The Metaphysical Status of Tractarian Objects 1

Transcription

1 Philosophical Investigations 24:4 October 2001 ISSN The Metaphysical Status of Tractarian Objects 1 Chon Tejedor I The aim of this paper is to resolve an ongoing controversy over the metaphysical status of Tractarian objects, to wit: are the objects posited by Wittgenstein in the Tractatus material or phenomenal? 2 This debate continues to occupy the foreground of the literature about the Tractatus and its ramifications are complex and far reaching. I propose to establish that Tractarian objects can be neither material nor phenomenal by putting forward two key arguments. Firstly, I will show that material and phenomenal particulars do not satisfy the constraints of simplicity, unalterability and subsistence laid down by Wittgenstein for objects. Secondly, I will demonstrate that the debate as to whether simples are material or phenomenal is, in any case, misguided from the start, because it relies on the mistaken assumption that Tractarian objects can be perceived. As we will see below, rejecting the claim that objects are perceivable will have a significant impact on our understanding of Wittgenstein's views on the connection between language and reality and on his remarks on solipsism. A novel aspect of the view I advance is that it doesn't rely on the argument which has hitherto been used to subvert the debate on the metaphysical status of simples. Proponents of this argument claim that it makes no sense to ask whether Tractarian objects are material or phenomenal, for the reason that Wittgenstein's notion of an object is a purely formal, logical one. This view is explicitly advanced by Ishiguro and is also implicit in the accounts of objects presented by 1. I am indebted to Malcolm Budd and Jane Heal, whose comments were crucial to the development of this paper. 2. I will, throughout this paper, use several expressions to refer to the `simple', `unalterable' and `subsistent' objects posited by Wittgenstein in the Tractatus: `(Tractarian) objects'; `simple objects'; `(Tractarian) simples'; `Wittgenstein's objects'. `the objects of the Tractatus'; etc., 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA.

2 286 Philosophical Investigations McGuinness and Sullivan. 3 I shall refer to this composite argument (advanced by Ishiguro, McGuinness and Sullivan) to the effect that the Tractarian notion of an object is a purely formal, logical one, as `the Argument for Formal Objects'. One of my subsidiary aims is therefore to show that the Argument for Formal Objects has had a distorting effect on the greater debate about the metaphysical status of simples. Indeed, the prevailing assumption has been that, in order to consider this debate misguided, one must agree with some version or other of the Argument for Formal Objects. The latter Argument has been regarded as the only plausible reason for questioning the validity of attempts to ascertain whether Tractarian objects are material or phenomenal. As a result, commentators wishing to engage in the debate on the metaphysical status of simples have relied on an explicit or implicit rejection of the Argument for Formal Objects, without considering that there might be other reasons why the debate is misguided. Part of the aim of this paper is to show that, even if the Argument for Formal Objects was mistaken and needed to be dismissed, it would remain the case that Tractarian objects are neither material nor phenomenal. My conclusion that simples cannot be ascribed a metaphysical status, understood in this sense, does not rely on the validity of the Argument for Formal Objects and would not be affected if this Argument was shown to misrepresent Wittgenstein's thinking. II The debate about the metaphysical status of Tractarian objects has generated a wide variety of differing views. One common denominator can, however, be extracted from these contrasting interpretations: participants in this debate about the metaphysics of simples begin with the assumption that Wittgenstein adopts a particular position on the realism-phenomenalism dispute, and then argue that Tractarian objects are the simplest units of the metaphysical type that is, in their view, favoured by Wittgenstein. 3. See Ishiguro (1969) p. 47, McGuinness (1981) and Sullivan (1996). Ishiguro futher argues that since it makes little sense to speak of `perceiving' something which is purely formal, Tractarian objects must be unperceivable ± Ishiguro (1969) p. 29.

3 Chon Tejedor 287 Thus, for instance, the Hintikkas argue that Wittgenstein espouses sense-data phenomenalism. In their view, the Tractatus holds that, when we perceive the world, we only ever directly perceive our own, strongly private sense-impressions. These sense-impressions form an impenetrable veil of sense-data accessible only to the person doing the experiencing. Tractarian objects are then presented as the simplest units of strongly private sense-impressions there can be, such as sense-data points in the visual field. 4 In contrast, Cook argues that Wittgenstein is a neutral monist who holds that it makes no sense to speak of there being two distinct substances in the world ± one mental and one physical. Only one substance exists, in Cook's understanding of the Tractatus: `pure experience'. This should not be taken to mean that everything in the world is `mental', however: according to Cook's interpretation, the dualist view that mental substance is to be contrasted with physical substance is mistaken and should be abandoned. Although everything in the world is experience ± and thus `phenomenal' in nature ± this is true only in the sense that everything, including other people's mental events, is directly perceivable. The traditional distinction between so-called mental events and physical ones is only a manifestation of the way in which we conceptually organise our experience: it is just a sorting out of our experience into two different conceptual categories. It is, in other words, a difference in modes of experience, not a difference in substance. As a result, Cook argues that Tractarian objects are the simplest possible units of `pure experience' there could be, such as weakly phenomenal points in the visual field. 5 These points are phenomenal, but in a weaker sense than that of the sense-data posited by the Hintikkas: they are not strongly private, since the substance that they make up can be directly perceived by all. Finally, realist interpretations of Tractarian objects hold that, according to Wittgenstein, material things exist and are directly perceivable. According to this view, Tractarian objects are the simplest type of material particulars there could be, such as material points or points of mass. Such a view is, for instance, advanced by Griffin and Pitcher, although their accounts differ somewhat in their details See Hintikka and Hintikka (1986), especially ch See Cook (1994) 6. See Griffin (1964) pp. 4±5 and ch. 11 and Pitcher (1964) ch. 5. This list of possible interpretations of the metaphysical status of objects is not meant to be exhaustive. The

4 288 Philosophical Investigations On the face of it, the textual evidence on this issue is extraordinarily ambiguous. Indeed, both the Notebooks and the Tractatus contain various apparently contradictory remarks on the metaphysical status of objects. On the one hand, Wittgenstein tells us in the Notebooks that: As examples of the simple I always think of points of the visual field (NB ) This comment could be taken to suggest that he thinks of objects in terms of phenomenal qualitied particulars. However, Wittgenstein also tells us in the Notebooks that: The division of the body into material points, as we have it in physics, is nothing more than analysis into simple components (NB ) which would appear to suggest that he thinks of objects in terms of physical points of mass. In the Tractatus too he speaks both of phenomenal points and of physical points of mass: A speck in the visual field, though it need not be red, must have some colour: it is, so to speak, surrounded by colour space. (TLP ) The laws of physics, with all their logical apparatus, still speak, however indirectly, about the objects of the world. (TLP ) We ought not to forget that any description of the world by means of mechanics will be of the completely general kind. For example, it will never mention particular point-masses: it will only talk about any point-masses whatsoever. (TLP ) The possibility of describing the world by means of Newtonian mechanics tells us nothing about the world. (TLP 6.342) The first of these entries would appear to suggest that Wittgenstein thinks of objects in terms of simple phenomenal points. In contrast, the latter three of these entries could, together, be taken to mean that he thinks of objects in terms of particular mass-points: the fact that Newtonian mechanics cannot say anything about the world because it does not speak of particular points of mass (but just of points of mass expressions `material points' and `mass-points' are used interchangeably in the literature ± see Griffin (1964), Pitcher (1964), Hintikka and Hintikka (1986), Cook (1994), and Pears (1987), amongst others.

5 Chon Tejedor 289 in general) could be taken to imply that Tractarian objects must be particular points of mass. Given the ambiguity of the textual evidence, Pears concludes that Wittgenstein may have believed both physical and phenomenal particulars to be plausible candidates for Tractarian objects. 7 I propose to show, on the basis of the textual evidence, that this cannot be correct. III In my view, it can be conclusively demonstrated that Wittgenstein's objects are neither material nor phenomenal. Furthermore, this can be done without appealing to the Argument for Formal Objects, by simply reviewing the textual evidence available and the way in which the Tractarian system operates. To this effect, let us give a brief overview of the key features of the Tractarian system. According to Wittgenstein, Tractarian objects concatenate to produce states of affairs (TLP 2.01 and TLP ). States of affairs are made up exclusively of Tractarian objects (TLP 2.03): they are the most elementary arrangements of objects there could be. More complex possible arrangements are the result of combining several existing and non-existing states of affairs (TLP 2.04 and TLP 2.06). I will use the expression `states of affairs' to refer to elementary concatenations of Tractarian objects, and the expression `(possible) situations (of the world)' to refer to combinations of more than one state of affairs. 8 The term `(possible) states (of the world)' will cover both states of affairs and possible situations of the world. Possible states of the world must both be capable of obtaining and be capable of failing to obtain: they are possible, and not therefore necessarily obtaining. Those possible states of the world which do obtain (or `exist' as Wittgenstein puts it) are called `facts' (TLP 2). I will use the expression `elementary facts' to refer to states of affairs which obtain, and the expression `non-elementary facts' 7. See Pears (1987) pp. 89± Wittgenstein doesn't explicitly define `possible situations of the world' in this way, but this notion is clearly the possible counterpart to a `Sachverhalt' or elementary fact. See the footnote below for a reference to `Sachverhalt'.

6 290 Philosophical Investigations to refer to possible situations that actually obtain. 9 The general term `facts' will cover both elementary and non-elementary facts. States of affairs differ from more complex possible situations of the world in that the former must be logically independent from each other, whereas the latter need not be (TLP and TLP ). To say that states of affairs are logically independent from each other is to say that the obtaining or non-obtaining of a given state of affairs cannot entail the obtaining or non-obtaining of another one (TLP 2.061). According to the Tractatus, all propositions must, in order to be senseful, represent a possible state of the world. Elementary propositions represent states of affairs: they assert that a given state of affairs obtains (TLP 4.21). Elementary propositions consist exclusively of Tractarian names (TLP 4.22) which are simple in the sense of not being analysable into other, simpler Tractarian names (TLP 3.325). Simple names designate the simple objects that make up the represented state of affairs (TLP 4.22 together with TLP 4.24 and TLP 3.22). Nonelementary propositions are produced by applying logical operations to elementary propositions (TLP 5.3) and represent more complex possible situations of the world (TLP 4.031). 10 Elementary propositions differ from non-elementary ones in that the former must be logically independent from each other, whereas the latter need not be (TLP 4.211, TLP and TLP 5.124±5.1241). Elementary propositions must be logically independent from each other in that the truth-value of one of them cannot determine the truth-value of another (TLP 4.211, TLP ). Thus, the simple names that make them up must meet the logical independence constraint: they must be capable of combining with each other to produce elementary propositions which are logically independent from each other. Wittgenstein lays down four key constraints for Tractarian objects. Objects must be `simple' in the sense of not being made up of other, even simpler objects (TLP 2.02 together with TLP ). 11 They 9. In his letter to Russel from Cassino ( ), Wittgenstein explains that a `Tatsache' is an elementary fact and that a `Sachverhalt' is a non-elementary fact. An editorial footnote explains the significance of this letter: `Wittgenstein had sent Russell a copy of the Tractatus by the hand of Keynes, and the following letter is a reply to Russell's queries about the book'. See NB p. 128, ff Wittgenstein simply calls non-elementary propositions `propositions', but I feel that using the term `non-elementary propositions' can be clearer in certain contexts. 11. This interpretation of the simplicity constraint is generally accepted as being correct. For more details on it, see Hintikka and Hintikka (1986) pp. 39±41.

7 Chon Tejedor 291 must be `unalterable' in the sense of being unchanging (TLP ). 12 Objects must also be `subsistent' (TLP ) in that they must be necessary: objects must be the ultimate constituents of all possible worlds. 13 Finally, objects must meet the logical independence constraint, in that they must form states of affairs which are logically independent from each other (TLP 2.01 together with TLP 2.061). Wittgenstein argues that, if objects did not satisfy these constraints, senseful language would not be possible (TLP 2±2.022). IV As was shown above, commentators engaging in the metaphysical debate about Tractarian objects have attempted to answer the question of whether objects are material or phenomenal. Their strategy has consisted in arguing that Tractarian objects are the simplest particulars belonging to the metaphysical type favoured, in their view, by Wittgenstein. Simples have thus been presented as being either material points or phenomenal ones (in some specific sense of `phenomenal'). In this section, I will show that neither material nor phenomenal points (in any of the sense of `phenomenal') are in fact candidates for objects. The reason for this is that neither satisfy the constraints laid down by Wittgenstein for simples. Material points or points of mass fail to meet the necessity constraint. Indeed, even if it was the case that all possible worlds contained some matter (e.g. even if the property of being material was instantiated in all possible worlds) there is no reason why this particular mass-point, located in this place at this particular time, should be the constituent of all possible worlds. After all, there could, 12. Again, this interpretation of the unalterability constraint is generally accepted as being correct. See amongst others, Pears (1987) p. 108, McGuinness (1981) p. 62, Hacker (1986) p. 33 and p. 65±72, and Kenny (1973) pp. 72± In Wittgenstein's terminology, a possible world is one which can possibly be represented in language and thought (see TLP 3.001, TLP 3.02, TLP 3.03, TLP 3.032, and TLP 5.61). This interpretation of the subsistence constraint is questioned by the Hintikkas: See Hintikka and Hintikka (1986) pp. 48±49. and pp. 55±56. It is possible to show, however, that their account of the subsistence contraint is mistaken (see my paper entitled `Sense and Simplicity: Wittgenstein's Argument for Simple Objects'). The view that objects are subsistent or necessary in the sense of being the ultimate constituents of all possible worlds is indeed accepted by most commentators. See, amongst others, Pears (1987) p. 108, McGuinness (1981) p. 62, Hacker (1986) p. 33 and pp. 65±72, and Kenny (1973) pp. 72±80.

8 292 Philosophical Investigations in a different possible world, be a vacuum in this location of space and time. Particular mass-points cannot therefore be said to be necessary. Similarly, material points fail to meet the unalterability constraint. For any given point of mass can conceivably be destroyed and replaced by a vacuum. The fundamental reason why points of mass fail to satisfy the necessity and unalterability constraints is that they are not simple in the Tractatus' sense. A mass-point consists in the possible Tractarian fact (i.e. in the possible composite arrangement) that a specific property (i.e. that of having a mass or that of being material) is instantiated at a particular location in space, at a particular time. 14 It is precisely because mass-points are composite, and not simple, that they fail to be necessary and unalterable. For, even if the property of having a mass and the spatial and temporal locations belonged to all possible worlds, there is no reason why this property should be instantiated at precisely this spatial location, at this particular time, in all possible worlds. The view that composites (i.e. complex arrangements or configurations of simples) are neither necessary nor unalterable is put forward by Wittgenstein in the following entries: Objects are simple. (TLP 2.02) Objects, the unalterable, and the subsistent are one and the same. (TLP ) Objects are what is unalterable and subsistent; their configuration is what is changing and unstable. (TLP ) The alterability and contingency of material points is intimately connected with their lack of simplicity. A point of mass is not a simple object, but consists in the possible fact (i.e. in the composite and thus alterable and contingent arrangement) that the property of having a mass is instantiated at a particular spatial location at a particular time. The same is true of phenomenal points (whether the term `phenomenal' is understood in its stronger or in its weaker sense). Phenomenal points are neither necessary nor unalterable: this particular red point (situated at this spatial location at time t) does not belong to all possible worlds since, in a different possible world, there 14. I should not be taken to imply here that the property of having a mass, the spatial location or the point in time are themselves simple objects. It could be that they are, according to Wittgenstein, complex possible facts which can, in turn, be analysed further.

9 Chon Tejedor 293 could be a green point at this spatial location, at time t; similarly, the point which is red at t could turn green at t 1. The reason for this is again that phenomenal points are not simple in the Tractarian sense: they are composite arrangements consisting in the possible fact that a specific phenomenal property (e.g. that of redness) is instantiated at a particular spatial location at a particular time. As a result, even if the property of redness belonged to all possible worlds, there is no reason why redness should be instantiated precisely at this location in space, at this particular time, in all possible worlds. This view is corroborated by the following entry: A speck in the visual field, though it need not be red, must have some colour: it is, so to speak, surrounded by colour space. (TLP ) [My italics.] In addition, some phenomenal points fail to meet the logical independence constraint. For, if it is the case that there is a red point at this location of space and time, then it cannot be the case that there is a green point at the same location of space, at the same time. Coloured phenomenal points cannot therefore be Tractarian objects, since they are unable to produce logically independent states of affairs. Wittgenstein was well aware of this when he wrote the Tractatus: For example, the simultaneous presence of two colours at the same place in the visual field is impossible, in fact logically impossible, since it is ruled out by the logical structure of colour. [...] (It is clear that the logical product of two elementary propositions can neither be a tautology nor a contradiction. The statement that a point in the visual field has two different colours at the same time is a contradiction.) (TLP ) 15 This section has therefore shown, without relying on the Argument for Formal Objects, that neither points of mass nor phenomenal points satisfy the constraints laid down by Wittgenstein for simples. Since material and phenomenal points are the simplest types of qualitied particulars there could be, Tractarian objects can be neither material nor phenomenal qualitied particulars. 15. When Wittgenstein wrote Tractatus, he was working on the assumption that colour properties would, one day, be shown to decompose into Tractarian objects which did satisfy his constraints: `If statements of degree were analysable ± as I used to think ± we could explain this contradiction by saying that the colour R [red] contains all degrees of B [blue] and that the colour B contains all degrees of B and none of R.' [My italics] (SRLF pp. 35±36). It is only later that Wittgenstein came to realise that this assumption was mistaken. See SRLF.

10 294 Philosophical Investigations Before we turn to the next section, it is worth making two final points. The first concerns those entries from the Tractatus and the Notebooks in which Wittgenstein explicitly mentions phenomenal and material points (see again NB , NB , TLP , TLP , TLP , and TLP ± cited above). Given what has just been established, it is clear that Wittgenstein cannot be regarded as implying, in these entries, that the phenomenal and material points to which he refers are examples of genuine simples. Instead, his point must be that, whilst phenomenal and material points are not genuine objects, they are the closest things to simple objects that he could think of. For Wittgenstein did not know any instances of simples when he wrote the Tractatus. No examples of objects are indeed given in this text. Furthermore, when, years after the publication of the Tractatus, Norman Malcolm asked Wittgenstein what, when he wrote the Tractatus, he would have regarded as an example of an object, Wittgenstein replied that, at the time, he thought: that he was a logician; and that it was not his business, as a logician, to try to decide whether this thing or that was a simple thing or a complex thing. 16 The view that neither Wittgenstein nor we (ordinary language users) know any instances of genuine objects also emerges in the Notebooks: Even though we have no acquaintance with simple objects we do know complex objects by acquaintance, we know by acquaintance that they are complex. (NB ) [My italics in the first instance.] This view is further supported by the Tractatus' remarks to the effect that we (including Wittgenstein) use ordinary language without knowing what takes place at the ultimate level of analysis ± which is, of course, where simple names and the objects which constitute their meanings emerge: Elementary propositions consist of names. Since, however, we are unable to give the number of names with different meanings, we are also unable to give the composition of elementary propositions. (TLP 5.55) Man possesses the ability to construct languages capable of expressing every sense, without having any idea how each word has meaning or what its meaning is ± just as people speak without knowing how the individual sounds are produced. (TLP 4.002) [My italics] 16. See Malcolm (1984) p. 86.

11 Chon Tejedor 295 In fact, in the Notebooks, Wittgenstein argues that it is impossible to know any instances of simple names (or, therefore, of their meanings ± the objects) without having carried out first a complete analysis of language; and he admits to never having carried out such an analysis: My difficulty surely consists in this: In all the propositions that occur to me there occur names, which, however, must disappear on further analysis. I know that such a further analysis is possible, but I am unable to carry it out completely. In spite of this, I certainly seem to know that if the analysis were completely carried out, its result would have to be a proposition which once more contained names, relations, etc. In brief it looks as if in this way I knew a form without being acquainted with any single example of it. (NB ) The second point which needs to be made before moving to the next section concerns the Tractatus' remarks on what it is to know an object: If I know an object I also know all its possible occurrences in states of affairs. (TLP ) If I am to know an object, though I need not know its external properties, I must know all its internal properties. (TLP ) For, whenever Wittgenstein speaks of knowing objects, he uses the conditional form (`If I know an object...'; `If I am to know an object...'). This renders these remarks fully compatible with the view that Wittgenstein didn't actually know any instances of objects when he wrote the Tractatus. Wittgenstein could here simply be expressing the view that, were one in the position to gain knowledge of objects (i.e. had one carried out a full analysis of language), then one would find that knowing an object involves knowing both `all of its possible occurrences in states of affairs' and also knowing `all its internal properties'. V The previous section has established that neither material nor phenomenal points are suitable candidates for simples, in that neither satisfy the constraints laid down by Wittgenstein for objects. In this section, I will show further that the debate about the metaphysical status of simples is, in any case, misguided from the outset. For, as noted above, this debate relies on the assumption that Tractarian

12 296 Philosophical Investigations simples can, in some sense, be perceived. 17 It is, however, possible conclusively to establish that Tractarian objects are neither directly nor indirectly perceivable. As I will now show, this can be done without relying on the Argument for Formal Objects, by focusing on the issue of logical independence. Before we examine the logical independence constraint, it is worth recalling once more the Tractatus' remarks on how we come to `know' simples (TLP and TLP ± quoted above). For it is clear that these remarks do not entail, in any obvious way, that knowing a simple consists in perceiving it. Moreover, the view that objects are given to us in perception is put into question by the following entry: If objects are given, then at the same time we are given all objects. (TLP 5.524) It is indeed hard to see what Wittgenstein could intend by this remark if he understood `given' as meaning `given in perception'. For, as Sullivan argues, there is no obvious reason why being perceptually acquainted with one object should entail that we are acquainted ± let alone perceptually acquainted ± with all other objects. 18 TLP , TLP and TLP thus show, at the very least, that Wittgenstein need not be understood as suggesting that identifying, knowing or being in any way acquainted with an object consists in perceiving it. However, they do not, in themselves, provide us with conclusive reasons for arguing that objects cannot be perceived ± however persuasive Sullivan's point may be. For it may just be possible to put forward an interpretation of these entries which makes them compatible with the claim that objects are perceivable and that it is precisely by means of perception that we come to identify and know them. The idea that knowing or identifying an object cannot consist in perceiving it (either directly or indirectly) can, however, be conclusively established by focusing on Wittgenstein's notion of logical independence. For, as noted above, Tractarian objects and simple names must be capable of producing ± respectively ± states of affairs and elementary propositions which satisfy the logical independence 17. This assumption is also made by Hacker and plays a crucial role in his account of the Tractarian connection between language and reality. See Hacker (1986) pp. 73± See Sullivan (1996) p. 207.

13 Chon Tejedor 297 constraint (TLP 2.061, TLP 2.062, TLP 4.211, TLP , TLP 5.134). Although this notion of logical independence has often been discussed in the literature, its full impact on the nature of objects has hitherto gone unrecognised. Commentators have thus felt able to argue that the logical independence of states of affairs is compatible with the view that objects can, in some sense, be perceived. 19 I propose to show that, in fact, adherence to this constraint entails that Tractarian objects cannot ± in any sense ± be perceived. In order to show this, let us briefly reflect on the notion of a material property. Material properties are mentioned in TLP , and there is no reason to suppose that Wittgenstein understood by this expression anything other than is traditionally understood by it: material properties are those properties which render particulars qualitied, such as the property of being red, the property of being solid, etc. Indeed, TLP is immediately followed by TLP , which refers to the issue of colour. An additional feature of at least some material properties is that they are those properties which enable us to perceive the world. To perceive an apple is, for instance, to perceive (via the sense of sight) that the apple is this particular colour and/or to perceive (via the sense of smell) that the apple has this particular smell and/or to perceive (via the sense of touch) that it is solid, etc. These material properties (redness, the smell of the apple, solidity, etc.) are those properties which enable us to perceive the apple: without them, there would be no such thing as perceiving the apple. For the sake of clarity, let us call those material properties which render perception possible `perceptual properties'. The reason why it is important to introduce the notion of a perceptual property is that perceptual properties are precisely those which violate the logical independence constraint: 20 if something is 19. Compare, for instance, Pears (1987) pp. 82±83 and pp. 156 and 162: in the former sections, Pears discusses the logical independence constraint; in the latter, he suggests that for objects to be known by me is for them to be perceived by me. Compare also Hacker (1987) pp. 71±72 and pp. 76±80: in the former sections, Hacker discusses the logical independence constraint; in the latter, he advances the view that simple names are connected to simple objects by means of ostension (i.e. by means of a combination of perception and pointing), and that it is also by means of ostension that objects are individuated. Hacker later reiterates the view that objects can be perceived (or `experienced') in p Canfield too attempts to argue that objects can, in some sense, be perceived ± see Canfield (1986) pp. 357± There may be other material properties which also violate this constraint. However, all that I need to establish, in order to secure my point, is that those material properties which render perception possible violate the constraint.

14 298 Philosophical Investigations this colour all over, then it cannot be that other colour all over; if something has this smell throughout, then it cannot have that other smell throughout; if something is solid throughout, then it cannot be liquid or gaseous throughout, etc. Since Tractarian objects must be capable of producing logically independent states of affairs, this means, however, that they can neither possess nor be perceptual properties. For, if they did or were, their combinations would not yield logically independent states of affairs. But this entails, in turn, that objects cannot be directly perceived ± for perceptual properties are precisely those properties which render perception possible. It is because objects neither possess nor are perceptual properties that Wittgenstein writes: In a manner of speaking, objects are colourless. (TLP ) The view that simples cannot be perceived is also corroborated by the following entry: Whatever we see could be other than it is. (TLP 5.634) Although discussions of this latter entry have hitherto restricted themselves to considering its impact on Wittgenstein's views on solipsism, this remark also supports the claim that objects cannot be directly perceived. For it underlines the Tractatus' view that, in order for something to be perceivable (e.g. visible), it must be capable of altering, of being `other than it is'. In other words, in order for something to be perceivable, it must be composite: that which is composite is alterable, that which is simple isn't. Since Tractarian objects are simple, unalterable and subsistent (TLP 2.02 and TLP ), they cannot be directly perceived. The view that simple objects neither possess nor are perceptual properties is reinforced by the following entry: It is only by means of propositions that material properties are represented ± only by the configuration of objects that they are produced. (TLP ) For, in this remark, Wittgenstein explicitly tells us that material properties (including, of course, perceptual ones) do not arise in simples but only in composite arrangements of simples (`configurations of object' being, by definition, composite). These considerations enable us to establish that Tractarian objects cannot be directly perceived. However, it could be argued that,

15 Chon Tejedor 299 although objects are not directly perceivable, they can be perceived indirectly. It might be, for instance, possible to directly perceive states of affairs, and thereby to gain an indirect perceptual acquaintance with simples. This view is, however, even less tenable than the view that objects are directly perceivable. For states of affairs could only be directly perceived if perceptual properties arose in them. However, since perceptual properties violate the logical independence constraint, they cannot arise in states of affairs. It is worth recalling that the logical independence constraint is set up by Wittgenstein precisely in order to avoid the possibility of logically dependent states of affairs (and elementary propositions). In other words, it is as impossible to have a direct perceptual acquaintance with states of affairs as it is directly to perceive objects: This point is red all over at time t cannot be a state of affairs, any more than redness can be a Tractarian object or be possessed by such an object. For, if this point is red all over, then it cannot be the case that this (i.e. the same) point is green all over at time t. The logical independence constraint entails that perceptual properties ± those properties which enable us to perceive the world ± arise neither in objects nor in states of affairs. As a result, neither objects nor states of affairs are directly perceivable. We do not perceive Tractarian objects directly, nor do we perceive them indirectly, by means of a perceptual acquaintance with states of affairs. Perceptual properties arise only at a higher level of complexity: at the level of possible situations of the world. Only at this level is perception possible, according to the Tractatus. Note indeed that TLP allows for the view that those material properties which render perception possible (i.e. perceptual properties) are represented only in non-elementary propositions. Indeed, saying that everything perceivable is composite (TLP ; TLP 5.634) is not the same as saying that everything composite is perceivable. One final objection could be raised at this stage. For it could be argued that it may be possible to individuate or gain knowledge of objects by perceiving those arrangements of them which are complex enough to be perceivable ± namely, possible situations of the world. Since possible situations can be perceived and since they are ultimately made up of objects, a direct perceptual acquaintance with them could, conceivably, yield an indirect (perceptual) knowledge of objects. Another way of putting this would be to say that, since perceptual

16 300 Philosophical Investigations properties ultimately supervene on objects, knowing or being acquainted with an object may consist in perceiving those possible situations in which these properties do arise ± and which the object ultimately helps to produce. This objection contains an element of truth and an element of confusion. It may well be true that perceptual properties supervene on simple objects, according to Wittgenstein. (This is indeed strongly suggested by TLP ; in fact, TLP provides us, precisely because of this, with the beginning of a reason for rejecting the Argument for Formal Objects.) However, the view that perceptual properties supervene on objects in no way entails that objects can be known or identified simply by perceiving the possible situations in which these properties arise. For the complexity of these possible situations obscures their ultimate logical structures, just as the complexity of non-elementary propositions obscures what takes place at the ultimate level of their analysis. According to Wittgenstein, it is the complexity of any given nonelementary proposition which prevents us from seeing that the proposition is ultimately analysable into elementary propositions, which, in turn, are made up of simple names, designating simple objects. Similarly, the complexity of possible situations of the world prevents us from seeing that they are ultimately made up of states of affairs, consisting of simple objects. The view that what takes place at the non-elementary level obscures what takes place at the ultimate level of analysis is, after all, repeatedly advanced by Wittgenstein (see again TLP 5.55; TLP and NB , quoted above). This is precisely why, in NB , Wittgenstein argues that it is impossible to identify simple names or their meanings (i.e. the simple objects designated by these names) before having carried out a complete analysis of language. It is also this impenetrability which allows philosophical mistakes and confusions to arise, according to the Tractatus (see TLP 3.323, TLP and TLP 4.003, amongst others). In summary, then, Tractarian objects are not perceivable. They can be neither directly perceived, nor `perceived' indirectly by virtue of a direct perceptual acquaintance with states of affairs or more complex possible situations of the world. States of affairs are as unperceivable as the simples themselves. And, whilst more complex possible situations are perceivable, a perceptual acquaintance with them does not provide us with a transparent view of simples. This, of course, helps to explain Wittgenstein's claim, in the Notebooks, that

17 Chon Tejedor 301 we have `no acquaintance with simple objects' (NB ): only a full analysis of language could yield such an acquaintance (NB ). Since the debate about the metaphysical status of simples relies on the assumption that objects can, in some way or another, be perceived, we must conclude that the debate has been misguided from the outset. VI In this paper, I have shown that the debate about the metaphysical status of Tractarian objects is misguided. Material and phenomenal points fail to satisfy the constraints laid down by Wittgenstein for simples. Moreover, the debate relies on an assumption which I have shown to be false: the assumption that Tractarian objects can be perceived. One important aspect of my argument is that its conclusions have been reached without recourse to the Argument for Formal Objects. It isn't necessary to espouse the latter in order to defend the view that objects are neither material nor phenomenal. Hence, my view that the metaphysical debate about simples is misguided is not affected if the Argument for Formal Objects is put into question. In addition, this discussion has two further important corollaries which stem from the conclusion that objects cannot be perceived. Firstly: if objects are unperceivable, they cannot be individuated or attached to simple names by means of ostension, as is argued by Hacker. 21 For unperceivable objects cannot be ostensively pointed at. In fact, as Ishiguro argues, it is only possible ostensively to point at an object which is currently being perceived. (If, for instance, I know the meaning of `Pablo Picasso' but have never perceived the man or any images of the man, then I cannot be said to have learnt the meaning of this expression by having had Pablo Picasso ostensively pointed out to me.) 22 Secondly: if objects cannot be perceived, then Wittgenstein's remarks on solipsism cannot be understood as implying that I perceive or could possibly perceive the totality of objects. Thus, those interpretations of the Tractatus' remarks on 21. This view is advanced in Hacker (1986) pp. 73± See Ishiguro (1969) p. 29.

18 302 Philosophical Investigations solipsism which suggest that this is the case ± such as those advanced by Pears and Hacker ± need, at best, to be seriously modified. 23 One final point needs to be made at this stage. My conclusion that the realism/phenomenalism debate about Tractarian objects is misguided does not entail that Wittgenstein holds no metaphysical views whatsoever. It may be, for instance, that debates as to whether Tractarian objects can be conceived in minimalistic terms remain legitimate. Moreover, whilst it is mistaken to discuss whether simple objects are material or phenomenal, it may not be mistaken to discuss whether possible situations of the world (which are complex, changeable and thus perceivable) are material or phenomenal. Whether these are material or phenomenal and whether it even makes sense to ask this question within the Tractarian context, is not however something that can be settled here. All that we know, at this stage, is that material and phenomenal points (which, as was shown above, are complex ± not simple) are mentioned both in the Tractatus and in the Notebooks. This might suggest that Wittgenstein did not, in fact, favour any metaphysical type over another when thinking of nonelementary possible situations of the world. However, in order conclusively to establish this, we would need to examine in detail his earlier views on solipsism. That is a task for another occasion. References Canfield, J. (1986) `A Model Tractatus Language', in The Philosophy of Wittgenstein: `The Early Philosophy ± Language as Picture', vol. 1/15, ed. J. Canfield, pp. 349±367. New York & London: Garland Cook, J. (1994) Wittgenstein's Metaphysics. Cambridge: Cambridge University Press. Griffin, J. (1964) Wittgenstein's Logical Atomism. Oxford: Clarendon Press. Hacker, P. (1986) Insight and Illusion. Themes in the Philosophy of Wittgenstein, rev. ed. Oxford: Oxford University Press. Hintikka, M. & J. (1986) Investigating Wittgenstein. Oxford: Basil Blackwell. 23. See Pears (1987) ch. 7, esp. pp. 156 and 162, Hacker (1986) pp. 73±80 and ch. IV.

19 Ishiguro, H. (1969) `Use and Reference of Names', in Studies in the Philosophy of Wittgenstein, ed. G. Von Wright, pp. 20±50. London: Routledge and Kegan Paul. Kenny, A. (1975) Wittgenstein. London: Penguin Books. Malcolm, N. (1984) Ludwig Wittgenstein: a Memoir, rev. ed. Oxford: Oxford University Press. McGuinness, B. (1981) `The So-Called Realism of Wittgenstein's Tractatus', in Perspectives on the Philosophy of Wittgenstein, ed. I. Block, pp. 60±73. Oxford: Blackwell. Pears, D. (1987) The False Prison, vol. I. Oxford: Clarendon Press. Pitcher, G. (1964) The Philosophy of Wittgenstein. Englewood Cliffs: Prentice Hall Inc. Sullivan, P. (1996) `The `truth' in Solipsism, and Wittgenstein's Rejection of the A Priori', in European Journal of Philosophy, vol. 4, pp. 195±219. Wittgenstein, L. (1961) Notebooks 1914±1916 (NB), ed. G. Von Wright and G. Anscombe, transl. G. Anscombe. Oxford: Basil Blackwell. Wittgenstein, L. (1974) Tractatus Logico-Philosophicus (TLP), transl. D. Pears and B. McGuinness. London: Routledge and Kegan Paul. Wittgenstein, L. (1993) `Some Remarks on Logical Form' (SLRF), in Essays on Wittgenstein's Tractatus, ed. I. Copi and R. Beard, pp Bristol: Thoemmess Press. University College, London Chon Tejedor 303