophie des Wiener Kreises abzugrenzen. Um 1915 nahm Schlick noch keineswegs eine logisch-empiristische Position im Sinne der Philosophie des Wiener Kreises ein und auch Franks und Neuraths Arbeiten hatten relativ wenig mit logischem Empirismus zu tun. In der Dekade nach dem Ende des Krieges scheinen Neurath und Frank zunächst nichts Philosophisches mehr veröffentlicht zu haben, sondern sich, was Publikationen anging, auf ihr Fachgebiet, die Ökonomie beziehungsweise die Physik, beschränkt zu haben. Schlick hingegen erlebte in dieser Zeit wohl seine produktivste Periode, die aber in WK nicht dokumentiert wird. Erst mit Beginn der „öffentlichen Phase“ des Wiener Kreises und der Gründung des Vereins Ernst Mach 1929 engagierten sich die „Gründer“ Frank und Neurath publizistisch wieder stärker in der Philosophie. Kapitel III Allgemeine Erkenntnislehre und Wissenschaftstheorie versammelt fünf Aufsätze

270

von Frank, Schlick, Hahn und Carnap, die die Erkenntnis- und Wissenschaftsphilosophie der „wissenschaftlichen Weltauffassung“ vom Rest der Philosophie in deutlicher, manchmal polemischer Form abgrenzen. Den Anfang macht Frank mit dem Beitrag Was bedeuten die gegenwärtigen physikalischen Theorien für die allgemeine Erkenntnislehre?, der ursprünglich den Eröffnungsvortrag auf dem Deutschen Physiker- und Mathematikertag 1929 in Prag bildete. In diesem Text bringt er die neue „wissenschaftliche Weltauffassung“ gegen die überkommene „Schulphilosophie“ in Stellung. Letztere ist gekennzeichnet durch eine naive Abbildtheorie der Wahrheit und eine Fixierung auf die veraltete aristotelische Logik, die „wissenschaftliche Weltauffassung“ hingegen wird präsentiert als eine Synthese, die das Beste aus Machs Sensualismus, Poincarés Konventionalismus, James’ Pragmatismus und Russells Logizismus vereint. Wie diese durchaus heterogenen Komponenten wirklich zusammengebracht werden können, bleibt allerdings unklar. Überdies vermeidet es Frank tunlichst, irgendeinen „Schulphilosophen“ (bis auf du Bois-Reymond, der als Sündenbock herhalten muß) beim Namen zu nennen. Insofern macht dieser Aufsatz ein wenig den Eindruck eines philosophischen Schaukampfes. Wie schnell die philosophische Entwicklung im Kreis voranschritt, oder wie verschieden die Standpunkte selbst von einander so nahestehenden Denkern wie Frank und Carnap waren, belegt Carnaps Beitrag Von der Erkenntnistheorie zur Wissenschaftslogik, der einem auf dem ersten Internationalen Kongress für Einheitswissenschaft 1935 in Paris gehaltenen Vortrag entstammt. Hier plädierte Carnap für eine „Reinigung“ der Erkenntnistheorie, um sie auf das Niveau einer Wissenschaftslogik im eigentlichen Sinne zu bringen, der keine psychologischen, soziologischen und wissenschaftshistorischen Beimengungen mehr anhafteten. Mit dem von Frank favorisierten Ansatz, der Elemente von Machs Sensualismus und James’ Pragmatismus in die Wissenschaftsphilosophie des Logischen Empirismus integrieren wollte, erscheint dies kaum kompatibel. Wäre man aufgefordert, die für die wissenschaftliche Weltauffassung des Wiener

Kreises typischen oder charakteristischen Themen kurz zu nennen, dürfte die allgemein akzeptierte Antwort wohl lauten: „Physikalismus“, „physikali(sti)sche Sprache“, „Einheitswissenschaft“ und „Antimetaphysik“. Einige der Schlüsseltexte der diese Begriffe betreffenden Debatte sind in Kapitel IV versammelt. Die Auffassungen der Protagonisten Neurath, Carnap und Schlick darüber, was unter Physikalismus, Einheitswissenschaft, und Antimetaphysik zu verstehen sei, waren dabei durchaus verschieden. Während Carnap in Die physikalische Sprache als Universalsprache der Wissenschaft zunächst für einen „methodischen Solipsismus“ plädierte, dem zufolge die Protokollsprachen verschiedener Subjekte in eine universale physikalische Sprache übersetzt werden könnten, lehnte Neurath in Soziologie im Physikalismus eine solche Vielfalt von Sprachen grundsätzlich ab. Für ihn gab es nur eine einzige Sprache, die „physikalistische Alltagssprache“ (WK, 278), die aus der von metaphysischen Beimengungen gereinigten Alltagssprache hervorging. Für eine ausführliche Darstellung dieser im Wiener Kreis lebhaft geführten Debatte um den „richtigen“ Begriff von physikali(sti)scher Sprache konsultiere der interessierte Leser Uebels Overcoming Logical Positivism from Within (Uebel 1992). In Soziologie im Physikalismus formulierte Neurath das Programm des einheitswissenschaftlichen Physikalismus in seiner radikalsten Form. Es gebe im Wiener Kreis, so dekretierte er, streng genommen keine Philosophie mehr, sondern nur noch Einheitswissenschaft: „Die Einheitswissenschaft ist in derselben Weise Ergebnis umfassender Kollektivarbeit wie bisher das Gebäude der Chemie, der Geologie, der Biologie, oder auch der Mathematik und Logik“ (WK, 270). Die Sprache dieser Einheitswissenschaft ist die Einheitssprache. Diese Sprache ist im wesentlichen die von metaphysischen Bestandteilen gereinigte Alltagssprache, in der wir raumzeitliche Vorgänge beschreiben. Die Einheitswissenschaft behandelt Aussagen, die raumzeitliche Vorgänge beschreiben, etwas anderes kommt in ihr nicht vor: „Aussagen werden mit Aussagen verglichen, nicht mit „Erlebnissen“ [wie Carnap im Aufbau behauptete], nicht mit einer „Welt“ [wie Schlick in seinen realistischen

271

Momenten verkündet hatte], noch mit sonst etwas“ (WK, 281). Eine Trennung von Geistes- und Naturwissenschaften war von einem einheitswissenschaftlichen Standpunkt sinnlos. Die Soziologie bildete als Sozialbehaviorismus einen Teil der Einheitswissenschaft (vgl. WK, 287). Das zweite für die „wissenschaftliche Weltauffassung“ des Kreises charakteristische Thema war die Enzyklopädie der Einheitswissenschaft, die man wohl als Neuraths Domäne bezeichnen kann. In zahlreichen Schriften zur enzyklopädistischen Problematik plädierte Neurath dafür, wissenschaftliche Theorien oder auch die Wissenschaft als ganze nicht unter der Perspektive logisch strikt geordneter „Systeme“ zu begreifen, sondern von „Enzyklopädien“ als „Modellen“ menschlichen Wissens auszugehen, deren mehr oder minder präzise Aussagen weniger streng mit einander in Beziehung standen als es in „Systemen“ der Fall war. Systematisierungen existierten immer nur lokal, in vielen Bereichen gab es keine Eindeutigkeit, sondern mehrere Möglichkeiten. Um es mit Neuraths drastischen Worten auszudrücken: „Das System ist die große wissenschaftliche Lüge.“ Die Vorstellung der traditionellen Philosophie, das Wissen der Wissenschaft als „System“ begreifen zu können, war eine Illusion, Ausdruck eines überzogenen Pseudorationalismus, der die Reichweite der menschlichen Vernunft überschätzte. Der logische Empirismus hingegen hatte sich nach Neurath von solchen metaphysischen Illusionen befreit und setzte zur Beschreibung des wissenschaftlichen Wissens auf das Modell der Enzyklopädie als „Symbol der Einheit der Wissenschaften und der Brüderlichkeit zwischen den neuen Enzyklopädisten“ (WK, 395). Eine aktuelle Erörterung der vielfältigen Beziehungen zwischen dem Wiener Enzyklopädismus und den französischen Enzyklopädisten des 18. Jahrhunderts findet man in den in Paris-Wien. Enzyklopädien im Vergleich (Wien 2005) versammelten Arbeiten, die vor kurzem von Elisabeth Nemeth und Nicolas Roudet herausgegeben wurden. Aus dem V. Kapitel Zum Basisproblem der empirischen Wissenschaften (Protokollsatzdebatte) scheint mir für die späteren wissenschaftsphilosophischen Debatten besonders

Neuraths frühe Kritik von Poppers Logik der Forschung wichtig zu sein, die er in Pseudorationalismus der Falsifikation vortrug. Popper fühlte sich bekanntlich in seiner Autobiographie bemüßigt, die rhetorische Frage „Who killed Logical Positivism?“ mit der eitlen Selbstbezichtigung zu beantworten, er selbst habe es getan, wenn auch nicht absichtlich. Neuraths Version des logischen Positivismus kann er dabei kaum gemeint haben. Neuraths Kritik an Popper ist eine der überzeugendsten Anwendungen seines Enzyklopädismus, die deutlich macht, daß dessen Falsifikationismus auf sehr stark idealisierenden „pseudorationalistischen“ Annahmen über die Struktur wissenschaftlicher Theorien beruht, die mit Blick auf die Praxis der Wissenschaften wenig plausibel erscheinen. Neurath schlägt vor, als Modelle empirischer Theorie nicht deduktive Systeme, sondern lockerer strukturierte Enzyklopädien zu benutzen, die Poppers strikte Falsifikationen als pseudorationalistische Artefakte entlarven. Von den in Kapitel VI behandelten Spezialproblemen einzelner Wissenschaften scheint mir Hahns Die Krise der Anschauung eine der interessantesten Arbeiten zu sein. Hahn, obwohl von Frank mit einigem Recht als der eigentliche Gründer des Wiener Kreises gepriesen, ist bis heute auch unter Philosophen eine der weniger bekannten Figuren des Wiener Kreises geblieben, da der Schwerpunkt seiner wissenschaftlichen Produktion auf seinem Fachgebiet, der Mathematik, lag. Gleichwohl sind seine wenigen wissenschaftsphilosophischen Beiträge, die sich hauptsächlich mit mathematikphilosophischen Problemen befassen, bis heute lesenswert. Sie zeichnen sich dadurch aus, daß sie sich nicht auf die allzu bekannten traditionellen Themen der elementaren Zahlentheorie und Geometrie beschränken, sondern Philosophie der Mathematik aus der erfahrungsgesättigten Perspektive eines praktizierenden Mathematikers treiben. Hahn behandelt die „Krise der Anschauung“ nicht wie üblich am bekannten Beispiel der nicht-euklidischen Geometrien, sondern erörtert sie an Beispielen raumfüllender Kurven und überall stetiger, aber nirgends differenzierbarer Funktionen, welche die Rolle der Anschauung für die mathematische Erkenntnis viel gründlicher erschüt-

272

tern als die üblichen in der Philosophie der Mathematik diskutierten Beispiele. Hahn war, den Präferenzen des Wiener Kreises entsprechend, einer der wenigen Mathematiker, die eine logizistische Begründung der Mathematik im Sinne von Russell und Whitehead befürworteten, und so zog er aus den neueren Entwicklungen der Mathematik das Resumé: „Der Raum der Geometrie ist nicht eine Form reiner Anschauung, sondern eine logische Konstruktion“ (WK, 539). Das letzte Kapitel von WK bringt mit Bergmanns Erinnerungen an den Wiener Kreis (1938/1939) ein wenig (auto)biographisches Kolorit in den Text. Bergmanns Erinnerungen sind ein von Neurath in Auftrag gegebener Brief Bergmanns an ihn, in dem dieser seine subjektiven Eindrücke des Wiener Kreises schildert. Bergmanns Erinnerungen bilden ein lesenswertes Gegengewicht zu den oft ideologisch geglätteten offiziellen Darstellungen Neuraths und Franks. Insbesondere sein Bericht über die Auseinandersetzungen zwischen den Anhängern Wittgensteins und dem „linken“ Flügel des Kreises und die Charakterisierung von Schlicks Persönlichkeit sind aufschlußreich. 3. Trotz des beeindruckenden Umfangs von WK sollte der Leser nicht glauben, mit WK die gesamte Philosophie des Wiener Kreises in handlicher Form ins Bücherregal stellen zu können. WK ist wesentlich unvollständig. Das gälte selbst dann, wenn WK alle relevanten Artikel des logischen Empirismus des Wiener Kreises im engeren Sinne enthielte, was nicht der Fall ist. Der logische Empirismus des Wiener Kreises kann nur mit sehr harten Schnitten aus dem Kontext seiner philosophischen Umgebung herausgelöst werden, so daß die Beschränkung auf Texte, deren Verfasser in der fraglichen Zeit gerade in Wien lebten, einigermaßen zufällig erscheinen muß. Die Feststellung, daß WK als Anthologie der wissenschaftlichen Weltauffassung Lükken hat, ist nicht als Kritik an den Herausgebern gemeint: Zwischen zwei Buchdeckel passen nun einmal kaum mehr als 600–700 Seiten, und die von den Herausgebern getroffene Auswahl der Texte für WK ist durchweg gut begründet. Der Grund für die Unvollständigkeit ist einfach, daß die „wissenschaft-

liche Weltauffassung“ mehr hervorgebracht hat als in einem einzigen Band Platz finden könnte. Es ist nicht schwer, ein oder zwei Dutzend Texte aufzuzählen, die „unbedingt“ noch hätten aufgenommen werden müssen. Dazu einige Beispiele: Schlicks Die philosophische Bedeutung des Relativitätsprinzips von 1915 ist ohne Zweifel ein Schlüsseltext in der philosophischen Debatte des 20. Jahrhunderts über die Relativitätstheorie. Er repräsentiert jedoch nur ihren Beginn. Bereits in Kritizistische oder empiristische Deutung der neuen Physik von 1921 nahm Schlick eine durchaus andere Position ein. Diese Wandlungen reflektieren ein wichtiges Moment in der internen Entwicklung der wissenschaftlichen Philosophie. Andere Lücken treten bei der Repräsentation von Carnaps Beiträgen auf. Wenn man – mit Recht – einige der frühen Arbeiten von Neurath, Frank und Schlick in eine Sammlung von Texten der Philosophie des Wiener Kreises aufnimmt, dürfte es kaum Argumente dagegen geben, das nicht auch mit einigen der frühen Arbeiten Carnaps zu tun, selbst wenn er nur im übertragenen Sinne ein “Gründer” des Wiener Kreises gewesen ist. Texte wie Über die Aufgabe der Physik (1923), Dreidimensionalität des Raumes und Kausalität (1924) oder Eigentliche und uneigentliche Begriffe (1927) gehören zu den zentralen Texten der wissenschaftlichen Philosophie im ersten Drittel des 20. Jahrhunderts. Auslassungen anderer Art machen sich im Zusammenhang mit späteren Arbeiten Carnaps wie Wahrheit und Bewährung bemerkbar, die letztlich nur im Zusammenhang mit Carnaps Übernahme von Tarskis Semantik richtig zu verstehen sind, was etwa Arbeiten wie Tarskis Der Wahrheitsbegriff in den formalen Sprachen (1935) und Grundlegung der wissenschaftlichen Semantik (1936) ins Spiel bringt. Analog kann man für Arbeiten der von Reichenbach, Hempel, Grelling und anderen gebildeten Berliner Gruppe argumentieren. Darüber hinaus wären die Verbindungen der europäischen wissenschaftlichen Philosophie mit dem amerikanischen Pragmatismus zu nennen, insbesondere mit Charles Morris, dem es in Arbeiten wie Semiotic and Scientific Empiricism oder Logical Positivism, Pragmatism and Scientific Empiricism explizit um eine Synthese zwischen Logischem

273

Empirismus und amerikanischem Pragmatismus ging. Man mag einwenden, Betrachtungen dieser Art wären müßig, da sie vom Hundertsten ins Tausendste führten. Ich halte diesen Einwand nicht für stichhaltig: Daß eine einigermaßen vollständige Sammlung von Schlüsseltexten des Logischen Empirismus kein Ding der Unmöglichkeit ist, zeigt die vor einiger Zeit von Sarkar herausgebene sechsbändige Sammlung Basic Works of Logical Empiricism (New York 1996). Sie umfaßt etwa 120 Texte. Damit ist m.E. eine Größenordnung gegeben, bei der man tatsächlich von Vollständigkeit in einem realistischen Sinne sprechen kann. Sarkars Sammlung ist

jedoch nur eine unkommentierte und hastige Kompilation von Texten, die einem Vergleich mit der vorzüglich edierten und kommentierten Anthologie von WK nicht standhält. Was man sich wünschte, wäre eine Synthese aus beidem – eine umfassende und zugleich sorgfältig edierte Sammlung der zentralen Texte des Logischen Empirismus. WK wäre ein exzellenter Anfang für ein solches Unternehmen. Thomas MORMANN University of the Basque Country UPV/EHU Donostia-San Sebastian, Spain

274

Grazer Philosophische Studien 76 (2006), 275–278.

Daniel COHNITZ & Marcus ROSSBERG: Nelson Goodman. London: Acumen, 2006. 288 pp. ISBN 1844650375. £ 15,99. Als Nelson Goodman Ende 1998 im Alter von 92 Jahren starb, herrschte in den Nachrufen Einigkeit darüber, einen der wichtigsten Vertreter des Faches verloren zu haben, dessen Werk sich insbesondere durch seine Breite auszeichnete, die von der Mereologie und Induktionslogik über die Wissenschafts- und Symboltheorie bis zur analytischen Ästhetik reicht. In allen Bereichen haben seine Arbeiten die Diskussion zeitweise mitbestimmt oder geleitet. Umso bemerkenswerter erscheint es, dass mit der Monografie von Daniel Cohnitz und Marcus Rossberg eines der ersten Bücher zu Goodman erschienen ist, das sein Werk in Gänze darzustellen versucht. Es folgt damit einer guten angelsächsischen Tradition, die darin besteht, hochwertige „Lehrbücher“ bzw. Monografien über Philosophen zu verfassen, bevor deren Werk – bereits kanonisiert – von der Philosophie in die Philosophiegeschichte übergangen ist. Dies mag dem stärker verschulten geisteswissenschaftlichen Studium geschuldet sein, führt aber dazu, dass man solche Werke, die einen guten Zugang bieten, der deutlich über Erstsemesterniveau liegt, in Reihen wie diesen findet. Der Grund für die spärliche Einführungsliteratur liegt wohl nicht nur in der Breite der Philosophie Goodmans, sondern in dem Umstand, dass diese von der Mereologie bis zur Ästhetik auf eine anspruchsvolle Weise miteinander verbunden ist. Und neben einer systematisch betriebenen Philosophie war es auch ein in seiner Biografie verankertes ganz praktisches Interesse an der Kunst, das ihn auf besondere Weise ausgezeichnet hat. In der biografischen Einleitung weisen die Autoren bereits darauf hin, dass Goodmans Leben in gewisser Weise zweigleisig verlief. Im Unterschied zu vielen Kollegen, die sich gegen Ende ihrer wissenschaftlichen Karriere ein weiteres Steckenpferd suchen, das sie mehr oder weniger geschickt abseits ihres bisherigen Forschungsgebietes reiten, setzte Goodmans Interesse an der Kunst erheblich

früher ein. Es bloß „Interesse“ zu nennen, trifft es auch nicht, da es in der langen Zeit zwischen seinem ersten Abschluss und seiner Promotion auch Broterwerb war. Als Leiter einer Kunstgalerie zwischen 1928 und 1941 war er erfolgreicher Praktiker im Bereich der Kunst. Dies ist ausgesprochen ungewöhnlich für die philosophische Ästhetik, der ein breiterer Kontakt gerade mit der zeitgenössischen Kunst und den außeruniversitären Diskursen häufig zu fehlen scheint, auch wenn einem mit A. C. Danto und seinen zahlreichen Kunstkritiken sofort eine weitere Ausnahme einfällt. Interessant ist in diesem Zusammenhang auch das von Goodman 1967 mitbegründete Project Zero, das sich der Entwicklung und Verbreitung der Kunsterziehung als besonderem Teil der kognitiven Entwicklung und des kreativen Denkens widmet. Neben der analytischen Philosophie gab es damit immer auch einen ganz praktischen Teil seines Lebens, der sich mit Kunst beschäftigte und gleichfalls der Grund dafür war, diese beiden Bereiche auch theoretisch miteinander zu verbinden. In einer gelungenen Kontextualisierung wird Goodman in seiner Opposition zu Henri Bergson auf der einen Seite und der analytischen Philosophie auf der anderen positioniert, und innerhalb des analytischen Lagers wiederum zwischen der ordinary language und der ideal language philosophy mit ihrem damals prominentesten Vertreter Rudolf Carnap. Es ist für die Autoren insbesondere der Einfluss von Whitehead, über den sich Goodmans Philosophie charakterisieren lässt. Von ersteren stammt wohl auch die für Goodman grundlegende Idee, dass Kunst, Wissenschaft und Religion (nur) unterschiedliche Sprachen verwenden. Ähnliches gilt für C. I. Lewis, der Wahrheit in Abhängigkeit von dem gewählten konzeptuellen Schema sieht, wobei die Wahl der Schemata pragmatischen Gründen unterliegt. Etwas überraschend und auch nicht unbedingt der Chronologie folgend beginnen die Autoren im zweiten Kapitel bereits mit dem sogenannten Goodman-Paradox, dem bekanntesten Teil seiner Philosophie. Dabei handelt

es sich um eine argumentative Bereicherung von Humes Rätsel der Induktion. So ist eine Gleichförmigkeit zu unterstellen, wenn wir aus Beobachtetem wenn-dann Relationen erstellen und auf noch nicht Beobachtetes projizieren. Nach Hume sei diese unterstellte Gleichförmigkeit nicht mehr als eine Form der Gewohnheit. Goodmans Antwort geht nun dahin, dass es die Verankerung der Regeln und Begriffe in unserer interpretativen Praxis ist, die eine Projizierbarkeit gewährleistet. Ohne eine funktionale Vorauswahl ist es möglich, Prädikate zu bilden, die zu paradoxen Aussagen führen. In seinem bekannten Beispiel ist es „grue“ für Saphire, die vor einer bestimmten Zeit t grün sind und danach blau. Dieses „neue Rätsel der Induktion“ hat eine Publikationswelle auch der bekanntesten Philosophen aus diesem Bereich ausgelöst, die immer noch nicht ganz abgeebbt ist. Anzumerken ist, dass weniger Goodmans auffallend pragmatisch wirkender Lösungsvorschlag einer Verankerung/Entrenchment diskutiert wurde, als die Frage, ob es sich tatsächlich um ein Paradox handelt. Als Roter Faden zum dritten Kapitel dient die so zentrale Frage nach der Projizierbarkeit. Dazu erfolgt eine Darstellung von Goodmans Selbstverständnis als Philosoph und der Rolle, die dabei Definitionen spielen. Dem locker angeschlossen ist Kapital 4, in dem sein Nominalismus und sein frühes Verhältnis zu Quine, neben der Mereologie, dargestellt werden. Mit den hier gewonnenen Voraussetzungen nähert man sich in Kapitel 5 dem anspruchsvollen Hauptwerk: The Structure of Appearence. Die Qualität dieses in weiten Teilen auf seiner Doktorarbeit A Study of Qualities beruhenden Werkes wird noch seltener bestritten, als es gelesen wird. Den Autoren gelingt es hier, zuerst durch eine Rekonstruktion der Bedeutung von Carnaps Der logische Aufbau der Welt seine Grundzüge herauszuarbeiten. Die Darstellung der Zusammenarbeit mit Leonard dient mit dazu, die Unterschiede zu Carnap und wiederum seine Opposition zu Lewis darzustellen. In der Folge lässt sich ein genaueres Bild von Goodmans Nominalismus, seinem Realismus und kurzzeitigem Platonismus nachzeichnen. Um es komplett zu machen, wird unter dem Schlagwort

„Goodman’s Realism“ seine Position noch kurz und prägnant mit Bezug auf die großen Gegensatzpaare wie Nominalismus versus Platonismus oder Realismus versus Partikularismus dargestellt. Insgesamt handelt es sich um das Hauptkapitel des Buches, in dem es den Autoren gelingt, ausgesprochen klar und überzeugend die Grundzüge der Philosophie Goodmans und seine Position innerhalb der großen philosophischen Kontroversen darzustellen. Weiter wird sichtbar, inwiefern diese frühe Arbeit Grundlagen für einen Zugang zu dem bereits diskutierten Induktionsproblem bereitstellt,wie auch für seine Symboltheorie, die im folgenden sechsten und siebten Kapitel vorgestellt wird. Sprachen der Kunst setzt mit seinem Entwurf einer allgemeinen Symboltheorie die von Carnap aufgenommene Idee konsequent um, dass es sich bei Kunst, Philosophie und Wissenschaften um unterschiedliche Sprachen handelt. Symboltheoretisch gesehen bedeutet dies, dass sie sich zum einen darin unterscheiden können, wie sich die jeweiligen Symbole auf ihren Gegenstand beziehen, und zum anderen, wie die Notationsbedingungen für die jeweiligen Symbolsysteme aussehen. Damit lassen sich Literatur, Musik, Tanz und Malerei unterschiedlich charakterisieren und Überlegungen zum jeweiligen Verhältnis von Original und Fälschung anstellen. Neben diesen an allgemeinen Eigenschaften von Symbolen und Symbolsystemen orientierten Überlegungen gibt es auch einen zweiten, in späteren Arbeiten stark ausgebauten Zugang zur Kunstphilosophie. Es ist dieser eher semantisch-pragmatisch zu nennende Aspekt, dessen Darstellung in der Monographie recht kurz kommt, obwohl er ein zentraler Bestandteil gerade des späteren bedeutungstheoretischen Konzeptes von Goodmans ist. So ist bedeutungstheoretisch gesehen gerade für die Kunst eine besondere Form der Bezugnahme interessant: die Exemplifikation. Grob gesagt meint dies eine Form des zum Ausdruck Bringens. Ähnlich wie bei einer Probe oder einem Beispiel läuft die Bezugnahme hier umgekehrt, von dem Gegenstand zum Symbol oder Label. Ein Bild exemplifiziert etwa den Begriff Reichtum – der wiederum denotiert es. Wahrheitsfähig wäre aber nur das Letztere, die Bezeichnung

276

eines Gegenstandes durch ein Symbol oder Label. Eine Probe oder ein Beispiel sind auch umgangssprachlich nicht wahr, sondern gut, gelungen oder eben richtig. Und Richtigkeit statt des engeren Begriffs der Wahrheit zu verwenden ist später Goodmans genereller Vorschlag. In Sprachen der Kunst ist hier eher noch von einem Passen die Rede, das gleichfalls recht pragmatisch ein Passen zu unserer vorgängigen Interpretationspraxis sowie unseren Zielen und Wünschen meint und ebenso ein zu vollziehendes interpretatives Einpassen. Ohnehin ist sein Zugang zur Kunst ein kognitionstheoretischer und kein stil- oder geschmacksorientierter. Das heißt, dass sich Kunst nach ihrer kognitiven Wirksamkeit beurteilen lässt, das heißt nach ihrem Einfluss auf unsere anderen Symbolwelten. Es sind auch dies Weisen der Welterzeugung beziehungsweise Weltveränderung. Zusammen mit ihrer Richtigkeit macht die kognitive Wirksamkeit die Beutung von Kunstwerken genau so wie von wissenschaftlichen Theorien aus. Aufgrund dieser Bedeutungskonzeption ist die Parität deutlich, die Goodman schon in einem Buchtitel ‚der Philosophie und anderen Künsten und Wissenschaften‘ zuweist. Und sie findet sich ebenfalls in dem Konzept des angesprochenen Projektes Zero wieder. Das siebte Kapitel nimmt zudem den zweiten breit rezipierten Teil der kunsttheoretischen Arbeiten auf, neben den bedeutungsund symboltheoretischen Aspekten, der sich mit der damals zentralen Fragestellung auseinandersetzt: Was ist ein Kunstwerk? Kunstintern bereits länger durch Ready-mades und Pop-Art thematisiert war es gerade die noch junge analytische Ästhetik, die diese Frage im Unterschied etwa zu der nach der Bedeutung der Kunst zu ihrem Hauptthema machte. Der klassische Werkbegriff und die Grenze zwischen Kunst und Kommerz konnten für den Gallerieleiter Goodman als aufgelöst gelten, auch wenn andere Bereiche der philosophischen Ästhetik daran festhalten. Eine geschlossene Definition für Kunstwerke ist damit nicht zu erwarten. Und kunsthistorisch gesehen passt die Antwort genau zu den Ready-mades, wenn er als alternative Frage vorschlägt: „Wann ist es Kunst“, das heißt, wann wird ein Gegenstand als Kunstwerk wahrgenommen. Die Antwort ist allerdings eher

technisch, wenn als Kriterien das fakultative Vorliegen von mehreren allgemeinen Charakteristika von Symbolen und Symbolsystemen vorgeschlagen wird wie etwa Exemplifikation oder semantische Dichte. Es erscheint sinnvoller, sich bei der Annäherung an einen offenen Kunstbegriff eher an dem Konzept einer kognitiven Wirksamkeit und einer interpretativen Kompetenz zu orientieren, wie sie im Spätwerk eingeführt wurde. Goodmans Philosophie ist nicht nur voller Label oder Etiketten, sondern auch voller Etikettierungen. In Kapitel 8 wird mit seinem Irrealismus das letzte größere thematisiert, das sich aus seinem Pluralismus und Konstruktivismus ergibt. Bei all den Weisen der Welterzeugung stellt sich die Frage, ob man verschiedene Welten erhält oder Weltversionen. Und falls man die Rede von Weltversionen bevorzugt, ist damit eine ihnen gemeinsame Welt unterstellt? Diese und ähnliche Fragen werden genutzt, um den vertretenen Irrealismus gegen einen Irrationalismus und andere postmoderne relativistische Positionen abzugrenzen, die Goodman fremd sind. Das neunte und letzte Kapitel ist in weiterführender Absicht seinen Kritikern vorbehalten, insbesondere C. I. Lewis, was die Absicht der Autoren unterstützt, die Aktualität von Goodmans Philosophie zu unterstreichen. So werden unter anderem Diskussionen zu den Themen Irrealismus, Pluralismus sowie permanente Akzeptierbarkeit und Wahrheit wieder aufgenommen. Gemessen an der Aufgabe, die Philosophie Goodmans nicht nur in ihrer Breite, sondern auch in ihrem Zusammenhang darzustellen, überzeugt das Buch von Cohnitz und Rossberg durch seine Klarheit in der Argumentation und die ausgezeichnete Lesbarkeit. Wünschenswert wäre nur eine stärkere Auseinandersetzung mit den die Kunst und das Ästhetische betreffenden bedeutungstheoretischen Aspekten gewesen, die die von Elgin so genannte „epistemische Wende“ hätte deutlicher werden lassen. Nach Einschätzung des Rezensenten liegt die aktuelle Bedeutung von Goodmans Philosophie am stärksten hier im Bereich einer kognitiven oder analytischen Ästhetik sowie einer allgemeinen Interpretationstheorie.

277

Aber es ist gerade die aus der nichtanalytischen Ästhetik, Kunsttheorie und -philosophie stammende Zielgruppe, der diese Monografie zu empfehlen ist. Nicht weil es sowieso die einzige ist, sondern weil es ihr gelingt, Goodmans Position in den voraussetzungsvollen wissenschaftstheoretischen und analytischen Grundsatzdebatten deutlich und nachvollziehbar zu machen. Denn leider hat

es sich gezeigt, dass ihre fehlende Kenntnis das größte Hindernis bei der angemessenen und wünschenswerten Rezeption eines der interessantesten Denkers ist.

278

Georg PETER Projekt ProtoSociology Johann Wolfgang Goethe-Universität Frankfurt am Main

Grazer Philosophische Studien 76 (2008), 279–285.

Dean ZIMMERMAN (ed.): Oxford Studies in Metaphysics, Vol. 2. Oxford: Oxford University Press, 2006. x + 400 pp. ISBN 019-929059-8. £55.00, €105.50, $99.00 hardcover; £18.99, €25.34, $35.00 paper. This book brings together an interesting selection of contemporary writing on a cross-section of important topics in analytic metaphysics. Oxford Studies in Metaphysics, under the general editorship of Dean Zimmerman, publishes new work in many areas of ontology, philosophy of mind, time and causation, and philosophical theology. The papers provide a snapshot of the topics and arguments currently of interest to professional metaphysicians. The same general function could in principle be served by journal publication, but there is currently no high profile philosophy journal dedicated specifically to analytic metaphysics, and the advantage of the Oxford Series is that it allows contributors the space they need to adequately address their topics virtually without restriction. In the present volume the papers range in length from 10 to 76 pages, with quite a few running more than 50 pages. The Oxford Series, additionally, unlike almost all journals, also has a useful index. The essays, thirteen in all, are organized into the following four sections with papers by the indicated authors: I. Symposium, Property Dualism: Ned Block, ‘Max Black’s Objection to Mind-Body Identity’; John Perry, ‘Mary and Max and Jack and Ned’; Stephen L. White, ‘A Posteriori Identities and the Requirements of Rationality’. II. The Open Future: Trenton Merricks, ‘Goodbye Growing Block’; Eli Hirsch, ‘Rashi’s View of the Open Future: Indeterminateness and Bivalence’; Peter Forrest, ‘General Facts, Physical Necessity, and the Metaphysics of Time’. III. Issues in Ontology: Thomas Hofweber, ‘Inexpressible Properties and Propositions’; Michael Loux, ‘Aristotle’s Constituent Ontology’; Phillip Bricker, ‘The Relation Between General and Particular: Entailment vs. Supervenience’; John Hawthorne, ‘Epistemicism

and Semantic Plasticity’. IV. Metaphysics and Theism: Brian Leftow, ‘God and the Problem of Universals’; Michael Bergmann and Jeffrey Brower, ‘A Theistic Argument Against Platonism (and in Support of Truthmakers and Divine Simplicity)’; Hud Hudson, ‘Beautiful Evils’. Taken together, the essays constitute a valuable sampling of current work in metaphysics, and there are surprising, surely unintentional thematic connections between papers from one section to another. What one learns from such a collection is not merely what topics are of interest to this medley of writers and what philosophical conclusions they find most defensible, but also the style in which metaphysics is now being practiced by these leading figures in the field. The first suite of three papers centers on a target essay by Block. The essays constitute a ‘symposium’, but it is unclear if the symposium existed outside of the Oxford Studies pages. Block mentions several venues in which ancestors of the paper received comments, but I could not discover whether the three participants, Block, Perry, and White, ever had a viva voce exchange at which Block’s paper was discussed. Block’s paper is in any case a remarkable study that takes its inspiration from Max Black’s objection to mind-body identity, but soon transitions into a detailed discussion of Frank Jackson’s so-called ‘knowledge’ argument for (some type of ) mind-body dualism (the least offensive version of which is often held to be property dualism) associated with Jackson’s thought experiment of Mary the color scientist. Interest in this much-discussed example is by no means exhausted, as the first three papers in this collection clearly demonstrate. Block enters into exquisite detail, pursuing the nodes of an extended decision tree of fine-grained distinctions describing exactly how the color scientist’s experience is to be described, with an eye at every step to its implications for the knowledge argument in defense in particular of property dualism. There is some difficulty, in Block’s view, as to just what the knowledge argument is, what it is trying to say, and how its implications for the mind-body problems should be under-

stood. I think I was less sure about what Jackson’s thought experiment is meant to describe after reading Block’s article than before, which might be either a tribute to its value or a criticism of its unnecessarily complicating an already thorny problem; nor am I even sure after working through Block’s paper several times which of these is the more appropriate evaluation. The conclusion toward which Block progresses inch by inch in the paper is that Jackson’s thought experiment does not adequately support property dualism, which clears the way for Block’s preferred interpretation and defense of a variant form of the doctrine, which is mentioned but not developed in the paper. What I found most wanting in Block’s exposition from my own philosophical standpoint—which happens to strongly favor an emergentist form of property dualism grounded on the irreducible intentionality of mental acts, and is ideologically friendly to anti-physicalist interpretations of Jackson’s color scientist example—is the failure to clarify the concept of property. Thus, when Block writes: ‘One can acquire new knowledge about old properties by acquiring new concepts of them. I may know that there is water in the lake and learn that there is H20 in the lake. In so doing, I do not learn of any new property instantiated, and in that sense I do not learn of any new fact’ (9), I do not know whether to think he is right or wrong; it would rather depend on what a property is and how properties are related to facts. In light of Saul A. Kripke’s lectures on Naming and Necessity, we have grown accustomed to regarding natural kind terms like ‘water’ as designating the same stuff in every logically possible world in which they designate anything at all. The situation is more complicated when we try to move beyond the property of being water (in any of its solid, liquid, or vaporous forms) to the property of containing water. Certainly, the property of having a mouthful of (liquid) water is not the same as the property of having a mouthful of H20 steam, even though (liquid) water = H20, and H20 steam = H20, in every logically possible world, despite the transitivity of identity. Moreover, the property of being (liquid) water is not the same property as the property

of being H20 steam, once again, even though (liquid) water = H20, and H20 steam = H20, in every logically possible world, and despite again assuming the transitivity of identity. It is the states of the stuff, liquid, solid ice, or vapor, after all, that matter in the truth conditions of these judgments, even if the stuff itself, H20 qua natural kind, is always necessarily the same. The states of a stuff, in turn, as in the case at hand, the mental and purely physical states of a neurophysiological system, are among its properties, that enter into psychological facts and that are relevant to interpretations of the knowledge argument in Jackson’s color scientist example as supportive of or indifferent to the merits of property dualism. Such aspects of a neurophysiological system are relevant in particular to whether or not Mary upon emerging for the first time from her black and white laboratory acquires new knowledge about the properties of and facts about the outside world of color. Block, much later in the paper, explores some of the ways of distinguishing facts and properties as having thick or thin, narrow or wide content, in roughly the way these distinctions are usually drawn. What Block turns to instead for the greater part of the essay is a discussion of different modes of presentation, cognitive (CMoPs) and metaphysical (MMoPs), of what he thereafter takes to be the same properties and same facts of the color world merely understood in different ways by Mary before and after her confinement to her original black and white environment, as the only ways of coming to terms with the significance of Jackson’s thought experiment. Block’s discussion in this regard arguably takes a decisive wrong turn at a crucial early part of the discussion, one that represents a missed opportunity, and that is bound unnecessarily to prejudice discussion against the relevance of the color scientist example in maintaining some version of property dualism. Block’s essay nevertheless is certain to provoke continued controversy as philosophers struggle to assimilate its dense and complicated interrelations of distinctions and theses in recent philosophy of mind, and its publication in this venue is an important event. Thus, I found myself much in agreement with Perry’s concise response to Block’s essay

280

immediately following it, when he says, at the very end of his article: ‘Block’s paper is rich and interesting—and long. I don’t claim to have digested all that he has to say, and have not tried here to discuss all of it. I hope to return to the issue of Block’s argument and the variety of modes of presentation at a later time, after having enlisted the help of seminar students in coming to grips with more of it’ (88–89). I also agree with Perry, as I have indicated above, in his sense that: ‘Pace Block, the knowledge argument is about knowledge … Mary has new knowledge when she steps out of the black and white room and sees a red fire hydrant … Conclusion: her new knowledge is of a non-physical fact’ (79). Perry discusses several features of Block’s essay, raising issues that are likely to add dimension to the problems Block raises, highlighting features of the paper and considering especially what he takes to be Block’s efforts to make the phenomenology of qualia appear mysterious, and his substitution of psychologically distinct modes of presentation for metaphysically distinct kinds of presentations concerning specific categories of knowledge, properties, and facts that are not easily explained if at all on the assumptions of physicalism or mind-body reductivism. White, in his commentary, examines ten areas of disagreement with Block. His central objection is that Block confuses direct access to phenomenal states with descriptive characterizations thereof. White acknowledges that there may be no adequate defense of property dualism by way of Jackson’s color scientist problem if we confine ourselves to alternative modes of presentation, since these are descriptive. He nevertheless sees a flaw in Block’s argument in limiting discussion, in terms of Bertrand Russell’s famous distinction, to knowledge by description as opposed to knowledge by direct immediate acquaintance. As White puts the criticism: ‘Block holds that our normal mode of access to our pains is via phenomenal concepts— descriptions in connection with which an instance of the pain itself occurs. That is, the pain is given in such a way that we stand in a demonstrative relation to it … But this is to confuse directness in Russell’s sense— acquaintance (in which sense data are given

directly and are their own modes of presentation)—with ordinary demonstrative access’ (96). White’s differences with Block shed light on Block’s argument and on the longstanding mind-body problem, from the perspective of Jackson’s thought experiment seen against a background of subtle distinctions in contemporary ontology and philosophy of language. Part II of the volume deals with the metaphysics of time. The three independent papers assembled here deal in different ways with the question of whether and in what sense the future can be regarded as open. The issues are intertwined also in interesting ways with questions about the nature of causation and the freedom versus determinism controversy, as stock topics of classical metaphysics. Merricks’ lead essay in this group nicely states three alternative approaches to the ontology of time as something real. Eternalism holds that all times, past, present, and future, are equally real; presentism is the view that only the present time is real; while the growing block theory of time maintains that the past is real and the present is its constantly growing edge, a pressure wave, so to speak, for the advance of real time, while the future is open and unreal. Merricks’ purpose in the essay is to refute and reject the growing block model, leaving the two first alternatives as the only possibilities for a proper metaphysics of time. Merricks’ objection to growing block stems from a distinction between subjective (temporally fixed in relation to the thoughts of a psychological subjective) and objective notions of time, and the complaint that growing block inevitably conflates the two. The implication is that growing block is logically consistent, but so implausible as to be unacceptable. The growing block theorist would like to make only the future subjective, but the problem is that the theory is committed to the view that it is most probably false for a subject to believe that anything is happening in the present moment, on the grounds that events can only be part of subjective and not objective time until they have decisively become part of the past. To the extent that I understand Merricks’ argument, I find it puzzling. I do not fully grasp why there could not be a growing block theory of time that cate-

281

gorizes past, present, and future as objective, nor do I quite comprehend why it should be that events in the subjective present should be regarded as most probably false according to growing block. The reasoning seems to be that at the growing edge of real time there is a transition from the open future in which any chosen event has no commanding probability, which is rightly attributed only after the event, so to speak, has safely entered the past and can be regarded as fully objective. Why not, however, consider growing block theory as treating all events at the growing edge of the present moment as just as highly probable as those that have become a part of the past? Indeed, in some ways, what is happening exactly now is more probable than what has happened in the past, concerning which as every instant passes we have less access and frequently less accurate or reliable information. In any case, why make the metaphysics of time depend on epistemology? Hirsch’s paper examines an argument in the late 11th, early 12th century Talmud scholar Rashi (Rabbi Schlomo ben Yitzchak), about the openness and indeterminacy of the future. Rashi’s position, developed from consideration of a problem in Talmudic law about the purchase of a specific house that is only determined after the transaction is supposed to have been completed, is contrasted with that of Aristotle in the famous puzzle concerning whether there will be a sea battle tomorrow. Hirsch addresses implications of the problem for bivalence and vagueness, and considers standard approaches offered by supervaluationists, like Aristotle, who accept excluded middle but reject bivalence, and Rashi’s ontic indeterminism, which accepts both bivalence and excluded middle but regards the future as genuinely undetermined. Rashi’s view is interesting precisely because it proposes to establish ontic indeterminacy within a roughly classical logical context. Hirsch evaluates Rashi’s position in application to vagueness and quantum indeterminacy from ontic and epistemic perspectives. As Hirsch concludes in the final sentence of his essay: ‘My main aim…has been to introduce Rashi’s position into the dialectical mix and to suggest that, if we are going to accept the notion of indeterminateness, then it is arguably best

to do so without abandoning bivalence’ (135). The section closes with Forrest’s essay on physical, time-dependent necessity. The approach derives in large part from a David Armstrong-inspired view of events as truthmakers, where the concept of an event is time-related and in that sense time-dependent. Forrest’s paper is interesting not only by virtue of its intrinsic content, but in the present context because it represents the development of a kind of growing block metaphysics of time. Forrest ultimately defends what he calls the Mortmain theory, which he characterizes in these terms: ‘[T]he dead hand of the past is constraining the future … It is much more intuitive to say that reality grows because more things, including more states of affairs, come to exist’ (142). If truth is established by events as truth-makers, and if, as seems reasonable, we cannot assign any definite times to projected future events because they have not yet occurred, then the future is indeed unreal, for there are no time-specific events to serve as truth-makers for propositions about what has not yet occurred. The objectivity of time-specific truth-making events in Forrest’s approach to growing block constitutes precisely the kind of theory that seems to me to avoid Merricks’ objections. The next section of the book is titled ‘Issues in Ontology’, which, in general terms, could obviously serve as well for any of the book’s main divisions. The papers presented here include Hofweber’s study of inexpressible properties and propositions, Loux’s historical examination of Aristotle’s ‘constituent’ ontology, Bricker’s discussion of the distinction between the general and particular, construed as a conflict between matters of entailment and supervenience, and Hawthorne’s treatment of semanticism and epistemic plasticity. Hofweber’s paper is the first winner of the Oxford Studies in Philosophy Younger Scholar Prize (details of the yearly competition, now in its fourth season, are found on pages ix-x). Hofweber describes a form of internalism, according to which properties are not entities (strict internalism) or properties are entities, but not mind- or language-independent (loose internalism). He argues at length that internalism is not refuted by consider-

282

ations of inexpressibility. Referring to a criterion proposed many years ago by Roderick M. Chisholm, Hofweber considers the argument for externalism in which properties as independent entities must be posited in order to account for the possibility of inexpressible truths. Recognizing divisions between extreme and moderate internalism, and between several different concepts of expressibility, enables Hofweber to argue that moderate internalism can accommodate the fact that for any real number there is a larger number than it, that there might have been different properties than there are, and that not every property is expressible in every language. The exact sense of ‘expressibility’ is indeed often the issue, and Hofweber does not always observe the fine nuances available as carefully as he might. An example is his discussion of diagonal arguments for generating properties excluded from any hypothetically complete list by means of diagonalization; for it might be said that the diagonalization procedure together with the list themselves constitute an expression of the excluded properties. A greater sensitivity to the ‘ible’ suffix in ‘expressible’ might also have improved this analysis of the prospects of internalism in light of expressibility and inexpressibility. We are reminded of Wittgenstein’s assertion in Tractatus Logico-Philosophicus 4.002, when he states: ‘Man possesses the capacity of constructing languages, in which every sense can be expressed …’. If Wittgenstein is right, then there are no inexpressible properties or propositions. The apparent disagreement between Wittgenstein and Hofweber on this issue of vital importance to Hofweber’s argument, centering on this passage that Hofweber does not consider or address, must obviously have something to do with a difference in what each means by a sense being expressible or having the capacity to be expressed. Without a clearer sense of what Hofweber means by the concept, however, we are unable to determine at virtually every stage of his argument whether and in what sense certain properties or propositions are supposed to be inexpressible. The problem is clear from the first examples introduced, in which Hofweber argues that the property of tasting better than Diet Pepsi was inexpressible for the ancient Greeks. It appears from

context that what Hofweber means by this is that the ancient Greeks did not have Diet Pepsi, nor the equivalent words in their extant vocabulary. On the same grounds, however, we would have to conclude that the property of being Plato is also inexpressible in ancient Greek up to the time before Plato’s birth or conception, simply because Plato lived only after this time and his (logically proper) name had not yet been added to the language until he was born and named. This would be a very narrow concept of ‘expressibility’, conspicuously different from Wittgenstein’s. It is also paradoxical, in the sense that of course Plato was named in ancient Greek (as Πλατων) by ancient Greeks, who must have found the property of being Plato expressible in what was then their still living and expanding language, which was simply extended in its available vocabulary to include a new name. We might instead find it more natural to hold that the property of being Plato is indeed expressible in ancient Greek, in the sense that the language provides all the resources by which such references can be incorporated without violating its grammatical and lexical rules and categories. In the same sense, the property of being Abraham Lincoln is equally expressible in ancient Greek, in something more like the sense that Wittgenstein may have intended. Loux’s insightful essay on Aristotle’s constituent ontology provides a useful historical treatment of Aristotle’s mereology, in which secondary substances are included among the more basic inherent constituents of primary substances. These are not ‘parts’ in the sense of limbs, bits or particles, but something more abstract, despite their ontic dependence on the primary substances in which they inhere. Loux does a superb job of discussing the historical background and philosophical implications of Aristotle’s theory, which he regards as contributing to the uniqueness of Aristotle’s ontology of properties. As Loux concludes: ‘The substantial form is that universal; it is a primitive causal principle, and it is what makes Aristotle’s ontology unique among those that implement the constituent strategy. His ontology includes composite beings which, nonetheless, exhibit an irreducibly unique or autonomous form of being. That autonomous form of being just is the form

283

of life that expresses itself in the functional organization and behavioral repertoire definitive of a biological kind; and what insures the autonomy of that form of being is the fact that anything that enjoys it does so in virtue of having as a proper constituent an irreducibly basic causal principle—the substantial form—that is equideterminate with that form of being’ (249). Bricker also considers a part-whole theory, holding as against the early Wittgenstein that the world is the mereological sum of all things rather than facts, although he blunts the confrontation with Wittgenstein by allowing facts also to be things. The issue for Bricker is whether, as Russell and Armstrong have held, general facts are needed in addition to particular facts because general truths are not entailed by particular truths, versus the position, accepted by David Lewis and others, according to which general facts are not needed in addition to particular truths because general truths supervene on particular truths. Here is an important opposition, a nice choice for metaphysicians, that reflects more deeply on presuppositions about the nature of reality and truth, and with interesting implications for traditional problems of ontology. Relying also on a truth-maker theory, Bricker evaluates the prospects of factualism in the dispute between entailment and supervenience approaches to the relation between particularity and generality, concluding, as he summarizes his argument early in the paper: ‘[W]hen entailment and supervenience diverge, it is only supervenience that is necessary for ontological determination; failure of entailment carries no ontological force’ (256). The further consequence, as Bricker hints at but does not fully or explicitly acknowledge, is that as a result particularism wins out over generalism; general truths are not required in order to understand all facts about the world over and above all particular truths. Hawthorne’s paper wraps up this section of the book with a detailed treatment of the problem of vagueness in predications. Hawthorne considers the advantages of Timothy Williamson’s epistemicism in application to questions of personhood and semantic properties. Epistemicism is the view that vagueness is nothing ontic, but rather a matter of

limitations in our knowledge. Thus, in the famous sophism of the sorites or heap and bald man, there seems to be no definite answer to the question how many stones are needed to make a heap or exactly how many hairs a head must lack to be bald or have to be nonbald. Hawthorne describes his paper as a report on work in progress, but it is clear that he has made considerable progress in understanding the implications of epistemicism in approaching questions of vagueness. Hawthorne develops two versions of epistemicism, relative to which he expresses a ‘mild preference’ for a metaphysically inflationary formulation. The final section of the book tackles two interesting problems of metaphysical theism. The first two essays by Leftow and Bergmann & Brower consider the prospects of eliminating Platonic Forms in favor of divine concept nominalism, of ideas originating in God’s mind by which all individuals are created by an act of divine will. The concluding paper by Hudson suggests a way of defending God’s existence in light of the problem of natural evil. The Leftow and Bergmann & Brower papers complement one another in the sense that a problem left dangling by Leftow is neatly solved by Bergmann and Brower. Leftow argues that if God exists then there would be no need for Platonism. God in that case possesses all the concepts for everything God chooses to create in single or multiple instances, and does not need to do so by reference to a higher Platonic Idea or Form. Leftow’s position, like Bergmann and Brower’s, is in a way a counter-Euthyphro-ian metaphysics. In Plato’s dialogue by that name, Socrates concludes that what is pious cannot be simply a matter of what the gods find pleasing, since then piety would be altogether arbitrary, subject to their whims as to what pleases or displeases them. The only alternative is that what the gods find pleasing is what is virtuous independently of their attitudes. The existence of an independent standard of what is virtuous points to a Platonic Form or Idea of Virtue to which the gods are also subject and subservient in their judgments as to what is virtuous and pleasing. Similar kinds of objections are often raised against divine command theories of ethics. Leftow and Bergmann & Brower in effect reverse the Euthyphro-ian stance by

284

holding that God’s activism in creating the world of things in accord with divine concepts dispenses entirely with the need for Platonic Forms or Ideas. All that is needed is for God to will to create the world of individual things as agreeing with God’s uniquely instantiated archetypal concepts. It is an intriguing picture of life without the Platonic universals, but it leaves open questions that are taken up by Bergmann and Brower, as to whether or not God’s concepts must not themselves in one way or another presuppose some kind of Platonism. Bergmann and Brower consider a variety of possible ways in which Platonism might yet slip in under the divine concept nominalism door, and argue systematically against each. They see the solution to putative problems under this head as consisting in the thesis of God’s divine simplicity and a commitment to a truth-maker theory. They come at last to the conclusion that the antiPlatonist argument supports the existence of God by appeal to reasons of ontic economy via Ockham’s razor, and additionally lends credence to the divine simplicity thesis and truth-maker semantics. The closing paper by Hudson undertakes to mitigate the negative implications of the traditional problem of evil. Hudson invites us to consider the conceivability of alternative dimensions in which aspects of the natural evils visited upon sentient beings in the phenomenal order are out-balanced by ‘compensating’ factors. He walks the reader through several preparatory imaginative exercises to pave the way for the final concept of evil as only a limited aspect of a more beautiful benevolent reality. The idea in the museum curator thought experiment is that, with respect to hidden dimensions of beauty in an unappealing two-dimensional gallery of statues, as Hudson writes: ‘A plurality of uglinesses may have a beautiful fusion’ (391). Similar lessons are drawn from a modified scenario based on Edwin A. Abbott’s fictional geometrical Flatland world. The overall good

in God’s created world exists, on Hudson’s conception, which he refrains from labeling a theodicy or ‘solution’ to the problem of evil; it is just that we cannot see it because we experience the world in only a limited number of its true dimensions from a limited phenomenal perspective. Does this fantasy offer any meaningful relief from the problem of reconciling the existence of natural evil with the world’s creation by an omnipotent, omniscient, and perfectly benevolent deity? Several things remain puzzling. Why should an adequate solution to the problem of evil need to appeal to ‘compensating’ factors? How does this solve anything? If I slap you mildly in the face causing you pain, and then offer to compensate you by presenting you with a zillion dollars tax-free in your bank account, I may have more than paid for my violation, and you might be so perfectly satisfied with the outcome as to turn the other cheek and wait for the second check to clear. After such an action, however, I now ask rhetorically, could I intelligibly be described as perfectly good, perfectly just or benevolent, as God in the traditional conception is imagined to be? Can there ever be compensating factors for the actions of a perfect being? Secondly, if God is indeed perfectly good, just or benevolent, why, if God is also omnipotent, would God not have created a world lacking natural evil in any of its dimensions, let alone the astonishing quantity and degree of pain in precisely that dimension of being to which conscious beings happen to be confined and to which human sensation and cognition happens to be limited? I fail to see how such considerations about the mere possibility of a layer cake of metaphysical dimensions offers any traction against the problem of evil as a formidable, and as yet unresolved logical-conceptual objection to the existence of God.

285

Dale JACQUETTE University of Berne

Grazer Philosophische Studien 76 (2008), 286–289.

Christian BEYER: Subjektivität, Intersubjektivität, Personalität. Ein Beitrag zur Philosophie der Person. Berlin, New York: Walter de Gruyter, 2006. 212 S. ISBN-10: 3-11018919-4. ISBN-13: 978-3-11-018919-3. € 49,95. Bei der vorliegenden Arbeit handelt es sich um die überarbeitete Fassung der Habilitationsschrift des Autors, die 2003 von der Philosophischen Fakultät der Universität Erfurt angenommen wurde. Sie liefert nicht nur einen „Beitrag zur Philosophie der Person“, wie es im Untertitel heißt, sondern zugleich auch eine Theorie des „intentionalen Bewusstseins“. Die darf schon deshalb nicht fehlen, weil sich eine Person oder ein „genuines Subjekt“ (Beyer verwendet die Begriffe der Person und des Subjekts synonym) wesentlich durch ein Bewusstseinsleben auszeichnet (117). Beyer verfolgt eine klar erkennbare argumentative Linie, die die folgenden Themenbereiche verbindet: die Bewusstheit intentionaler Zustände, die synchrone und diachrone Einheit von „Bewusstseinsströmen“, die Konstitution von Subjekten über Intersubjektivität, das Problem der Identität von Personen über die Zeit hinweg und schließlich die moralische Personalität. An Husserl anknüpfend, verfolgt er dabei einen „transzendentalphänomenologischen“ Ansatz (2). Die Arbeit beginnt mit der Frage, worauf das intentionale Bewusstsein, d.h. unsere „Fähigkeit, bewußt auf etwas Bezug zu nehmen“ (15), letztlich basiert. Beyer vertritt hier eine „indexikalische Metaüberzeugungstheorie“. Nach dieser Auffassung ist sich ein Subjekt, zunächst etwas vereinfacht gesagt, genau dann eines intentionalen Zustands bewusst, wenn es aufgrund dieses Zustands (unmittelbar, nicht-inferentiell) glaubt, in eben diesem Zustand zu sein. Z.B. nehme ich bewusst wahr, dass es regnet, wenn ich selbst davon überzeugt bin, dass ich wahrnehme, dass es regnet. Im Unterschied zu der von Beyer kritisierten Metaurteilstheorie postuliert die Metaüberzeugungstheorie nicht, dass mir ein intentionaler Zustand nur dann bewusst ist, wenn ich tatsächlich das Urteil fälle, dass ich in die-

sem Zustand bin. Die fragliche Überzeugung über mich und meinen intentionalen Zustand äußert sich lediglich in der Disposition, unter passenden Umständen ein entsprechendes Urteil zu fällen – d.h. bei erwachsenen Menschen normalerweise: ihm einen sprachlichen Ausdruck zu verleihen (18). Damit entgeht er dem u.a. bei Carruthers diskutierten Einwand von der kognitiven Überbelastung (cognitive overload): Wir haben kaum die neuronalen Kapazitäten zur Verfügung, die nötig wären, wenn jeder plausiblerweise bewusst zu nennende intentionale Zustand durch ein aktuelles Urteil begleitet würde (18). Natürlich ist diese Konzeption des intentionalen Bewusstseins immer noch recht anspruchsvoll. Offenbar können nur solche Wesen bewusste intentionale Zustände haben, also z.B. bewusst etwas wahrnehmen oder glauben, die über eine ganz spezielle Klasse von Begriffen verfügen: nämlich über solche intentionalen Konzepte wie die der Wahrnehmung, der Überzeugung usw. Das scheint z.B. Kinder unter vier Jahren aus dem Kreis der bewusst Wahrnehmenden und Glaubenden auszuschließen, denn wie aus entwicklungspsychologischen Untersuchungen wohlbekannt ist, kann man von ihnen kaum sagen, dass sie etwa über einen vollständigen Begriff der Überzeugung verfügten. Das wäre eine missliche Konsequenz, zumal bei Beyer „bewusst“ soviel wie „erlebt“ bedeuten soll (16) und man kaum bereit sein wird, kleinen Kindern Wahrnehmungserlebnisse abzusprechen. Dieses Problem versucht Beyer zu lösen, indem er annimmt, dass Kinder etwa in typischen Wahrnehmungssituationen glauben würden, dass sie etwas wahrnehmen, wenn sie schon über den Glaubensbegriff verfügten – und dass ein solcher potentieller Glaube für das Verfügen über ein Wahrnehmungserlebnis schon ausreicht (24 f.). Das mutet nun allerdings, vor allem in der Kürze, in der es vorgetragen wird, wie eine Verlegenheitslösung an. Es ist nicht ohne weiteres zu sehen, wie die Tatsache, dass jemand beim normalen Gang der Dinge in einigen Monaten oder Jahren über einen Glaubensbegriff verfügen

wird, dazu führen kann, dass er jetzt bewusste Wahrnehmungserlebnisse hat. Hier hätte man sich eine etwas ausführlichere Diskussion und eine Auseinandersetzung mit der einschlägigen Literatur gewünscht; dasselbe gilt für die Frage nach Wahrnehmungserlebnissen bei Tieren, für die sich ja ein ganz analoges Problem stellt. Generell wäre eine etwas längere Verteidigung der Metaüberzeugungstheorie nicht fehl am Platz gewesen: Tatsächlich wird auch schon am Anfang nur recht kurz erläutert, wie die durch sie postulierte Beziehung zwischen Überzeugungen über eigene intentionale Zustände und deren Bewusstheit bzw. Erlebnischarakter überhaupt zustande kommen soll (15 f.). Nach einer subtilen und überzeugenden Verteidigung der These, dass der Indexausdruck „ich“ in indexikalischen Metaüberzeugungen der Art „Ich bin überzeugt, dass ich jetzt wahrnehme, dass p“ etwas bezeichnet, gegen Wittgensteins und Anscombes Angriffe (33–45) wird der Metaüberzeugungs-Ansatz in Kapitel 2 verwendet, um die Frage zu klären, wann zwei gleichzeitig bestehende oder aufeinander folgende intentionale Zustände i und j zu einem einzigen Bewusstsein(sstrom) gehören. Kurz gesagt, gehören zwei simultane Zustände i und j zum selben Bewusstseinsstrom, wenn sie Gegenstand einer Überzeugung der Art „Ich habe gerade die Erlebnisse i und j“ sind; zwei aufeinander folgende Zustände i und j gehören zum gleichen Bewusstseinsstrom, wenn sie Gegenstand einer Überzeugung der Form „Ich hatte ein Erlebnis der Art i und dann eines der Art j“ sind (51 ff.). Solche Fragen können sich natürlich kaum aus der Innenperspektive eines bewusstseinsfähigen Subjekts stellen: Ich kann mich ja nicht gut selbst fragen, ob zwei intentionale Zustände i und j, derer ich mir bewusst sein muss, um entsprechende Fragen überhaupt erst einmal zu stellen, zu meinem Bewusstsein gehören. Sinnvoll sind sie bestenfalls aus einer Außenperspektive auf ein Subjekt, in dem es womöglich, vielleicht auch nur temporär, verschiedene Bewusstseinsströme geben könnte. Gedacht ist hier an die typischen Parfit-Fälle, denen Beyer mit seiner Theorie gerecht werden will: Etwa an das (imaginäre) Beispiel

des Prüfungskandidaten, der sein Bewusstsein aufspaltet, um zugleich zwei verschiedene Lösungswege für eine Aufgabe in einer Physikklausur durchzurechnen (48 f.). Die Frage, wie sich der Begriff meiner selbst als eines solchen aus einer Außenperspektive wahrnehmbaren Subjekts konstituiert, diskutiert Beyer in Kapitel 3. Der Grundgedanke ist dabei, dass ich mich selbst als ein derartiges Subjekt verstehen kann, wenn ich begreife, dass es andere Subjekte gibt, die mich ihrerseits als ein Subjekt mit bewussten intentionalen Zuständen auffassen können. Die Annahme, dass ein Konzept meiner selbst als Subjekt eng mit dem Begriff der Intersubjektivität verbunden ist, ist dabei freilich als solche nicht neu. Man denke etwa an Tugendhats an Mead anschließende Überlegungen in Selbstbewußtsein und Selbstbestimmung1, die auch bei Habermas aufgenommen werden. Beyer bezieht sich aber nicht auf diesen Diskussionsstrang, auch nicht auf die Kritik, die insbesondere Manfred Frank an derartigen Vorstellungen geäußert hat.2 Vielmehr knüpft er an die phänomenologische Idee der „Einfühlung“ an, insbesondere an die Gedanken, die Edith Stein in ihrer von Husserl betreuten Dissertation Zum Problem der Einfühlung entwickelt (65 ff.). Einfühlung spielt sich dabei in drei „Vollzugsmodi“ ab. Im ersten Modus der Einfühlung ist mir z.B. der Andere als in einer traurigen Stimmung befindlich unmittelbar gegeben. Ich erlebe den Anderen als traurig, ich stehe ihm gleichsam nicht gegenüber, sondern versetze mich in ihn hinein. Vielleicht spiegelt sich seine Stimmung im entsprechenden Gesichtsausdruck wider, aber ich muss sie nicht erst in einem problematischen Analogieschluss aus dem Gesichtsausdruck erschließen, sondern Trauer und Ausdruck bilden von vornherein eine „natürliche Einheit“ (67) – was impliziert, dass der Andere mir per Einfühlung auf eine gewisse Weise von vornherein als ein Subjekt mit bewussten Erlebnissen gegeben ist. Wenn man so will, erlebe ich ihn als Subjekt: ihn als Subjekt wahrzunehmen ist keine Angelegenheit theoretischer Schlüsse. Im zweiten Modus der Einfühlung versetze ich mich in den Anderen hinein und stelle mir anschaulich vor, wie es wäre, an seiner Stelle z.B. Traurig-

287

keit zu empfinden; oder etwa wie er etwas links von sich sieht, was ich rechts von mir sehe. Dazu muss ich ihn bereits als ein eigenständiges „Orientierungszentrum“ wahrnehmen, das Dinge unter bestimmten Aspekten wahrnimmt (77). Im dritten Modus, dem der „zusammenfassenden Vergegenständlichung“, ist mir der Vorgang des „Sich-Hineinversetzens“ bewusst. Beyer versteht dieses Bewusstsein dann wieder im Sinne der Metaüberzeugungstheorie: Z.B. kann ich die Metaüberzeugung haben, dass ich glaube, dass ich den Anderen als jemanden auffasse, der einen Tisch unter diesen und jenen Aspekten sieht (78). Der letzte Schritt ist dann Steins Idee einer „iterierten Einfühlung“ (im dritten Modus): Ich glaube jetzt, daß ich den Anderen als jemanden erlebe, der mich, das Erkenntnissubjekt, kraft seiner Wahrnehmung eines bestimmten lebendigen Körpers als Besitzer eines Leibes ansieht“ (80) – wobei ein Leib ein mit Bewusstsein begabter Körper ist. Auf diese Weise konstituiert sich mein Begriff von mir als einem von Anderen wahrnehmbaren, leiblichen Subjekt, in dem es dann auch verschiedene Bewusstseinsströme geben kann. Der Zusammenhang zwischen Subjektivität und Intersubjektivität wird bei Beyer also im Anschluss an Stein über das Konzept der „Einfühlung“ hergestellt. Steins Begriff der Einfühlung erscheint verwandt mit dem Begriff der Simulation in der so genannten Simulationstheorie der Alltagspsychologie. Tatsächlich bezieht Beyer im dritten und vierten Kapitel seiner Arbeit dann auch Stellung in der aktuellen Debatte über die „Theorie-Theorie“ und die Simulationstheorie und spricht sich für eine moderate Variante der letzteren aus. Argumente dafür, die über die Anknüpfung an Edith Stein hinausführen, finden sich vor allem in Kapitel 4. Beyer setzt sich dort kritisch mit Davidsons Überlegungen zur „radikalen Interpretation“ auseinander und versucht zu zeigen, dass das Konzept der Einfühlung eine zentrale Rolle spielt, wenn es darum geht, den Äußerungen eines Anderen Sinn zuzuschreiben. Bekanntlich ist eine solche Sinnzuschreibung nach Davidson nur möglich, wenn man voraussetzt, dass die Äußerungen des Anderen in der Regel wahr sind. Im Anschluss an eine Überlegung Føllesdals betont Beyer (99 ff.), dass nicht unbedingt Wahrheit im absoluten

Sinne der entscheidende Punkt ist, sondern Wahrheit, sofern ich Grund zu der Annahme habe, dass der andere aus seiner Position die Wahrheit überhaupt erkennen kann. Man nehme an, der Andere reagiert gewöhnlich mit „gavagai“ auf die Anwesenheit von Kaninchen, und entsprechend habe ich die Hypothese gebildet, dass dieser Einwortsatz in seiner Sprache soviel bedeutet wie „Da ist ein Kaninchen“. Nun sehe ich ein Kaninchen auf der Wiese sitzen, nehme aber zugleich wahr, dass ein Baum zwischen dem Anderen und dem Kaninchen steht, der das Tier für ihn verdeckt. Eine (fälschlich) negative Reaktion des Anderen auf meine Frage „gavagai?“ spricht nun sicherlich nicht gegen meine Hypothese, sondern eher für sie: Schließlich kann der andere das Kaninchen ja nicht sehen und sollte deshalb im Einklang mit meiner Hypothese nicht zustimmen. Beyer sieht das als einen Beleg für die Simulationstheorie an, weil die Wahrnehmungsperspektive des Anderen hier per Einfühlung berücksichtigt werden müsse. Er meint: „Die Applikation geläufiger theoretischer Prinzipien der Alltagspsychologie verbietet sich schon deshalb, weil die Anwendbarkeit dieser Prinzipien auf das Verhalten des Sprechers durch radikale Interpretation allererst etabliert werden muß“ (99). Das ist, jedenfalls in dieser Kürze, nicht recht einzusehen. Erstens wird auch jeder Anhänger der Theorie-Theorie der Forderung zustimmen, dass die „egozentrische Wahrnehmungsperspektive“ eines Sprechers mit „ins Kalkül gezogen“ werden muss – so dass sich allein daraus wohl eher kein „Prinzip der Einfühlungsbasiertheit radikaler Interpretation“ (vgl. 101 und Punkt 2 auf 102 f.) im Sinne einer substantiellen Simulationstheorie ableiten lässt. Und zweitens wäre ausführlicher dafür zu argumentieren, dass „die Applikation geläufiger theoretischer Prinzipien der Alltagspsychologie“ sich im Prozess der radikalen Interpretation verbietet: Zumindest auf den ersten Blick scheint man doch sagen zu können, dass man das Unternehmen der Radikalinterpretation auch als Versuch auffassen kann, den Anderen im Sinne der alltagspsychologischen Theorie als ein sprach- und bewusstseinsbegabtes Subjekt aufzufassen.

288

Generell kann man bezweifeln, ob das Gedankenexperiment der radikalen Interpretation über die „kognitiven Mechanismen Aufschluß“ gibt, „die unserer sprachlichen Kommunikation über raumzeitliche Objekte zugrunde liegen“, wie Beyer meint (103, Punkt 3). Es gibt sicherlich Aufschluss darüber, was zu leisten ist – nämlich die Berücksichtigung der Perspektive des zu interpretierenden Sprechers. Die Frage, wie und durch welche „kognitiven Mechanismen“ diese Aufgabe dann faktisch gelöst wird, könnte eher eine Angelegenheit echter (psychologischer) Experimente sein. Um Gedankenexperimente geht es auch im fünften Kapitel, wo Beyer die Probleme der diachronen Personenidentität diskutiert. Nach einer gut begründeten Zurückweisung von Chisholms These der „strikten, kriterienlosen Identität“ (120 ff.) widmet er sich den bekannten Teleportations-Fällen und entwickelt eine „kontext-sensitive Konzeption“ (128 ff.). Sie läuft darauf hinaus, dass physische Kontinuität und psychischer Zusammenhang je nach Kontext unterschiedlich wichtig sein können. Wird mein Körper in einem Fall „funktionierender“ Teleportation an einem bestimmten Ausgangsort zerstört und woanders genau so wieder zusammengesetzt, dass er die gleichen physischen und psychischen Eigenschaften hat wie zuvor, so spielt die physische Diskontinuität im Transport keine entscheidende Rolle: Die Person ist nachher dieselbe wie vorher. Wird jedoch aufgrund eines Defekts mein ursprünglicher Körper am Ausgangsort nicht zerstört, obwohl am Zielort ein neuer, gleicher Körper hergestellt wird (Kopie statt Transport), so wird die physische Kontinuität relevant: Die Person am Ausgangsort ist nach dem Vorgang dieselbe wie vorher, und am Zielort ist eine neue Person entstanden. Eine ganze Reihe interessanter Überlegungen enthält das sechste und letzte Kapitel

über „moralische Personalität“. Beyer argumentiert dort für die These, jemand sei eine moralische Person und damit für seine Handlungen verantwortlich zu machen, wenn er über „De-jure-Willensfreiheit“ verfüge. De jure willensfrei ist jemand, wenn er in der Lage ist, „auf die eigenen relevanten Handlungsabsichten (Willensbestimmungen) kritisch zu reflektieren“ (150). Das kann sich etwa darin manifestieren, dass er wünscht, er hätte bestimmte handlungsleitende Absichten nicht – also in einer „Volition zweiter Stufe“ im Sinne Harry Frankfurts. Freiheit de jure impliziert dabei keine Freiheit de facto: Frankfurts „widerwillig Drogenabhängiger“, der von seiner Sucht nicht lassen kann, obwohl er sich das wünscht, ist faktisch nicht frei, aber trotzdem de jure moralisch verantwortlich. „Er hat – moralisch gesehen – Pech, daß er drogenabhängig und gleichzeitig moralisch einsichtsfähig ist“ (151). Diese Konzeption „moralischen Pechs“ bietet sicherlich Stoff für substantielle und kontroverse Diskussionen, ebenso wie Beyers interessanter Versuch, die Simulationstheorie im Anschluss an Husserls Konzept des „Nachverstehens“ auch auf das Verstehen von Subjekten als moralischen Personen zu übertragen (168 f.). Alles in allem ist es Beyer gelungen, eine beeindruckende Menge von Themen und Problemen in einem stringenten Argumentationszusammenhang zu verbinden und zu Lösungsvorschlägen zu gelangen, die in der einschlägigen philosophischen Diskussion zweifellos Beachtung verdienen. Uwe MEYER Universität Osnabrück 1. Frankfurt/M.: Suhrkamp, 61997. 2. Vgl. etwa dessen Selbstbewußtsein und Selbsterkenntnis. Stuttgart: Reclam, 1991.

289

heiten‘ tatsächlich Wahrheiten sind.“ Das

auch auf William James’ und John Hicks Aus-

geschrieben […], die mit den Wörtern ‚wahr‘ und ‚falsch‘ schon lange nichts mehr anfangen

LISTE EINGELANGTER BÜCHER (Jänner – Dezember 2007) R. Barthes: Wie zusammen leben. edition suhrkamp 2402. Frankfurt/Main: Suhrkamp, 2007, 283 S. D. G. Camhy / R. Born (Hg.): Encouraging Philosophical Thinking. Proceedings of the International Conference on Philosophy for Children [= Conceptus-Studien 17]. Sankt Augustin: Academia, o. J., 149 S. P. Engel: Va savoir! De la connaissance en général. Paris: Hermann, 2007, 256 pp. E.-M. Engels: Charles Darwin. [Beck’sche Reihe 575]. München: Beck, 2007, 256 S. H.-P. Leeb: Sachverhalte und Extensionalität in der freien Logik [= ProPhil – Projekte zur Philosophie 7]. Sankt Augustin: Academia, 2006, 177 S. S. Loewe: Lebensbeschreibung des ehemaligen Salzburger Philosophieprofessors Johann Heinrich Loewe. Dargestellt anhand von Briefen von seiner Tochter. Aus dem Nachlaß von Professor Eduard Winter hrsg. v. Edgar Morscher u. Otto Neumaier [= Beiträge zur Bolzano-Forschung 19]. Sankt Augustin: Academia, 2005, 125 S. Th. Mormann: Bertrand Russell [Beck’sche Reihe 560]. München: Beck, 2007, 180 S. E. T. Olson: What Are We? A Study in Personal Ontology. New York, Oxford University Press, 2007, ix + 250 pp. J. Padilla Gálvez (Hg.): El laberinto del lenguaje. Ludwig Wittgenstein y la filosofía analítica/The Labyrinth of Language. Ludwig Wittgenstein and the Analytic Philosophy. Cuenca: Ediciones de la Universidad de Castilla-La Mancha, 2007, 188 pp. A. Rami / H. Wansing (Hg.): Referenz und Realität. Paderborn: Mentis, 2007, 287 S. H.-J. Simm (Hg., in Verbindung mit J. Assmann, U. Beck, K. Berger, M. von Brück, W. Frühwald, C. Levin, M. Mulsow, A. Neuwirth, P. Schäfer und H. Schmidt-Glintzer): Die Religionen der Welt. Ein Almanach zur Eröffnung des Verlags der Weltreligionen. Frankfurt/Main–Leipzig: Verlag der Weltreligionen, 2007, 415 S. E. Sosa: A Virtue Epistemology. Apt Belief and Reflective Knowledge, Volume I. Oxford: Clarendon, 2007, xiii + 149 pp. K. R. Stueber: Rediscovering Empathy. Ageny, Folk Psychology, and the Human Sciences. Cambridge, Mass.: MIT Press, 2006, xi + 276 pp. P. van Inwagen / D. Zimmerman (Hg.): Persons. Human and Divine. Oxford: Clarendon, 2007, ix + 380 pp. D. Zimmerman (Hg.): Oxford Studies in Metaphysics. Volume 3. Oxford: Oxford University Press, 2007, x + 299 pp.

THOMAS-MANN-STUDIEN „W as war das Leben? Man wusste es nicht!“ Thomas Mann und die Wissenschaften vom Menschen. Die Davoser Literaturtage 2006 Herausgegeben von Thomas Sprecher 2007. Etwa 280 Seiten Ln etwa e 59.ISBN 978-3-465-03553-4 Thomas-Mann-Studien Band 39

Vom weltläufigen Erzählen 50 Jahre Thomas-Mann-Archiv. Symposion Zürich 2006 Herausgegeben von Manfred Papst und Thomas Sprecher 2007. Etwa 196 Seiten Ln etwa e 49.ISBN 978-3-465-03548-0 Thomas-Mann-Studien Band 38

Vom Nachruhm Beiträge zur Lübecker Festwoche aus Anlass des 50. Todesjahres von Thomas Mann Herausgegeben von Ruprecht Wimmer und Hans Wißkirchen 2007. 278 Seiten mit zahlreichen Abbildungen. Ln e 49.ISBN 978-3-465-03541-1 Thomas-Mann-Studien Band 37

THOMAS MANN Briefe an Jonas Lesser und Siegfried Trebitsch 1939–1954 Herausgegeben von Franz Zeder 2006. 234 Seiten. Ln e 59.ISBN 978-3-465-03500-8 Thomas-Mann-Studien Band 36 Im Geiste der Genauigkeit Das Thomas-Mann-Archiv der ETH Zürich 1956–2006 Herausgegeben von Thomas Sprecher 2006. 576 Seiten mit 76 z.T. vierfarbigen Abbildungen Ln e 74.ISBN 978-3-465-03498-8 Thomas-Mann-Studien Band 35

Bitte fordern sie den aktuellen Prospekt an: Verlag Vittorio Klostermann Postfach 90 06 01 D–60446 Frankfurt am Main E-Mail: [email protected]

VITTORIO KLOSTERMANN

r o d o p i [email protected]–www.rodopi.nl

The Courage of Doing Philosophy Essays Presented to Leszek Nowak Edited by Jerzy Brzeziński, Andrzej Klawiter, Theo A.F. Kuipers, Krzysztof Łastowski, Katarzyna Paprzycka & Piotr Przybysz

In Recent years, the problem if idealization has been one of the central issues discussed in philosophy of science. This volume gathers original essays written by well-known philosophers. The papers address the method of idealization and its applications in science as well as ontological and epistemological problems that have arisen. Among the questions addressed are: What is the logical form of idealizational statements and how should they be interpreted? Is the possible worlds semantics useful in understanding idealization? What is the relation between idealization and truth? The volume is a celebration of Leszek Nowak’s sixtieth birthday.

Amsterdam/New York, NY, 2007 471 pp. Bound € 94 / US$ 141 ISBN: 9789042023369 USA/Canada: 295 North Michigan Avenue - Suite 1B, Kenilworth, NJ 07033, USA. Call Toll-free (US only): 1-800-225-3998 All other countries: Tijnmuiden 7, 1046 AK Amsterdam, The Netherlands Tel. +31-20-611 48 21 Fax +31-20-447 29 79 Please note that the exchange rate is subject to fluctuations

B4821.indd 1

3/10/08 9:32:23 AM

r o d o p i [email protected]–www.rodopi.nl Amsterdam/New York, NY, 2007 VI-380 pp. (Currents of Encounter 33) Bound € 78 / US$ 109 ISBN: 9789042022317

Probing the Depths of Evil and Good Multireligious Views and Case Studies Edited by Jerald D. Gort, Henry Jansen and Hendrik M. Vroom

In the few years since the attack on the World Trade Center on September 11, 2001, evil has become a central theme in the media and human consciousness: the evil of terrorism, the evil of secular culture, concern for poverty, and climate change .... Yet different cultures and religious traditions have different ideas of what evil is and what its root causes are. Although there is no massive clash of cultures, many disagreements and also conflicts in the world arise from the deep differences in views of evil. This volume explores religious views of evil. Scholars from different religions and from various parts of the world describe how people probe the depths of evil—and by necessity that of good—from their own background in various worldviews. In their explorations, almost all address the need to go beyond morality, and beyond legalistic definitions of evil and of good. They point to the radical depths of evil in the world and in human society and reinforce our intuition that there is no easy solution. But if we can gain a better understanding of what people from other worldview traditions and cultures consider evil, we are that much closer to a more peaceful world.

USA/Canada: 295 North Michigan Avenue - Suite 1B, Kenilworth, NJ 07033, USA. Call Toll-free (US only): 1-800-225-3998 All other countries: Tijnmuiden 7, 1046 AK Amsterdam, The Netherlands Tel. +31-20-611 48 21 Fax +31-20-447 29 79 Please note that the exchange rate is subject to fluctuations

B4729.indd 1

12/11/07 6:03:48 PM

www.rodopi.nl

Grund- und Methodenfragen in Fichtes Spätwerk Beiträge zum Fünften Internationalen Fichte-Kongreß »Johann Gottlieb Fichte. Das Spätwerk (1810–1814) und das Lebenswerk« in München vom 14. bis 21. Oktober 2003. Teil IV. Herausgegeben von Günter Zöller / Hans Georg von Manz Amsterdam/New York, NY, 2007 XIII-276 pp. (Fichte-Studien 31) Paper € 58 / US$ 81 ISBN: 9789042022935 ISBN Vol. 1-5: 9789042020450 Inhalt Vorwort Siglenverzeichnis Daniel BREAZEALE: »Der Blitz der Einsicht« and »der Akt der Evidenz«. A Theme from Fichte’s Berlin Introductions to Philosophy Jürgen STAHL: Von der Form der Anschauung zur Anschauung der Form. Zu Fichtes Verständnis des Formbegriffs Albert MUES: Die Position der Anschauung im Wissen oder Die Position der Anschauung in der Welt. Der Unsinn der Subjektphilosophie Christoph ASMUTH: Transzendentalphilosophie oder absolute Metaphysik? Grundsätzliche Fragen an Fichtes Spätphilosophie Marek J. SIEMEK: Unendlichkeit und Schranke. Zum Fichteschen Entwurf einer transzendentalen Ontologie des Wissens Tom ROCKMORE: On Fichte and Idealism Sabine AMMON: Realismus oder Idealismus? – Irrealismus! Akira OMINE: Der Begriff des Übersinnlichen in der Philosophie Fichtes Roderich BARTH: Wahrheit als Sein von Einheit. Die gewißheitstheoretische Reformulierung des absoluten Wahrheitsbegriffs in Fichtes Phänomenologie von 1804-II Arkadij LUKJANOW: Auf der Suche nach einer neuen Theorie des Absoluten. Die Idee der Synthesis in den späteren Systemen von Fichte und Schelling« Lu De VOS: Der Gedanke des Lebens in den späten Schriften Fichtes Michael GERTEN: Geistige Blindheit und der Sinn für Philosophie. Das systematische Problem einer Einleitung in Fichtes Wissenschaftslehre Kai GREGOR: »Revolution der Gesinnung« und »Vollendung der Freiheit« – Wesen und Möglichkeit höherer Lebensformen bei Kant und Fichte Violetta L. WAIBEL: Die bildende Kraft des Wissens vom Wissen in der Spätphilosophie Johann Gottlieb Fichtes Elvira GAREEWA: Wissen als ein freies und selbständiges Leben in den »Thatsachen des Bewußtseyns« Franco GILLI: Die Präsenz der ›Populärphilosophie‹ im Spätwerk Fichtes Hans Georg von MANZ: Die Funktion der »Tatsachen des Bewußtseins« im Blick auf die Wissenschaftslehre George di GIOVANNI: Sacramentalizing the World: On Fichte’s Wissenschaftslehre of 1810 Stamatios D. GEROGIORGAKIS: Der Begriff Schema in Fichtes Spätwerk. Seine Unterschiede zum Schemabegriff in Fichtes Frühwerk und seine Einbettung in der philosophischen Tradition vor Kant Jacinto Rivera de ROSALES: Die transzendentale Logik (1812). Ihr systematischer Ort und ihre Bedeutung Alessandro BERTINETTO: Die transzendentale Argumentation in der Transzendentalen Logik Fichtes Marc MAESSCHALCK: Attention et réflexivité dans la Logique de 1812 et la dernière philosophie de Fichte

USA/Canada: 295 North Michigan Avenue - Suite 1B Kenilworth, NJ 07033, USA. Call Toll-free (US only): 1-800-225-3998 All other countries: Tijnmuiden 7, 1046 AK Amsterdam, The Netherlands Tel. +31-20-611 48 21 Fax +31-20-447 29 79 Please note that the exchange rate is subject to fluctuations

B4768.indd 1

2/2/08 6:07:51 PM