The Metaphysics of Meaning by Geoffrey Klempner




Geoffrey V. Klempner




D. Phil thesis

University of Oxford

Trinity term 1982





© Geoffrey Klempner 1982, 2016


All rights reserved.









It is all one to me where I begin; for

I shall come back there again in time.









1 Beginnings

2 How is false belief possible?

3 The idealist’s challenge

4 Kant’s ‘refutation of idealism’

5 The refutation of Kant’s transcendental idealism

6 The metaphysic of action

7 Objective idealism

8 Perception and action; and primary qualities

9 The epistemology of the passive observer

10 Psychological explanation

11 Solipsism and self-consciousness

12 Mind and body

13 Thought and language

14 Does thought entail the possession of a language?

15 Concept and object

16 Sense and reference

17 Name and object

18 What are concepts?

19 What is truth?

20 The meaning of ‘realism’

21 The idea of an objective world

22 The refutation of realism (1)

23 The refutation of realism (2)

24 What is knowing a proposition’s truth conditions?

25 God and realism

26 The law of excluded middle

27 Intuitionist logic and mathematics

28 The reality of time; and of other times

29 The antinomy of phenomenalism and realism

30 The metaphysical attitude

31 The rejection of transcendent metaphysics

32 The transcendence of the ego and of truth

33 The source of the illusion


Prefatory note (1982)

Abstract (1982)

Writings referred to in the text


1. Beginnings


1. The metaphysics of meaning is a system of metaphysics. It earns the name ‘metaphysics’ because it articulates a vision of ultimate reality. It constitutes a ‘system’ because the articulation aims to be complete and whole. It reduces neither to a metaphysics of experience nor to a theory of meaning. Yet the negative example provided by each of these false projects yields a partial clue to its meaning: that its orientation is not the concept of experience but rather the concept of meaning; that its project is not the construction of a theory but the working through of a dialectic. In their positive aspect, the examples serve as necessary reminders that to ‘construct a theory’ is merely one choice of project, not the only possibility; and that the orientation of a philosophical project need not consist in the attempt to comprehend the fact of experience but may instead aim to comprehend the fact of meaning. However, these assertions prove meaningless outside the context of the dialectic: If they seem to have been understood then they have been misunderstood. Still, even misunderstanding can serve a useful purpose. There are morals to be drawn from the fact of misunderstanding and lessons to be learned from its eventual correction. But it is in any case impossible to guarantee against misunderstanding; to attempt to give preliminary definitions of ‘theory’, ‘dialectic’, ‘project’, ‘orientation’ or even to explain at this stage what could be meant by the ‘fact’ of experience or of meaning would be futile.


2. The essence of ‘system’ lies wholly within the demands of completeness and wholeness and not in any supposed method of carrying them out. Indeed, metaphysics must repudiate the very idea of ‘method’; either for ‘discovering truths’ or for ‘presenting results’. Metaphysical investigations may discover truths, and works of metaphysics present those truths, in a variety of different ways. But, contrary to surface appearances, these amount to no more than differences of style or tone. Moreover, the style of presentation is nothing other than an idealized version of the style of discovery. What different metaphysical systems have in common is that they are dialectical; their realization in the form of ‘works’ serves essentially to record and communicate a process of metaphysical argument. That the word ‘system’ should irresistibly call to mind geometrical and linear notions of a beginning, middle and ending, of ‘definition’, ‘exposition’, ‘axioms’ and ‘proof’ indicates how far we still lack the ability to think dialectically. It also accounts for the strength of the contemporary prejudice against metaphysics. For one may easily demonstrate the impossibility of a ‘system of metaphysics’ if one understands the concept ‘system’ in the commonly accepted sense (Ch.2/para.14). To see the point of the twin demands of system is to see the point of metaphysics; to understand what it could mean to desire to know the ultimate reality and to attempt to gain that knowledge by rational means. And that entails viewing metaphysics as an open possibility.


3. Metaphysics demands completeness and wholeness because ultimate reality is not something of which one could rest content with a partial view. The very uniqueness of metaphysical knowledge, by contrast with other forms of knowledge, its lack of corroborating evidence from any other field of inquiry renders insecure any knowledge of ultimate reality which does not not only know it completely but also in such a way as to integrate all partial perceptions into an interconnected, meaningful whole. For metaphysics must aim at complete clarity, even if it knows that such an ideal is practically unattainable; a ‘metaphysics’ which stops short of attempting to solve all the problems which present themselves in the course of its investigations simply risks reduplicating those very problems for which it claims a ‘solution’ in the form of an ineliminable residue of unanswered questions or unsatisfied intuitions.


4. The deepest paradox of metaphysics lies in the mode of existence of its object. Ultimate reality is not something which exists to be perceived by or through the dialectic, an object with which our minds endeavour to make contact through the medium of metaphysical argument. That ultimate reality presents itself as such an object is the illusion of transcendent metaphysics; a project which entails the self-refuting demand for a ‘system’ of metaphysics in the rejected sense (2/14). We must view the dialectic rather as a means of instruction: One works a dialectic through until one’s knowledge reconstitutes itself under the aspect of ultimate reality; until the desire to know the ultimate reality has been fully expressed, worked through and finally appeased. The effort and pain taken to work the dialectic through shows that one has ‘grasped ultimate reality’. One cannot describe what one has grasped; one can only repeat the dialectic and in so doing help someone else to work it through. The truths of metaphysics cannot be expressed; they are rather something which a certain process of training enables one to make one’s own; a style and a manner of taking reality.


5. The essence of dialectic is to reject and negate. Metaphysics is essentially controversial. Its most simple propositions are not assertions but negations; the controverting of propositions which cannot therefore themselves be ‘propositions of metaphysics’. In this way the motive force of dialectic, the rejected ‘given’, becomes at the same time the gravest question mark against its very possibility: How can metaphysics tolerate any kind of given, when the very notion of being ‘given’ something puts the determination of what is to be given and why it is given outside the scope of metaphysical inquiry? How can such an inquiry fail to be merely arbitrary? How can metaphysics justify its rejection of the given if it cannot appeal to the criterion of a prior knowledge of ultimate reality? And what kind of meaning can a metaphysical negation possess if the negated proposition is thereby denied a meaning? - Through being obliged to answer these questions the dialectic entails an inquiry into the very possibility of metaphysics.


6. The rejected given is no mere ‘error’ or ‘mistake’ but an illusion; a manner of wrongly taking reality which forces itself upon our consciousness. This ‘metaphysical’ illusion first appears as a seeming opacity which clouds our knowledge of reality and renders the very idea of knowledge problematic We find ourselves confronted by seeming ‘questions’ which aim at a reality ‘beyond’ or ‘behind’ the world disclosed by perception and science, a transcendent reality of ‘metaphysical facts’. The symptoms of the illusion consist not only in the disposition to ask such questions but in the resulting dissatisfaction with the forms of language or else the attempt to imbue those forms with a ‘significance’ which they do not possess. That is how the ‘desire to know the ultimate reality’ (para.4) first shows itself. We must make two observations, First, the objectivity and non-arbitrary nature of metaphysical inquiry presupposes that metaphysical illusion is in some sense universal; that the dialectic speaks to an audience, articulating something which it may be taken for granted that they all share; in just the same way that poetry speaks to the ‘universal’ conditions of human sensibility. Second, the diagnosis and critique of metaphysical illusion amounts to the pursuit of the illusion of transcendent metaphysics itself. The crucial task of establishing the possibility of metaphysics in the face of the rejection of transcendent metaphysics thus becomes the topic of the inquiry rather than merely a condition for its successful completion,


7. The dialectic refutes the transcendent ‘vision of an ultimate reality’. But in that rejection it neither employs a vision of its own with which to combat the rejected vision, nor even possesses any vision which could serve to fill the vacuum left by the removal of illusion. We gain our vision only by working through the dialectic of illusion; a mere consequence of that critique rather than the starting point or criterion which might otherwise have served to guide it. And the vision we gain is none other than that of reality itself, only now taken in the ‘right way’. One might describe this as the determination to keep silent about that which provides the ultimate ground of the dialectic, which yet remains beyond the range of dialectical expression. But this silence is an innocent silence. It contrasts with the silence to which a person who remains under the spell of a metaphysical illusion finds himself reduced. The victim struggles to express what he now realizes cannot be expressed, while still demanding expression. His silence is the silence of paralysis. Here the dialectic threatens to come to an end prematurely, with the victim stubbornly refusing to renounce the illusion. And here too one is tempted to break one’s own innocent silence in order to help the argument along. But to adopt that tactic would amount to an admission of defeat. One should bear in mind that silence can be the right response; that the objectivity of the dialectic does not mean that it should be psychologically impossible for a rational being to fail to follow it through to the end.


8. The progressive expression and elimination of metaphysical illusion necessarily stops short of explaining what the illusion is in itself, of uncovering its source. This is a corollary of the ‘givenness’ of the illusion (5). For example, that a symptom of the illusion consists in a certain ‘misuse of language’ does not mean that the illusion itself, that which causes the symptom, is only ignorance or misperception of the ‘rules of correct usage’. The cause cannot be deduced from its effect. And to frame hypotheses concerning the possible cause is clearly beyond the scope of metaphysics. That metaphysics may legitimately accept this ‘beyond’ while nevertheless claiming to satisfy the demands of wholeness and completeness can be settled only by working through the dialectic itself, by getting to see its point. The dialectic asserts its independent reality on the grounds that ‘it does not matter where the illusion ultimately comes from’.


9. The givenness of illusion (5, 6), the absence of a prior metaphysical vision (7) and the impossibility of uncovering the source of illusion(s) are merely different aspects of the question of the possibility of metaphysical knowledge which constitutes the continuous theme of the dialectic of illusion. None of these problems arises for a transcendent metaphysics, for the very reason that it purports to begin with the discovery of an ultimate reality and then proceeds to elaborate its seeming ‘attributes’. To misperceive or fail to perceive this ‘ultimate reality’ is called ‘illusion’ and nothing more need be said. For that same reason the deepest and most revealing aspect of the thematic question can arise only for a dialectic of illusion: How can the propositions of a system of metaphysics give so much as the appearance of being understood? They are in some sense ‘negative’; but not negations in the ordinary sense, for a negation presupposes the meaningfulness of the proposition negated; and that would entail the meaningfulness of those seeming ‘questions’ which we reject as symptoms of metaphysical illusion (6). Neither can the propositions be said to deny that certain forms of expression have a meaning, for the linguistic manifestations of the illusion have their own non-metaphysical meaning. The denial is directed, we should like to say, only at what those forms are ‘meant’ to mean by a proponent in the grip of the illusion, at a ‘meaning’ which it only seems possible to mean but which is really meaningless. But what kind of meaning is that?


10. That question may be reformulated as the demand that the illusion be adequately expressed. We can identify the illusion only from within its dialectic; through the argumentative context which aims to express it and to which it contributes the motive force. The meaning of metaphysical propositions thus consists solely in their functional role in a dialectic. But now the completely circular nature of the demand for adequate expression becomes manifest, A dialectic which adequately expresses its illusion is one for which the problem of the meaningfulness of its propositions no longer arises! We of course want to know why and how that problem ‘no longer arises’. But once again the answer to this question, and indeed the meaning of ‘adequate expression’ can be discovered only by working through the dialectic and coming to see its point.


11. Any attempt at a preliminary description of the illusions which reveal themselves through the dialectic of a metaphysics of meaning is bound to be misunderstood. However, that remark will not deter the anxious reader. My own justification for attempting to ‘anticipate’ what can be expressed only in the dialectic is, to repeat, that even misunderstanding may turn out to serve a useful purpose. We begin with an assertion which conveys very little information: In its most general form, the particular illusion which is the concern of a metaphysics of meaning consists in the seeming ‘belief’ that the proposition, ‘Thought serves to represent reality’ means something other than, or besides, what it actually means. This illusion manifests itself as two illusions, sharing the same general form, but so related as to strongly suggest that they share the same ultimate source; that they are indeed the ‘same illusion’. I shall now try to say what these two illusions are.


12. ‘Mind aims its thoughts at reality.’ — What does that really mean? What must one add to that bare statement in order to obtain a coherent description of this ‘relation’ between mind and reality? What is the peculiar mode of relating to reality which we call ‘thinking’? The questions suggest the following project of transcendent metaphysics: to determine ‘the place of the thinking subject within ultimate reality’. We reject the suggestion: The questions can serve only as clues to some underlying illusion; to something which will motivate a series of incoherent conceptions of mind’s relation to reality: In the course of the dialectic, mind appears first as a bare ‘point of view’ on reality, simply and essentially a constructor of ‘representations’, which passively observes the show of its ‘experience’. Then the ‘representations’ acquire a life of their own independent of the activities and situation of the thinking subject; and their relation to reality becomes a matter of simply being ‘made’ true or ‘made’ false by the obtaining or failure to obtain of the possible states of affairs which they represent. The illusion of representation thus splits apart into twin illusions concerning the nature of the ego and the nature of truth. By pursuing these illusions through their various disguises and subterfuges one reaches a point where one no longer needs to enquire how mind relates to reality; for the relation becomes manifest. And that concludes the dialectic.


13. The illusion of representation provides the metaphysics of meaning with its orientation towards the concept of meaning, as will become apparent in due course. And the demands of system, that the articulation of the metaphysics be complete and whole (2), are met through an adequate expression of the illusion (10) which is at the same time a satisfying explanation of the very possibility of a dialectic of illusion (5). However, a partial perception of the illusory nature of both the ego and truth illusions has been embodied in the idea of a system of a very different kind. The project of describing the ‘form’ of a ‘systematic analysis’ of the complex practical and linguistic knowledge in which possession of a concept consists (Dummett ‘What is a Theory of Meaning?’) takes as its point of departure the incoherence of that conception which would reduce concepts to incommunicable ‘rules’ applied to the ‘show of experience’; concepts are necessarily embodied in a language which can be used to communicate one subject’s thoughts to another (Dummett Frege Philosophy of Language P.638-42). Similarly, the ‘systematic explanation’ of how a speaker’s perception of the functional interrelationship of concepts in a proposition yields a knowledge of its meaning involves the rejection of any seeming ‘explanation’ which appeals to his knowledge of the conditions for that proposition’s truth, conditions whose ‘obtaining’ or failure to ‘obtain’ transcends the speaker’s ability to determine whether or not they ‘obtain’ (Dummett ‘What is a Theory of Meaning? (II)’). I have called Dummett’s insight only partial because the demand for such a ‘theory of meaning’ must, in both cases, be refused. We shall indeed see how this demand results in a distorted account of the two ‘illusions’ which it means to reject. But that involves a three-cornered fight: In addition to pursuing the two illusions we have to fend off the meaning theorist’s erroneous conception of what their rejection entails.


2. How is false belief possible?


14. ‘The difficulty with constructing a system of metaphysics is knowing where to start.’ — That philosophical joke embodies the crucial objection against any attempt at making a work of metaphysics ‘systematic’ in the commonly understood sense (1/2). All that would have constituted the articulation of its metaphysical vision lies buried in the extra-systematic choice of a starting point. in premisses, method and criteria of validity which, within the system, remain necessarily ungrounded. The ‘results’ of working out the system become merely the inevitable outcome of its own beginnings. What really matters is lost in the execution. One might now be tempted to object that ‘what really matters’ cannot be expressed; that its ‘ungrounded’ assumptions can in fact be ‘discharged’ only by an incorrigible and inexpressible metaphysical vision of the ‘ultimate reality’ which the system describes. But that is transcendent metaphysics (1/4). Against the objection that any number of other metaphysicians might intuit different, incompatible starting points the transcendent metaphysician has no defence except his own conviction; in which case no further discussion is possible.


15. By contrast, we only need a question suitable for setting the dialectic in motion. Choosing the ‘right’ starting point reduces to the choice of an interesting and fruitful strategy. But there remain pitfalls. To begin with the questions which most deeply concern us may turn out to block further progress. In a dialectic of the relation between mind and reality the obvious candidates are: ‘What is the self?’, ‘What is truth?’ (the mind, reality), or: ‘What is knowledge?’, ‘What is thought?’, ‘What is action?’ (mind’s relating to reality). However, these questions prove too inflated; their indeterminacy of point, the freedom allowed by a multiplicity of possible approaches only leads to futile battles between rival ways of ‘taking the question’. To acknowledge that there is something right in each alternative, a partial perception of the truth in each competing position, may be the right thing to say but quite useless dialectically. Even the arbitrary, contentious choice of some particular construal of an indeterminate question is better than no choice. And better than that would be a choice of question whose construal was beyond dispute.


16. The questions which so interest us must be postponed until later. Provided that the investigation begins in the right way, a definite angle on the notions of truth, the self, thought, knowledge and action should materialize at the appropriate point in the dialectic; when the multiplicity of approaches has narrowed to just one. For the initial question I instead propose the following. How could there be such a thing as a false belief? One might well react that its ‘point’ was so far from being a question of dispute that it does not seem to have any: If a belief is wrong, if what is believed fails to be the case, then the belief is false. No-one will deny that some beliefs are false. And it seems just obvious how they can be false: Even the most perfectly constructed believers could not help getting tricked by reality occasionally. But of course that is not the point of the question.


17. If having beliefs is being in some manner related to reality then it is a very peculiar kind of relation. For the very same belief ‘relates’ to reality whether it is true or false. Yet surely a true belief is more ‘related’ to reality than a false one. Should we then say that there exist two possible ‘relations’ between a belief and reality? how are those ‘relations’ related to one another? or that true beliefs are indeed ‘related’ to reality but that false beliefs remain ‘unrelated’? or that they ‘relate’ not to reality but to something else? what could that be? — Why indeed should these be regarded as serious questions? They express what might well strike one as a real ‘perplexity’, but that does not render our choice of an initial question itself any less puzzling. But perhaps the following may serve as adequate motivation: Just as a relativist who denies the very existence of objective truth might accuse the sceptic who merely denies that we could ever know the truth of assuming the existence of that truth’ which we ‘fail to know’, so one might in turn accuse the relativist of assuming that it is possible for beliefs to fail to be true, that beliefs retain some kind of objective status, even if only as failed attempts at the impossible. If one attempts to put into question the relation between thought and reality then one cannot avoid being forced to go all the way. — Still, it doesn’t matter if the reader remains unconvinced, The proof of the interest and fruitfulness of an initial question remains in its working out.


18. For the sake of the later development of the dialectic, we shall first consider a reaction which refuses to see false belief as the locus of the ‘real problem’, ‘A belief is called ‘false’ when what is believed to be the case turns out not to be the case; when the content of the belief, the proposition which it expresses, possesses the truth value ‘falsity’. Provided that one can grasp how the nature of a proposition allows for the possibility of either truth or falsehood, nothing more is needed to comprehend the possibility of false belief.’ — The attractiveness of this idea lies in the seemingly ‘direct’ way in which it tackles the question of how language itself ‘works’, simply examining how propositions are constituted provides all that is necessary for grasping how language can serve to express our thoughts concerning reality. But the proposition theory gets things back to front: It turns out that the notion of a false proposition, far from explaining the possibility of false belief, requires an explanation of how false belief is possible in order to account for its own ‘possibility’.


19. How are propositions constituted? What could the enormous and varied range of possible propositional contents and indeed of possible ways of expressing those contents in writing and speech ever have in common? The crucial insight here is to ask instead what function propositions perform. In that way all possible contents and modes of expression will have been gathered together under the same intelligible principle, the ‘Form’ of propositionality itself.


20. Propositions communicate information. In order to perform that function they must be understood. Therefore understanding them must not presuppose possession of that very information: Is that a necessary truth or might things have been otherwise? Let us imagine two gods who, because they ‘know everything’, do not need to communicate information. Instead, they have agreed to associate an arbitrary number with each actual and possible state of affairs. Their ‘conversation’ consists in calling out these numbers to each-other; like the prisoners who, having run out of jokes, referred to them by numbers ‘in order to save time’. The comparison gives the game away; the number language is parasitic upon the kind of language which can be used to communicate information. Even allowing the fiction that the gods could ‘point out’ and number actual states of affairs without resorting to language, there is manifestly no way they could refer to possible states of affairs without describing them, prior to assigning a number. Propositions do not name states of affairs; if language is possible at all there must exist a means of expression whose own constitution determines which possible state of affairs it describes, and consequently its aptness for truth or falsehood, without presupposing prior ‘identification’ of that state of affairs.


21. Mere ‘names’ of states of affairs would acquire their ‘truth’ or ‘falsity’ by our arbitrary choice. Of course, if one thinks of the names as shorthand for some description of those states of affairs then choosing the names is indeed arbitrary but the truth value of the propositions for which they stand is not. But now remove that intermediate ‘description’: The point of this ‘counternecessaryfactual’ is that the denial that propositions name states of affairs is dialectically equivalent to the denial that truth is ‘determined by arbitrary choice’. We construct propositions. Their constitution is indeed a matter of our choice; for we choose both the content and the manner of expressing it. Yet their truth or falsity is not a matter of choice. The proposition itself determines which possible states of affairs it describes, and, once that is fixed, the matter of its truth or falsity is taken out of our hands. How could that be?


22. In the Tractatus Logico-Philosophicus Wittgenstein seemed to have solved the problem: Whatever the medium of expression, a proposition consists in a particular combination of ‘elements’. The world which propositions describe may similarly be seen as one of many possible combinations of ‘objects’. In the ideal case, each combination of propositional elements would be correlated with a unique possible combination of objects. Our ‘arbitrary choice’ determines which elements stand for which objects. The possibilities of combination of elements allowed for by the grammar of the language aim to match the possibilities of combination of each object with those of the element which stands for it (not an arbitrary choice because the grammar must be adequate for the representational purposes of the language). These two relations of ‘standing for’ and ‘matching’ fix the correlation of propositions with states of affairs independently of any further choice on our part: The propositions genuinely serve to ‘represent’ those states of affairs.


23. What are the ‘elements’ of propositions? and what are the ‘objects’ of which reality is composed? — We may pass over the atomistic metaphysic of the Tractatus; that concerns a different problematic from ours (which is not to deny that a dialectical route may be constructed from one to the other). Call each separate word of a given language an ‘element’. Then the ‘object’ which it names is either an object in the ordinary sense or else a suitably construed function, the simplest kind being that from ordinary objects to truth values. The philosophical interest in the project of working out a consistent and complete theory of the functions to be associated with each kind of word is in its possibility and not necessarily in the details of its working out. We do not need that detailed theory to tell us how, provided one is familiar with the grammar and with the meanings of the words of a given sentence, one understands what it means simply by reading off the state of affairs which it ‘represents’.


24. There remains room for detailed and possibly philosophical argument over the degree of respect with which we should treat the words of ordinary language and over the relative illumination provided by different theories which are otherwise equally consistent and complete. But the point about the representational nature of language has already been made. And our concern now is not with the truth which it contains but with the possible illusion which it conceals. The notion of a ‘representation’ seems to contain all that is needed to comprehend the possibility of false belief. However, in that case either the metaphysical import of the question has not been properly grasped, or else the ‘theory of the proposition’ embodies a metaphysical illusion: At a crucial point a lacuna appears in the theory which only a false metaphysic could lead one to suppose was without consequence.


25. Suppose I draw this line: ‘______’ claiming that it represents, non-pictorially, the shape of the North coast of Africa. We can indeed construct ‘rules’ according to which that line ‘represents’ the particular arrangement of points which constitutes the shape of just that particular coastline: Each point in the line will ‘stand’ for a point in the represented shape and ‘match’ its ‘possibilities of combination’. By accident or design, a complex of points in a system of representation according to which each point lies at a certain distance above or below a given line might in this unique case be such that the distance was zero; just as ‘ba ba ba ba’ could conceivably be a sentence of some language. But now suppose that we complicate the rules of the system of representation in the following way: For each new coastline a different length of line is arbitrarily chosen such that for just that length of line the distance of each point in the complex from the line is once again zero. By making up the ‘rules’ of the ‘system’ as we go along, we have reduced its ‘propositions’ to mere names. In order to grasp the ‘meaning’ of each line, the coastline which it names must first be presented in some other, way. Now as Wittgenstein was aware, the rules of a genuine system of representation cannot be made up ad hoc; they must be in some sense lawlike. The state of affairs represented by a proposition is determined by a ‘law of projection’ (Tractatus 4.0141). But the theory of representation has nothing to say about what makes a rule of projection ‘lawlike’. We so far possess only the negative criterion that ‘propositions should not reduce to names’ and do not yet comprehend how it is possible for a language to satisfy that criterion.


26. What prevents me from making up the ‘rules’ of my ‘language’ as I go along? — Other people would not understand what I was saying. But would even ! understand the language? I should be doing no more than ‘naming states of affairs’. If I already possessed a proper language then the naming might be some form of private code. If I didn’t then my words would amount to no more than a meaningless babble. A ‘proper’ language is one for which a distinction between being right and wrong about a particular rule applies. Of course, if I could just arbitrarily ‘decide’ what was ‘right’ and ‘wrong’ the distinction would once again vanish. It must be possible for me to be mistaken, to get things ‘wrong’. But getting things wrong means having false beliefs. The theory of representation cannot employ its account of a false proposition to explain the possibility of false belief; for if we do not understand how false belief is possible then we cannot make sense of whatever it is that the theory of representation does ‘explain’. — We must now examine the metaphysical illusion which would make false belief appear impossible. Only then shall we be able to see what makes false belief a problem.


3. The idealist’s challenge


27. The following story describes the genesis of the metaphysical vision of the ‘pure idealist’. For all the multitude of arguments and positions inspired by idealism, its metaphysic first appears in a guise best described neither as ‘argument’ nor even the ‘description of a position’. Nor, on the contrary, should we expect a simple historical account of an actual experience which some philosopher once had. We shall now witness the first, most naive manifestation of the ego illusion; and because the experience is a metaphysical illusion, the story alludes to a universal experience belonging to all those to whom the dialectic addresses itself (1/6), serving merely to direct attention to a vision which would otherwise have remained only at the very edge of our consciousness.


28. ‘Just now, while gazing at a lampshade, I became aware of a new experience. It seemed as if the lampshade dissolved away, while in its place I perceived an altogether different object: my mental impression of the lampshade. Nothing actually happened. The lampshade did not change its shape or its colour; it didn’t even go fuzzy. But whereas previously I took myself to perceive something in the external world, what I now perceive is in my own mind. The redness of the lampshade, which before was a property of the lampshade-out-there, now belongs to the lampshade-in-my-mind. Nobody can look into my mind to see the colour that I see; for the redness of the lampshade is mine alone. Here I have indeed discovered a perfect example of knowledge I My mind is in direct contact with what it knows; nothing can come between me and the redness of the lampshade because it is part of my own mind. Now I compare my example with other forms of so-called ‘knowledge’. The ‘things in the world’ which I previously took myself to know dissolve away, like the original lampshade. The only things I really know are in my mind. Do I then lack knowledge of the things which are not in my mind? Whenever I even try to think about a ‘thing-in-the-world’ it dissolves away. In that case I fail to mean anything when I say that ‘there are things in the world which are not in my mind’, ‘


29. To the person who recounted that experience we should retort: You think you have invented a new concept ‘mental redness’ which describes the ‘colour’ of your ‘mental lampshade’. But it isn’t really a concept at all. Whenever one uses a concept to describe something — there is indeed no other way of ‘describing’ things or ‘saying’ anything at all — it must be possible to distinguish two questions: First, does it seem to me that the concept applies? Second, does the concept really apply, or am I mistaken? But when you call your mental lampshade ‘red’ your very ‘act of judgement’ makes it true. Since there is no such thing as ‘failing’ in your ‘attempt’ to describe it you cannot be said to ‘succeed’ either. It is as if a marksman decided to call whatever he happened to hit ‘the target’; in that case he does not succeed in ‘hitting the target’. Of course, there is ‘something which the marksman ‘succeeds’ in doing; he is not doing nothing; and indeed uttering a sequence of sounds which seem to have a meaning is not ‘doing nothing’. But what you are doing is just that and nothing more.


30. The objection cannot be evaded by looking to past or future ‘mental contents’ as a possible ‘target’ for judgement. It may seem to the idealist as if his ‘judgement’ concerning past mental contents would be false if what he seems to remember is not in fact the case; if his memory impression fails to correspond with what ‘really happened’. That is to say, having reduced all of spatial reality to ‘my mental contents’ he nevertheless wishes to allow these contents an objective temporal existence. Let us then allow, for the sake of argument, that ‘mental contents’ exist objectively in time: With what right can one speak here of ‘me’ and ‘mine’? For we may freely substitute, in place of the self which ‘experiences’ and the later self which ‘correctly remembers’ having had that experience, the hypothesis of two different selves, thus rendering the ‘memory’ incorrect. But nothing after all turns on the ‘difference’ between these two seeming ‘hypotheses’; the supposed ‘distinction’ is merely verbal. In which case there is no difference between ‘correct’ and ‘incorrect’ memory either. Thus, the idealist does not really ‘remember’ anything; he only seems to remember. Any distinction which he feels inclined to draw between ‘correct’ and ‘incorrect’ memories is merely a distinction between different ‘seeming memories’, similarly, judgements concerning the future may seem to ‘turn out’ to have been true or false. But the statement. ‘I anticipated that’, conceived as a recognition of the truth of a judgement about the future, presupposes a distinction between genuine and seeming memory. The idealist does not ‘succeed’ or ‘fail’ in his ‘attempt’ to anticipate his mind’s future contents; he merely seems to succeed or fail.


31. A philosophically more ‘sophisticated’ idealist may, however, be willing to admit the objections: ‘I realize that my knowledge of what is ‘in my mind’ is not the applying of ‘concepts’ to a state of affairs which only I am able to ‘perceive’. The very notion of ‘applying a concept’ involves a separation of the knowing subject from its object which I regard as impossible. The use of the word ‘perceive’ is misleading: I am not some ‘entity’ which stands in a ‘relation’ to my mind’s contents; rather, I am identical with those contents, I am the redness of the lampshade and also the whiteness of the lamp, the brownness of the stem etc. I cannot conceive how what you call ‘judgement’ could possibly occur. For judgement necessarily ‘aims’ at some ‘reality’ beyond the ‘contents of my mind’; that is to say, it aims at nothing. I accept that normally I find myself ‘making judgements’ both about things in the world and things in my mind; but that is only because I fall victim to the unavoidable illusion that there exists a ‘reality’ outside my mind. To the extent that I attempt to use language to ‘describe’ how ‘reality’ comes to be ‘all in my mind’ I perpetuate that illusion. In reality, there is no ‘impression of a lampshade’, no ‘mental redness’, nor indeed is there an ‘I’, an ‘entity’ subsisting through time which ‘perceives’ and’ judges’. There is only ‘this...’.’


32. One might now be tempted to point out that the idealist quite happily indulges in what he takes to be illusion’ along with the rest of us-when he is not doing metaphysics. It is only then that he finds himself reduced to silence. The idealist’s life ‘contradicts his own philosophy’. But that makes a cheap argument. It simply begs the question of the idealist’s sincerity. Even if the only consistent philosophy were silence, the philosopher continues to live in the world. The idealist sees his ‘indulgence’ in the illusion as only a surrender to the inevitable, rather than as something one may permit oneself on days off or when one’s opponent is not looking. Nor is the illusion of the kind to be dispelled by bluff common sense; as if it were only a mental aberration induced by spending too much time in contemplation instead of getting on with the business of living. In order to address the idealist fairly we must first lower our defences and experience his illusion. This is not a matter merely of extending our sympathies but of deepening our self- knowledge. For the illusion is in us too. That is what makes the metaphysic of idealism something which we are compelled to address.


33. How does one address an opponent whose very philosophy forbids him to reply? — Here at least there appears a foothold. Idealism is ‘meant’ to be a ‘philosophy’, to be a ‘position’ which one ‘holds’, Now that is not obviously incompatible with silence. One thinks of holding a position as a refusal to move; prompted in this case by a vision which ‘strikes one dumb’. However, the idealist must maintain that what his silence means is the perception of how things really are. Otherwise his silence remains indistinguishable from other silences with different ‘meanings’ or none. But now remember what the idealist’s ‘position’ is. Judgement is impossible because there is no reality; or, equivalently, that appearance is the only reality, is not the appearance of anything. The idealist foregoes not only thought and language; he must relinquish the very idea that his ‘foregoing’ means anything at all, that it even goes so far as to express the vision of an inexpressible ‘something’.


34. These remarks do not yet address idealism but merely force the idealist further into his shell. But there remains a strategy open to us: to find an indirect means of expressing the idealist’s ‘inexpressible’, through its effects upon our philosophizing about the nature of the relation between mind and reality. In developing the ability to perceive the idealist’s vision we become able, paradoxically, to speak for him even while he cannot speak for himself. By pursuing the ego illusion through all its ‘various disguises and subterfuges’ (1/12), we come to perceive the illusion in our own selves and develop the weapons with which to destroy its influence. The way to do that is to take up the idealist’s challenge: to discover what we have to say in opposition to his silence, even if what we say cannot be taken to ‘contradict’ his ‘position’. Let us therefore assert that there is such a thing as ‘judgement’, an activity of aiming thoughts at reality. We do not yet know what that means except that there must therefore exist conditions for the possibility of false belief; if there is a target for thoughts to aim at then at least something must count as ‘failing’ to hit it. The attempt to discover those conditions will indeed direct us towards the heart of the ego illusion.


4. Kant’s ‘refutation of idealism’


35. By analogy with mathematical proof, one may speak of metaphysical arguments as either ‘constructive’ or ‘non-constructive’: The bare, non-constructive reduction of a proposition to absurdity, or contradiction, not only concludes the proof but closes off further inquiry; we do not know how to take the conclusion, lacking insight into the reality which makes it true. But the discovery of non-constructive proof in metaphysics presents a special problem: The absurdity or contradiction now threatens our own state of knowledge; presenting an obstacle which must be overcome. Moreover, there is no question here of a wholesale rejection of the coherence of non-constructive proof, by analogy with intuitionist mathematics (cf. Ch.27). One therefore has little cause to be satisfied by the pure idealist’s ‘reduction to silence’. The possibility of ending up trapped in the idealist’s predicament remains for us as long as we lack an account of false belief; and that entails finding a constructive refutation. However, the sole rationale for attempting to ‘locate’ the concept of false belief lies in the project of the diagnosis and critique of the idealist’s ego illusion (3/34), It best serves that concern to examine first a constructive argument which failed to accomplish its stated intention: For it is within Kant’s ‘refutation of idealism’ that the ego illusion finds its new home.


36. Kant shows his respect for ‘problematic’ idealism, the idealism which stops short at scepticism with respect to the existence of an external world (3/28), by taking the trouble to refute it (Critique of Pure Reason B274-9). ‘Dogmatic’ (i.e. pure) idealism, the doctrine that ‘space, with all the things of which it is the inseparable condition, is something which is in itself impossible’, is ‘unavoidable, if space be interpreted as a property that must belong to things in themselves’ (B274). Since, on the contrary, space is only the ‘form of experience’ (vide ‘Transcendental Aesthetic’ B33-45) dogmatic idealism rests upon a false premise. But that inverts the true structure of the dialectic. That space is a property of unknowable ‘external reality’, and consequently a ‘thing in itself’, is one of the conclusions of the argument for problematic idealism; that I can really know only the contents of my mind, an entity which is ‘immaterial’. Any refutation of’ pure idealism must at least entail a demonstration of the invalidity of that argument.


37. If Kant’s ‘refutation’ were valid only against problematic idealism, then talk of pure idealism ‘resting upon a false premise would, however, remain an unjustified ‘explanation’ of the idealist’s ‘mistake’. Problematic idealism, when coherently thought out, certainly leads to pure idealism: As we have seen from the idealist’s ‘story’, construing what was previously taken to be perception of an external world as merely the mind’s apprehension of its own contents leaves no room for the conception of an ‘external world’, and consequently robs the sceptic of the terms which he needs to frame his question (3/28). Now even if problematic idealism were the premise for an argument for pure idealism, there might remain other routes to pure idealism. But, worse, since problematic idealism is not a premise of the argument which ‘leads’ to pure idealism but rather an assumption discharged by reductio, the pure idealist may justifiably claim a ‘refutation of problematic idealism’. Fortunately, contrary to Kant’s own extra-systematic comments, it turns out that the argument itself aims to refute both problematic and pure idealism.


38. I shall state Kant’s argument in my own way: The application of a concept to an object involves an act of recognition; a subject’s awareness of that object’s resemblance, in the relevant respects, to objects which he has previously judged to fall under the concept. ‘Recognition’ implies the identity of the subject who performs these acts of judgement. But the mere act of seeming recognition is not by itself sufficient to constitute the identity of the subject: It must make sense to ask whether a previous act of recognition which he ‘seems to remember’ really took place (compare 3/30). That question can be raised only for a conscious experience which is so ‘structured’ as to determine its interpretation as the perception of an external world of things-to-be-encountered from a point of view within the world; experiences which ‘really take place’ belong to the objective history of a subject tracing a path through that world. Moreover, the subject possesses self-knowledge only on the condition that he has knowledge of an external world: If the only way to raise a question concerning his past experience immediately begs further questions concerning events which occurred in the external world, then in order for him to be able to answer it that experience must be adequate for knowledge of that same world.


39. The ‘anti-sceptical’ component of the argument is conditional: Kant’s intended target was Descartes (B274), whose ‘method of doubt’ stopped short at knowledge of one’s own subjective states, thus challenging the anti-sceptic to prove the existence of an external world given only the ‘indubitable’ knowledge of those states. The argument does not refute the global sceptic, who denies that we can have any ‘knowledge’ about anything. This will become significant later (Ch.9). For now, we should note that although global scepticism and scepticism with respect to an external world are indistinguishable in certain contexts, they arise at different points in the dialectic and it is a fallacy to identify them.


40. Our subject’s knowledge of an external world is not an ‘interpretation’ of his subjective experience in the sense which implies inference from an ‘indubitable’ knowledge of his ‘Cartesian mental states’. For that would require the application of ‘concepts’ which could not be mistakenly applied; for example, the ‘description’ of one’s ‘mental redness’ (3/29). The only ‘objects of experience’ are objects of perception, objects concerning which it is possible to have false beliefs. The subject must therefore perceive those objects directly. However, as we shall soon discover, Kant’s ‘objective world’ reduces to something which is only an ‘interpretation’, in a sense which is indeed vicious. The fatal flaw reveals itself, not in the ‘theory of perception’ but in its underlying metaphysic.


41. That metaphysic reveals itself when we ask for the underlying intention of the ‘refutation of idealism’: Kant is concerned not merely to refute the Cartesian epistemology but also to answer the challenge of the pure idealist, to explain the possibility of false belief. To take up that challenge is tantamount to answering the question: What is objective reality? or, equivalently, what is an objectively real object? We have seen that the very notion of a conscious experience entails both the notion of a ‘mind’ and that of a mind-transcending ‘reality’ at which mind aims its thoughts. Kant understands that as saying what mind and reality really are. The thinking subject simply is and does not merely have ‘experience with the structure necessary for objective experience’. This move from the necessary conditions for objectivity to sufficiency is what makes Kant’s ‘refutation’ constructive (35). Now in attempting to ‘contrast’ the metaphysical concepts of ‘being’ and ‘having’ in this way we are dangerously close to falling into sheer nonsense. For we are dealing with the manifestations of metaphysical illusion and therefore with ‘meanings’ which are in reality nonsensical. And yet somehow we must contrive to give them meaningful expression (1/9, 10), One cannot hope to achieve this simply by the coinage of ‘metaphysical’ concepts. The expression which we seek is not to be gained by enunciating ‘propositions’ but only by working through a dialectic.


42. Let us consider how an idealist might react to Kant’s challenge: The problematic idealist regards Kant’s ‘objective world’ as a poor substitute for the ‘external reality’ whose existence was meant to be in doubt. But here we may join forces with the pure idealist: The sceptic’s idealist premise, when ‘coherently thought out’, robs his notion of an ‘external reality’ of its seeming content (37). The more ‘sophisticated’ pure idealist, by contrast, was prepared to reject the notion of ‘applying a concept’ and along with it the notions of the ‘subject’ and its ‘experiences’: What idealism’ means’ is something inexpressible: ‘this... is all there is’ (3/31). At most, Kant’s argument serves to sharpen pure idealism, exposes the incoherence in cruder forms of the doctrine. It is not merely an ‘accident’ that what I find myself uncritically ‘describing’ as ‘my experience’ has the character of ‘perception of an external world’, as the naive idealist mistakenly supposes. But that only means that ‘self-reference’ stands or falls with the use of ‘concepts’ to ‘describe’ the states and doings of an ‘external world’. In that case Kant’s metaphysic is pure idealism, The addition of the ‘constructive’ component to the ‘refutation of idealism’ does not address the ego illusion; and in the context of the dialectic, to fail to address the ego illusion is to embrace it. (Note that Kant speaks of a different ‘ego illusion’, associated with Cartesian dualism (‘Paralogisms of Pure Reason’ A341-405 and also B399-432); that may be seen merely as a contribution to the project of ‘purifying’ idealism.)


43. Kant says little about the ‘structure’ which would be sufficient for objective experience. (The ‘transcendental deductions’ of the first and second editions (A95-130, B129-69) do not contain an adequate answer to the question.) The omission does not look accidental; the role which the concept of ‘structure’ would have to perform within the demands and limitations of Kant’s metaphysic cannot be performed by any such concept. It is essential to the constructive component of Kant’s argument that the very ‘structure’ of conscious experience itself is the objectivity of the objective world. But now consider the experience which consists solely in ‘travelling back and forth along a row of objects’. What is there in the experience itself to prevent us from reinterpreting a purported ‘change in direction’ as one in which, for example, the subject encounters a second set of objects, forming a mirror image of the first, while travelling in the same direction? Given enough richness of experience, we should rule out competing ‘interpretations’ on the grounds that they did not make sense of the experience, even though admittedly ‘consistent’ with it. However much detail one puts into a ‘model’ of the objective world, the ‘objective world determined by the structure of the experience’ can mean only an interpretation of that experience (not in the sense of inference from a prior ‘given’, 40) subject the holistic constraint of coherence. The answers to the interconnected questions: ‘Where am I?’, ‘How are things at place P?’ are determined by what makes sense of the subject’s overall experience. But this ‘making sense’ is necessarily making sense to the subject. Suppose now one raises the question whether perhaps the subject is ‘wrong’, has a false belief about what ‘makes sense’. This is a question which the ‘structure’ of the subject’s ‘experience’ cannot answer by itself. For the Kantian subject, however, there exists nothing else in reality which could answer it; there is nothing ‘outside’ its ‘possible experience’. It follows that the question does not have an answer; that there is no ‘reality’.


44. If one is thrown off balance by the discovery that ‘experience’ cannot alone ‘determine’ any thing, no matter what ‘structure’ the experience ‘possesses’, that might show that one finds oneself gripped by Kant’s metaphysical vision. But still we have not yet said what that vision is; only that it is the finding of certain questions ‘gripping’. But there remain the projects which constitute the external manifestation of the vision. That Kant should have taken up the project of destroying ‘incoherent’ formulations of the idealist metaphysic (naive and problematic idealism, Cartesian dualism) together with the project of drawing the necessary consequences for the structure of objective reality from the necessary structure of objective experience (in the ‘Analogies of Experience’ B218-265) fixes his position within the dialectic of the ego illusion. Addressing the latter project — although we shall not follow the details of Kant’s own ‘programme’ — will serve as a contribution towards the ‘adequate expression’ of the ego illusion.


5. The refutation of Kant’s transcendental idealism


45. The ‘necessary structure’ of reality (4/44) is determined by the sole condition that it constitute an objective world for a conscious subject which retains an identity through time; its compliance with the conditions for the ‘transcendental unity of apperception’. That same condition entails a model of the thinking subject: to be essentially without properties or any kind of structure. The subject cannot come across itself anywhere within its objective world; for, even supposing that the unity of apperception entailed the necessary existence of an object with which the subject was ‘intimately associated’, in the manner in which one is associated with one’s own body, that object would not be its own self, that ‘transcendental ego’ for the sake of whose unity the structure of reality must comply; while outside the objective world there is nothing (at least, nothing with ‘properties’ or ‘structure’; we shall return to the idea that this ‘nothing’ might remain in some sense ‘something’, 50). Kant calls the seeming ‘self’ that one does ‘come across’ the ‘empirical self’; but it would be more accurate to withhold the term ‘self’, and moreover reserve judgement on whether or not its ‘association’ with a body constitutes a merely empirical feature of the world experienced by the transcendental ego. The Kantian subject has no function whatsoever except simply to be a conscious ‘point of view’ on an objective world: That is at least part of what was ‘meant’ by saying that the subject ‘simply is and does not merely have ‘experience with the structure necessary for objective experience’ (4/41). But it remains deeply nonsensical.


46. Kant’s model of the thinking subject encapsulates what phenomenalists ought to mean by their claim that statements about the external world ‘reduce’ to statements about ‘subjective experiences’. But the model avoids a crucial error which has been associated with that doctrine: There are not two kinds of ‘objects’, ‘subjective experiences’ and ‘external objects’, truths concerning the latter being in some sense ‘supervenient’ upon truths concerning the former (‘supervenience’ has indeed usually been conceived in the strongest possible sense of translation, cf. below). There can be no such ‘way out’ from the problematic idealist’s predicament. Kant’s ‘refutation of idealism’ involves the rejection of the ‘private object’, the object to which concepts cannot be mistakenly applied. The subject’s ‘experience’ is interpreted as the perception of external objects; but not in any sense which implies an inferential connection between knowledge of those objects and a prior knowledge of ‘private, inner objects’ (4/40). We have seen, furthermore, that since the ‘interpretation’ of experience can only be ‘holistic (4/43), statements about the external world could not, in any case, be translated into statements about how things seem to the subject of experience; irrespective of whether or not ‘how things seem’ consisted of ‘private objects’; and that destroys the possibility of embodying the phenomenalist vision in a project of ‘semantic construction’. What Kant’s ‘phenomenalism’ adds to the silence of the pure idealist is rather the attempt to embody his metaphysic in the different project of a ‘transcendental deduction’ from the unity of the transcendental ego: The external world conforms to the condition of being the possible object of experience of the transcendental ego. Now: In the idealist’s silence there was only ‘this...’ (3/31); and what the discovery of a self and a world within the ‘this...’ adds to the ‘this...’ that I alone can ‘mean’ can only be the transcendental ego that! myself am, and the objects of my possible experience. The world is my world (compare Wittgenstein Tractatus 5.62).


47. Let us now put to the phenomenalist the following question: ‘you should not allow that the constitution of your ‘objective world’ is a matter for your arbitrary choice. Rather, your conscious experience ‘determines’ how that world is to be constructed. But how do you know that this ‘world’ is not a mere fantasy which it only seems to you as if you are ‘determined’ to construct out of the experience which you only seem to have had?’ — It would not suffice to describe a possible future experience which would seem to ‘show’ the phenomenalist that his present experience had been a fantasy; for example, an experience of ‘waking up from a dream’. If his present experience, together with his memory of all his past experience, does not contain some feature in virtue of which it may be either correctly or incorrectly described as ‘objective experience’ then neither does that putative ‘experience’. The point of our apparently ‘sceptical’ question is not scepticism; for the phenomenalist’s ‘thinking subject’, the subject which the phenomenalist must indeed suppose himself to be, all ‘objective experience’ is seemingly objective experience. Any ‘distinction’ which the subject wishes to draw between ‘reality’ and ‘fantasy’ merely distinguishes seemingly objective experience from seemingly ‘seemingly objective experience’.


48. Whether or not a person’s experience is a mere fantasy is a question which it indeed makes sense for another person to ask. Someone who is in the grip of a fantasy, or dreaming, fails to react to his environment in ‘appropriate’ ways; i.e., in ways determined by his perception of that environment. But the phenomenalist is committed to search for a criterion by means of which to determine the objectivity of his own experience; he is committed to asking that question about himself. And we have seen that nothing could’ count as a satisfactory answer. The phenomenalist refuses to admit the existence of questions for which ‘my possible experience’ does not provide the answer. But whether one’s experience is not a fantasy is just such a question; for that concerns one’s activity within an environment, as potentially viewed by another person: The question therefore is a question only for that other person, It has been held to be a merest platitude that ‘one does not really know whether one is dreaming’. Of course, we cannot expect a person to answer the question: ‘How do you know that you are not dreaming?’ by describing those features of his experience which ‘prove’ that it is not a dream. But if the idea is that one can coherently ask that question about oneself while lacking only the ability to answer it, then the ‘platitude’ becomes the expression of an incoherent metaphysics.


49. We should note that Kant does not regard the question, ‘What is the difference between objective experience and a dream?’ as posing any threat to phenomenalism (‘Refutation of Idealism’ Note 3, B278-9). Confident in his proof that ‘inner experience in general is possible only through outer experience in general’ Kant says merely that whether a supposed experience is not ‘purely imaginary’ can be ‘ascertained from its special determinations, and through its congruence with the criteria of all real experience’. This immanent criterion is directed against Cartesian scepticism and, as we have already noted, does not shake the global sceptic who accepts that self-knowledge presupposes knowledge of an external world (4/39). But that objection arises at a quite different point in the dialectic, where the coherence of Kant’s metaphysic has not yet come into question (although the sceptical objection certainly suggests that it ought to be put into question). The global sceptic does not need to go so far as to put into question the very existence of an ‘objective world’, but only the possibility of gaining any particular knowledge of its states; thus accepting the point of Kant’s ‘transcendental’ version of idealism. By contrast, our ‘apparently sceptical question’ (47) questions the very foundation of the putative distinction between ‘objective experience’ and ‘seemingly objective experience’; and not merely whether we can know, in any particular case, which is which.


50. In response to our challenge, the phenomenalist may now be tempted to shift ground. He will allow that there are truths ‘about’ experience which are not about the world of one’s experience. Rather they concern its ultimate foundation. One can raise the ‘question’ whether one’s experience is determined by things as they are ‘in themselves’ rather than merely by some faculty hidden within one’s own mind (the self as it is ‘in itself’); but one cannot answer it. He will appeal to Kant’s distinction between ‘phenomena’ and ‘noumena’, Facts about possible experience are phenomenal facts; facts which one is capable of representing to oneself. Whether or not one’s experience is really ‘objective’ is a question about noumena; one cannot even represent the fact that such facts ‘exist’. Now these statements are clearly inconsistent with phenomenalism. But there remains a way of taking the point of the challenge to phenomenalist, that the question of objectivity ‘cannot be answered within the subject’s experience’, while seeming to maintain the ‘coherence’ of the phenomenalist’s question. Stripped of talk of ‘things in themselves’ or ‘noumena’, the only expression of this dialectical ‘position’ is the refusal either to accept or renounce the phenomenalist metaphysic.


51. A ‘passive observer’, a pure ‘point of view’ with no physical identity (45), could not be a ‘thinking subject’. The question of the objectivity of a subject’s experience can be raised only by another person who observes that subject’s behaviour (48). Yet the subject must indeed be so constituted that the question can be raised. The subject is for that reason necessarily embodied as an agent. At this point one must guard against a facile misinterpretation: ‘That is just like Dr. Johnson kicking the stone; while everyone remains unimpressed. That we are agents and not passive observers is an observation which the phenomenalist is happy to accept.’ According to this view, recognition of what we all know to be the case, viz. that we are agents, involves an ‘enrichment’ of Kant’s model; we do not only ‘have experiences’, we initiate changes in the external world. The most that can be said against the original model is that it confines our attention to only one set of phenomena. But it wasn’t intended to do anything else. For there is no question of using Dr. Johnson’s observation as a criterion to ‘prove’ either to oneself or to others that one’s own experience is objective.


52. That objection misunderstands the point of the preceding argument. For the phenomenalist, ‘being an agent’ means having the experience of being an agent (45). But the subject whose experience it is, the transcendental ego, is not itself anything; and a fortiori not an agent (45). ‘Action’ reduces to the experiencing of ‘desires’, together with correlated changes in sensory input. The phenomenalist is a passive observer within his own active body. His response to idealism: ‘False belief is possible because experience has its characteristic structure’, commits him irrevocably to a ‘passive observer’ model; for the self which considers the question, ‘What is it about the nature of my experience which makes it objective experience?’ must construe every question about objective reality as a question to be answered either through the ‘deduction’ of the necessary structure of reality itself, or by observation. The transcendental ego discovers that it is ‘associated’ with a body which behaves in characteristic ways. Meanwhile, we deny the coherence of the question, rejecting the notion of a ‘criterion’ for the objectivity of one’s own experience (47). But the question of the meaning of ‘objectivity’ itself, the possibility of false belief, remains urgent; for if false belief is impossible, then there is no ‘reality’ (3/34, 4/43). If the question of the objectivity of a subject’s experience can be meaningfully raised then it can be raised only by another person (48). But it must be possible to raise the question; the self concerning whom the question is raised is therefore necessarily an agent (51): That is the meaning of ‘objectivity’.


53. If that ‘metaphysical assertion’ seems only a platitude then it has been misunderstood. We should find it shocking, an affront to our deepest intuitions. The phenomenalist in us will object: ‘What use is it to me that another person can know that my experience is objective, that there is a ‘reality’ for me, if I cannot know myself? And if I can so much as question the reality of reality, then the reality of that other person is equally in question! And anyway, what ‘authority’ can he have if the question whether his experience is objective can only be answered by yet another person? And where does the chain of ‘other persons’ end?’ — One can only reply that what the other person knows about me is no ‘use’ to me at all, But that is only because there is nothing which I fail to ‘know’. The reality of the ‘reality for me’ does not rest upon the seeming ‘foundation’ that it seems to me that it must rest upon. For that reason there is no ‘authority’ on the matter which either I or anyone, else either ‘has’ or ‘lacks’. Even if we fail to find the reply satisfying, as we indeed must, we can at least grasp that this is what one ‘has to say’. So there remains the possibility that we shall eventually be able to follow the dialectic through to its conclusion; that ‘grasping’ the argument will in the end give way to conviction.


54. ‘Phenomenalism’ is not the name of a philosophical error; no mere error could exert the power that the phenomenalist’s seeming ‘question’ has over us. The source of that power is called ‘illusion’, the ‘ego illusion’, because only an illusion can continue to tempt us even after its illusory nature has been recognized. To acknowledge the philosophical ‘refutation’ of phenomenalism is more than simply to accept a ‘proposition’ or an ‘argument’; it involves the rejection of entrenched habits of thought. That is why the dialectic has first to be ‘worked through’; and also why we find that work so hard.


6. The metaphysic of action


55. There is such a thing as false belief. There is a genuine distinction between objective and only seemingly objective experience, between reality and fantasy. But the conditions which make false belief possible, the criteria for objective experience are not to be excogitated by assuming the point of view of the subject of experience. That road is b1ocked off. There can be no answer to the idealist’s challenge which accepts the idealist’s own terms of reference; and just that constitutes our answer (5/47-8). So the terms of reference have to be changed: The point of view which determines the objectivity of a subject’s experience can only be attained by a spectator of that subject’s behaviour. The subject of experience is ‘necessarily an agent’ (5/51-2).


56. We know that the metaphysical assertion: ‘The self is an agent’ must be true. Yet we can hardly bring ourselves to believe it; while at the same time not really comprehending what it would mean to ‘believe’ or ‘disbelieve’ it. It is as if an unknown ‘something’ obstinately blocked our vision while itself remaining hidden in shadow; so that we do not even know how to describe how different things seem from the way they ought to seem. That very expression of perplexity contains more truth than we realize; for that unknown ‘something’ has already made an appearance in the dialectic: at the point where Kant’s ‘refutation’ placed the idealist’s illusion out of reach, in the form of a ‘this...’ which cannot be described but only ‘meant’ (4/42, 5/46). Its inexpressibility seemed to render the ‘this...’ immune to all argument. Now in the context of Kant’s metaphysic, ‘this...’ becomes the bare givenness of that ‘material’ to which concepts of an objective world are introduced. The material cannot be described or pointed to in any way because to describe or point to something entails that it is an object; that it falls under concepts. Kant calls this ‘given’ intuition. To assert its existence is dialectically equivalent to, asserting the sufficiency of the Kantian criterion of objectivity: Provided only that the material of intuition is ‘structured’ in the right way, there can be thought of a ‘mind-transcending reality’. Consequently, to reject that criterion entails rejecting the ‘this...’ : There is no room in ultimate reality for both a genuinely objective world and Kantian intuition. But the ‘this...’ obstinately refuses to go away.


57. What does it mean to reject the ‘this...’? All fact and all existence is objective fact and existence. There is nothing to be found in reality such that the objectivity of a person’s experience of seeming to ‘find’ it could not in principle be determined by another person. The insistent presence of the ‘this...’ manifests itself as the experience of finding’ one’s own self; a self whose essential nature sets it apart from the ‘self’ that others can know: The ‘this...’ becomes both outer world and inner self (5/46). This inner self is in itself nothing, a ‘transcendental ego’ without properties or structure (5/45). Thus, when I tell another person ‘who I am’, the information which gets conveyed is of a different order from what that same telling ‘means’ to me. If one strips away the information conveyed there remains only the tautology: ‘I am me’. Yet that seems so far from mere tautology as to express the most profound metaphysical truth. But if we are to reject the ‘this...’ then that seeming truth’ can only be a metaphysical illusion. The ‘essential I’, the ‘inner self’ do not exist.


58. We must now return to the constructive’ refutation of idealism (4/35): In explaining action one at the same time assesses its general appropriateness to the environment, its aptness for attaining the subject’s goals and its success or otherwise in attaining them. A mistake over the application of a concept is potentially manifested in inappropriate, inept or unsuccessful action. That means that the supposition that a ‘mistake’ has been made is always open to the question: ‘so what?’; one must be able to conceive of how that mistake might show up. The naive idealist’s concept of ‘my mental redness’ (3/28) is not a concept because it cannot in that sense ‘mistakenly’ applied. Let us then suppose that the idealist decides to coin a word for his seeming ‘concept’. The practical consequences would not be those of introducing a word into a language, for the subject may now employ it in any way he chooses (cf. Wittgenstein Philosophical Investigations I/268; I take this to be the crux of the so-called ‘private language argument’). The constructive component of our argument lies in its location of the concept of false belief. ‘Mistakes’ (in the widest sense) of judgement and of action are not asserted to be the ‘same thing’; rather, we comprehend the possibility of the former in terms of their role in serving to explain the latter.


59. That the concept of false belief finds its ‘location’ within the explanation of action does not mean that we can define false belief in terms of behaviour (see 14/136-7). We shall indeed have something to say concerning the concept of belief; the ego illusion will reappear in the form of the conviction that definitions of psychological concepts must be available (10/97). We shall also say more about the concept of a concept; the ego illusion reappears there too (18/177-8). For now, we must content ourselves with only the assertion that every concept is a potential link in the movement from thought to action; while putting off until later the explanation of the nature of that ‘linkage’ (18/176, 179-80).


60. ‘Thinking is possible only for an agent.’ — Setting our faces against the ego illusion, we shall cleave to that proposition no matter what might turn up to shake our faith. However, the ego illusion is not the only source of incomprehension; the concept of ‘metaphysical possibility’ is itself deeply problematic. Just when all the meaning of a dialectic seems to have crystallized into a proposition, that proposition reveals a crucial ambiguity. It first appears as a seeming criterion which one could use to resolve disagreements in a particular case. If a purported ‘subject’ is not an agent then it cannot be said to ‘have thoughts’. But consider the case of a person who is paralysed and can only lift one finger in answer to our questions. Provided he seemed to understand us we should regard that as proof that he was in some sense an ‘agent’. Then that same ‘criterion’ could be taken to show that a statue in a garden which ‘moved’ one of its ‘fingers’ to signify ‘agreement’ or ‘disagreement’ with our remarks, say, about the weather was an ‘agent’ too. But that could at most show that some agent was communicating to us through the medium of the statue. But why should one not say the same thing about the paralysed person? Obviously because of what we know of persons in general and in particular the history. of this ‘person’. But that only rules out the hypothesis that this living body is the ‘medium’ for another agent’s communications; or does it? Why can’t the ‘other agent’ be this living body’s former active self? (not literally but in the sense that knowledge of its history determines our attitude towards the living body). — Consider a different tack: The paralysed person can ‘show’ that he has ‘needs’ and ‘desires’ by, for example, responding to the question whether he would like the window open; for that means he is affected by his environment in the sense that the environment matters to him. But the statue can ‘respond’ too. If one appeals to the possibility of distinguishing the ‘causal source’ of the ‘responses’ in the two cases, then consider instead a computer which appeared to ‘communicate’ with us intelligently. — There is a way of answering all these questions; but only when our seeming ‘proposition’ dissolves once more into a dialectic (a ‘dialectic of analysis’, cf. 13/120).


61. And just as we cannot reduce the results of the dialectic to a proposition, so the dialectic cannot be taken to ‘express’ any ‘position’. In destroying the very possibility of rival ‘positions’, the idea that one has merely chosen the best metaphysical ‘explanation’ or ‘theory’ for some independently understood ‘question’ or ‘problem’, one loses one’s own sense of ‘position’; one cannot take up a position in empty space. Despite this, there remains some heuristic value in ‘identifying one’s position’, if only that it provides a convenient name with which to reactivate a memory of the argument. Thus: To explain the possibility of false belief is to explain how thought is made to fit reality, how thought is so much as possible. For Kant, the very notion of ‘structured experience’ is the notion of subject and object, of mind and reality (5/45). Experience is the substance in which mind and reality inhere as necessary correlates; we comprehend the relation between thought and reality through a metaphysic of experience. Kant’s metaphysic fails, not because the concept of ‘experience’ as such is incoherent, but because he places upon it a burden which it cannot bear. (A ‘metaphysics of experience’ (1/1) is, by contrast, only a bad choice of starting point for a metaphysical inquiry (2/15); which does not necessarily entail illusion but may simply block the possibility of attaining an effective critical stance towards it.) Now the search for an ‘adequate’ metaphysic involves the identification of a concept which we may take as ‘fundamental’, in the way that the concept of experience fails to be. That concept is the concept of an agent; our metaphysic is a metaphysic of action.


62. One may speak of a ‘shift in perspective’ from a metaphysic of experience to that of a metaphysic of action. For as we lose all the philosophical projects inspired by the rejected metaphysic, so we gain the new project of a ‘diagnosis and critique’ of the illusion which inspired the old. That will involve a reassessment of what have been taken to be the philosopher’s ‘central tasks’. We cannot expect that what were previously regarded as ‘philosophical questions’ will necessarily retain their status, once the change in our terms of reference has been carried through. Moreover, we shall acquire the means to express questions which could not previously have been put.


7. Objective idealism


63. When Kant set out to destroy naive, ‘empirical’ idealism his underlying intention was to save the idealist ‘insight’ from the self-refuting consequences of its naive formulation (4/42). Against the viciously subjective ‘world’ of ‘private mental particulars’ he placed a world of mind-transcending ‘objects of the mind’s perception’ as the sole condition for the existence of a ‘mental subject’; an objective world (4/38). The programme of the Critique of Pure Reason is so convincingly worked out that one may now be forgiven for doubting whether there could be a more ‘objective’ idealism than transcendental idealism. some commentators have even balked at calling Kant’s metaphysic an ‘idealism’, as if there were nothing one could add to Kant’s own ‘refutation of idealism’. (Strawson, in The Bounds of Sense, views what he call’s Kant’s ‘transcendental idealism’, i.e. the interconnected doctrines of the ‘psychology of the faculties’ and of ‘phenomena and noumena’ (cf. 5/50), as an aberration in what would otherwise have been a perfectly acceptable programme of ‘drawing the bounds of sense’ (Pt.I sec.I, 4 and Pt.IV). This reaction will prove significant later (Ch.ll). However we fail to do justice either to the history or to the dialectic of idealism if we identify ‘idealism’ with empirical idealism, or indeed regard ‘transcendental’ idealism as the most objective form of idealism. Our critique of Kant’s metaphysic has already provided reasons for denying the former identification. The purpose of this chapter is to show how the dialectic still allows room for an ‘objective’ idealism which is entitled to call the Kantian idealism which it rejects ‘viciously subjective’; and to show how this new metaphysic remains an idealism nonetheless.


64. Idealism begins with an ‘experience’; the ‘primary manifestation’ of the ego illusion (3/28). Of course that is our description. Each successive critique of a metaphysical experience diagnoses a different ‘illusion’; and in so doing sees more or less ‘truth’ in the original ‘vision’, hidden by the illusion but revealed in the critique. At its bare minimum that ‘truth’ is just the fact that the experience is nothing but illusion: That is how the metaphysic of action sees the ego illusion. Transcendental idealism, by contrast, sees not only a different ‘illusion’ (empirical idealism, Cartesian dualism) but also a more substantial ‘truth’: The idealist’s ‘this...’ is a vision of something true; but how to express that insight without illicitly using the notions of ‘concept’ and ‘object’, how to express recognition of the ‘this...’ without falling victim to the illusion of empirical idealism is the real project; diagnosing the illusion serves only as a preliminary. Now one may think that the very idea of such a ‘real’ metaphysical project is itself an illusion; the only coherent project is the dialectic of illusion. That is indeed the stance taken up by the metaphysics of meaning (Ch.l), and realized in its ‘metaphysic of action’ (6/61). But we can also grasp the further possibility of a ‘position’ which not only finds an ‘illusion’ to diagnose in the idealist experience but also sets itself a different project from that of the transcendental idealist; rejecting his project only on the grounds that it fails to describe coherently the idealist’s vision. If there is such a possibility within the dialectic of idealism then by finding a means to express it we stand to learn more about the nature of the ego illusion; and nothing less than learning all there is to learn about the illusion will suffice to arm ourselves adequately against it.


65. This new idealism should occupy a position in dialectical space in between transcendental idealism and the metaphysic of action. That is simply where we are now looking. In that case it will partially concur with our own diagnosis of the illusion in transcendental idealism; while at the same time falling short of that diagnosis. (That of course does not determine whether the new position will be either more or less acceptable to our unrefined intuitions.) If we are to identify that position our first task must therefore be to gain a fuller understanding of the project which, for partially dissimilar reasons, we both reject: the project of transcendental idealism.


66. A ‘metaphysic’ is a constructive rejection of metaphysical illusion; any lesser response must fail to determine the truth in the fact of illusion, how its experience is to be understood (4/35). The constructive aspect of the metaphysic of experience consists in its treating the ‘material’ which it regards as necessary for experience as also sufficient (4/41, 5/45, 6/56). Its project is to show how this material ‘suffices’: a stream of ‘intuition’ structured in such a manner as to determine’ its interpretation as a ‘self’ and a ‘reality’ which the self perceives from a point of view. How could the idea of such a project have arisen? To assert only that the material of structured intuition is necessary for experience would only be to deny that the empirical idealist’s ‘material’ is sufficient. Talk of the ‘material of intuition’ might be regarded here as merely ad hominem, not to be taken seriously. But then, under the influence of the ego illusion, one finds oneself wanting to express the ‘truth’ in the idealist experience; that ‘this...’ is all there is’. Of course that will not do because talk of ‘what there is’ is necessarily talk of what objects there are in the world of one’s perception. In that case talk of the structure’ ‘sufficing for experience’ must be construed metaphorically; the truth in the metaphor can only be exhibited in the project of ‘showing how the structure of the material suffices’.


67. Bearing in mind the essential element of metaphor, let us call the ‘this...’ which is all there is the ‘primary being’ Now a certain kind of primary being suffices for experience. Let us forget the problem of specifying the ‘right kind’ of primary being (4/43) and ask instead: For whose experience does the primary being ‘suffice’? The question cannot be permitted. If one were to answer: ‘anybody’s’, then one is forced to choose between two equally inadequate conceptions of the primary being: If each person’s experience is constituted by the material of its own primary being then it becomes false that the primary being ‘is all there is’. Reality is at least a plurality of primary beings. But if, instead, one says that each person’s experience is merely of the same primary being then one loses one’s grip upon the ‘this...’, the ‘truth’ in idealism. Of course one wants to answer that the primary being suffices for ‘my experience’ (5/46). But the use of the concepts of ‘me’ and ‘mine’ is legitimate only within the world of objective particulars where I exist as an ‘empirical subject’ alongside other ‘I’s’, each with its own experience. And in this legitimate sense it is once again false that ‘this... is all there is.’ There are other ‘thises’ besides my ‘this...’. The question one is permitted to ask in place of the impermissible: ‘For whose experience does the primary being suffice?’ is only: ‘Whose task is it to show that the primary being suffices for experience?’ The answer here is indeed: ‘mine’. That task is the project of transcendental idealism. By recognizing the existence of such a task I refuse to allow that there can be any question concerning my cognitive relation with reality, the validity and objectivity of my experience, which I cannot satisfactorily answer for myself (5/48). ‘Taking up’ that task ‘exhibits’ my idealism: that ‘this...’, that which is given to me for the accomplishment of the task, is all the material that counts in my determining the relation between my mind and reality.


68. However, that task is an impossible task. My seeming to have ‘taken it up’ is only a symptom of my ego illusion. What then becomes of the ‘primary being’? Does it not exist? — That is the ‘position’ maintained by the metaphysic of action (6/56), Any attempt to ‘express’ or ‘exhibit’ the ‘existence of the primary being’ is idealism; and all idealism is an illusion. But putting the matter that way raises the suspicion of a non-sequitur, Perhaps there remains an overlooked possibility; a way of saving the primary being without falling into the trap of attempting to take up the transcendental idealist’s impossible task, Previously, we rejected the possibility that experience is a ‘plurality of primary beings’. Such an ‘experience’ is not the primary being that I ‘mean’ when I say ‘this...’. But that construal does not seem to be forced upon us by the concession that ‘each person’s experience is constituted by the material of its own primary being’. Why not say instead that each person’s experience merely partakes of its own portion of the very same primary being, while yet remaining the ‘this...’ to which mythis...’ was ultimately meant to refer? The ‘this...’ which is ‘all there is’ would then be shared by a plurality of centres of experience; my ‘this...’, contrary to the interpretation of transcendental idealism, would then be only an incomplete fragment of a greater totality in which all ‘thises’ belong together. From the point of view of such an idealism, to regard my ‘this...’ as ‘all there is’ is a vicious subjectivization of the primary being. This new idealism, by contrast, is an ‘objective idealism.


69. The objective idealist’s talk of partaking, of ‘portions’ of the primary being, of ‘fragments’ of a ‘greater totality’ must, as in the case of transcendental idealism, be converted from mere metaphor into a metaphysical project. The following suggests itself: Mythis…’ does not seem to be a mere fragment’; it seems absolutely immune to all challenge, complete in itself. Just because nothing can be said about it, nothing can be wrongly said about it either. But suppose now that the ‘world of appearance’ which I construct on the basis of the structure of the ‘this...’ cannot, contrary to appearance, be ‘coherently thought out; that subsuming ‘intuition’ under ‘concepts’ of an ‘objective world’ in the manner of transcendental idealism cannot result in a self-consistent account of reality. Suppose further that it is of the very nature of conceptual thought that each one of us can know reality only in this defective way. For that reason our ‘knowledge’ is only of appearance and not the true manner of knowing reality. The project of metaphysics is then to show up the ‘inconsistent nature’ of the world of appearances; and moreover show it as an inconsistency which can be overcome only by describing the principles according to which ‘ultimate reality’, the undivided primary being, divides up into a world of subjects of experience and the inconsistent appearances which they perceive.


70. For the objective idealist, the reality we thought we knew breaks up into a ‘mere appearance. But this appearance is not the unquestioned ‘given’ of the naive idealist’s experience; a ‘given’ whose mere ‘recognition’ immediately renders the existence of an objective world ‘problematic’ (3/28). It is the result of a critique which takes the ‘reality we thought we knew’ as its far from unquestioned ‘given’; and which then proceeds to demonstrate its inconsistency. Now saying that does not replace the metaphor of a ‘this...’ to which my ‘this...’ belongs as part to whole. The bare ‘schema’ for a project awaits realization in the form of convincing arguments directed against the world of ‘appearances’ (for example, something which would justify Bradley’s attack, in Appearance and Reality, on the coherence of relational properties). Only then would objective idealism become a challenge to the metaphysic of action.


8. Perception and action; and primary qualities


71. The challenge of an objective idealism lies in its rejection of the phenomenalist project, while purporting to remain within an idealist metaphysic. To achieve that result, it must attack the coherence of the idea of a world ‘external’ to mind; an idea which phenomenalism and the metaphysic of action both share, although differing in their accounts of its possibility. The challenge which we shall now consider aims to undertake, by contrast, the project of an internal critique of phenomenalism; to show that the model of the passive observer and its ‘phenomenal’ world are merely the wrong conclusions drawn from the right starting point, viz., the transcendental unity of apperception (5/45). We know that the challenge cannot succeed, before even examining its arguments (5/51-2). It is nevertheless worth doing so; for the model of the ‘self and its world’ which it purports to derive from the Kantian criterion of objectivity appears identical to that determined by a metaphysic of action. In learning to see through that superficial identity we shall have strengthened our grip upon what the rejection of the ego illusion really means.


72. A system of metaphysics must aim to harmonize its consequences with what we take ourselves to ‘know’ about the world of our experience. This demand is a corollary of the twin ideals of completeness and wholeness (1/3). It does not rule out the possibility of a revision of what we ‘take ourselves to know’ when that is seen under the aspect of ultimate reality; but the discrepancy must then be explained away through the concept of illusion (an extreme example would be Parmenides’ ‘One’). The project of phenomenology may be seen as the attempt to satisfy that demand of ‘empirical adequacy’ in the case of the phenomenalist metaphysic; by showing that those aspects of mind and reality which cannot be thought away without fundamentally revising our pre-systematic understanding of how they are ‘related’ belong to the ‘necessary structure’ of the world of the transcendental ego; for example, that the self’s experience is that of an agent and not a disembodied passive observer. The ambitions of the phenomenological project thus threaten the very possibility of expressing the phenomenalist’s illusion through its ‘external manifestations’ (4/44). If every assertion aimed at contradicting the phenomenalist’s model of a self and its world can be translated into a form acceptable to the phenomenalist, if every consequence of his metaphysic which appears ‘revisionary’ can be explained away, then our charge of ‘illusion’ becomes otiose.


73. But that is all superficial. The phenomenalist’s illusion is identified, not in the supposed ‘consequences’ of his metaphysic — as if the metaphysic of experience and of action meant the same thing by ‘determining the necessary structure of reality’ and were held to differ only in what they took that ‘structure’ to be — but in the project of working those consequences out. The transcendental ego, for whom alone that project is a ‘project’, remains a ‘passive observer within its own active body’ (5/52), regardless of whether or not being an agent is a necessary feature of the world of its experience. And we have seen moreover how the rejection of that project ‘affronts our deepest intuitions’ (5/53, 6/56-7); so far is the metaphysic of action from being opposed to the ‘revision of our pre-systematic understanding’ of how mind and reality are related.


74. The first phenomenological project we shall consider concerns the rejection of the passive-observer model of perception. The resume of the argument will appear strangely familiar: ‘The concept of an object which belongs to an external world necessarily includes a potential to causally interact with other objects. Only on that condition could there be inference from things seen to things unseen, and consequently the possibility of correcting perceptual judgements by means of knowledge of an external world gained from past perceptions. Now a hallucinated ‘object’ is just an object one seems to ‘perceive’ but which turns out not to participate in the scheme of causal interactions. But that opens up an unforeseen gap between two concepts of ‘causal interaction’, The predicament of the passive observer prevents him from making sense of any notion of ‘causal interaction’ besides that of a mere predictability of changes within the passing show of his experience. All his ‘objects’ are merely hallucinated objects, and ‘interact’ only in this weak sense. Only for a subject whose experience of ‘interaction’ with objects as an agent within the world necessitates a conceptual distinction between ‘real’ and ‘hallucinated’ objects, between objects which can and cannot be acted upon directly or indirectly, can there be said to exist a genuinely objective world, as opposed to a merely predictable ‘dream world’.’ — That argument illustrates how the anti-phenomenalist assertion that ‘only for an agent can there be said to exist a distinction between objective and seemingly objective experience’ appears, on the surface, to get translated into a form acceptable to the phenomenalist.


75. We should first observe that the phenomenological argument for the necessity of the experience of agency succeeds only through a subtle change in the terms of reference of the phenomenalist project of deducing the necessary consequences of the transcendental unity of apperception. A ‘predictable dream world’ suffices to account for the possibility of false belief, for the existence of a subject of experience with a point of view on an objective world, provided we do not go so far as to question the very possibility of a ‘first-person criterion’ of objectivity. That there is something else which it does not suffice to explain may be very interesting but strictly beside the point. That is the way with ‘phenomenological’ arguments (compare Ch.ll). But let us tackle the argument head-on, For the phenomenalist, action is merely the ‘experiencing of desires, together with correlated changes in sensory input’ (5/52). Now a further phenomenological argument would establish that desires could be intentions only in a world where it was in some sense ‘not accidental’ that certain movements of one’s own body follow the occurrence of certain kinds of desire. Let us grant that. For the transcendental ego whose experience is that of an agent, objective experience of real, non-hallucinated objects remains indistinguishable from a merely seemingly ‘objective’ ‘experience of real objects’. The experience of agency, albeit necessary, is not the ‘necessity of agency’ entailed by the rejection of the transcendental ego.


76. Our second phenomenological project aims to rectify a ‘lacuna’ in the phenomenalist’s account of the relation between experience and its objects. However, we should not be surprised that, from an apparently different starting point, the second argument ends up covering the very same ground. Once more, the crucial failing of phenomenalism is only half-perceived, and its ‘rejection’ merely ‘translated’ into a form which does not challenge the phenomenalist’s metaphysic: ‘In a passive observer’s world of, say, purely visual objects there can be no room for a conception of what the subject’s perceptions are perceptions of, of a world of space-occupying substances whose states and doings can alone determine the truth or falsity of perceptual statements. But if their truth value remains ‘undetermined’ then there is no difference between perceiving and only seeming to ‘perceive’: Only the experience of being an agent contains materials sufficient to justify a distinction between primary and secondary qualities of objects; the former constituting the ‘potentiality of the object as causal agent’ (cf. the first argument), and which, in combination with the properties of sensory organs, both explain the existence of the latter and consequently determine the truth value of perceptual statements. Therefore, only such a world may be regarded as genuinely objective.’ — The interest in this argument lies in its suggestion that the problem which tempts the phenomenalist to ‘shift ground’ and distinguish between the phenomenal world of ‘appearances’ and its ‘noumenal reality’ (5/50) can be solved within the world of phenomena, merely by enriching experience. We of course know that the ‘problem’ cannot be ‘solved’ by any such manoeuvre; that, by contrast with the world of a metaphysic of action, the only ‘primary qualities’ which the phenomenalist is able to comprehend reduce to the ‘constructs’ of mere ‘percepta’.


77. One might now proceed to expand the foregoing remarks to a ‘critique of all phenomenology’; but that should hardly be necessary. Instead, we shall consider the essentially novel meaning which the determination of the ‘necessary structure of reality’ (73) acquires within a metaphysic of action. Our metaphysic is a full-blooded ‘dialectic of illusion’; that means there can be no room for metaphysical ‘assertions’ which do not serve the purpose of a ‘diagnosis and critique of metaphysical illusion’. Consequently, any statement that such-and-such is ‘necessary’ can be understood only as the rejection of a correlative ‘illusion of possibility’: We reject the picture of a constant ‘form’ underlying the fluctuations of ‘experience’ which metaphysics seeks to ‘discover’.


78. The phenomenalist sees the actual world only in the distorting mirror of his account of the ‘conditions for the possibility of an objective world’: The transcendental ego discovers a plurality of spatially unrelated ‘spaces’; visual, auditory and tactile ‘worlds’ whose objects can, and indeed must be correlated but for which it remains empty to assert an additional, supervening identity. This seeming ‘plurality’ simply cannot be seen from the point of view of a metaphysic of action; for its spatial world is the world of primary qualities; and primary qualities cannot be reduced to the ‘constructs of mere percepta’ (76). For the same reason, the phenomenalist allows the possibility of a plurality of spatially unrelated ‘spaces’ of the very same categories of percepta: One imagines a ‘dream-world’ to which one ‘returns’ every night and whose ‘coherence’ makes it as ‘real’ as the ‘real world’, so that there is nothing to distinguish between them (compare Kant on the ‘criteria’ of dreaming, 5/49). Allowing a certain degree of ‘correlation’ between one’s experiences of the unrelated spaces is all the phenomenalist requires for a unitary ‘experience of an objective world’ (the simplest kind would consist of ‘interactions’ between objects in the two worlds). Now, the metaphysic of action remains agnostic on the question of ‘spatially unrelated spaces’; so described. We assert only that the mere ‘experience’ of spatially unrelated spaces must be reinterpreted so that either the spaces are thought to be in space — while we remain ignorant of their spatial relation — or that one of the ‘experiences’ is merely a collective hallucination (i.e. the ‘experience’ we are not having now). For the hypothesis of ‘hallucination’ could be disproved only by the discovery either of the spatial relation or of some other form of ‘causal connection’ which did not reduce to the ‘lawlike correlation’ of seeming ‘perceptions’. The experience must be reinterpreted because it can be; and because it could be coherently taken at its face value only within a metaphysic of experience.


9. The epistemology of the passive observer


79. Why should metaphysics take any interest in epistemology? There are two bad reasons: First, the rejection of the illusion appears to depend, at a critical point, upon the possibility of knowing something; for example, the transcendental ego’s ‘knowledge’ that it is not dreaming (5/47-8). But that amounts to no more than a crude misreading of the argument: The rhetorical question: ‘How do you know…?’ serves merely as a device for exhibiting the identity of the ‘hypotheses’ that a given ‘experience’ is ‘objective’ or ‘non-objective’ (5/49). Second, the question whether one can ‘know anything at all’ appears to derive its motivation independently of the illusion and the projects which it motivates. Thus, Descartes’ question: ‘How do I know that I am not dreaming?’ might seem a question anyone might choose to initiate a search for the ‘foundations’ of knowledge, whether or not one had ever been gripped by the ego illusion, simply because knowledge which falls short of ‘proof’ fails to correspond to a model of mathematical knowledge which, for some other, undisclosed reason one prefers as the ‘ideal’. The idea of an ‘independent’ epistemology indeed carries the weight of tradition on its side; but we can simply turn a deaf ear, refusing either to accept or reject the ‘interest and fruitfulness’ (2/15) of its questions. The only valid reason for our ‘taking an interest’ in epistemology is this, since a metaphysic describes mind’s relation to reality, and since knowing is a manner of mind’s relating to reality, we should expect the articulation of that metaphysic to have consequences for our understanding of ‘knowledge’. And that means: The rejection of illusions concerning mind’s relation to reality entails the rejection of illusions concerning the nature of knowledge.


80. In a metaphysic of experience all knowing becomes, in a vicious sense, ‘only an interpretation’ (4/40, 3). The thinking subject consists solely in a stream of experience, its sole activity the discernment of objects through the application of concepts (5/45); a passive observer (5/51). For a passive ‘observer’ in the literal sense, a disembodied eye, all knowledge would derive from visual perception. But the phenomenalist is a ‘passive observer within his own active body’ (5/52). What the active body ‘knows’, the inner, ‘transcendental self’ must also know, in a manner derived purely from perception of the world through the medium of the outer, ‘empirical self’. For the determination of what, and how the subject knows is a question which only the inner self can answer. Its knowledge of objects necessarily holds those objects at a distance apart from its own self; and that metaphysical ‘distance’ remains even when, in the case where the outer ‘self’ perceives, by means of touch or taste, the spatial distance reduces to zero.


81. Of course, encountering an obstacle or manipulating a tool is not a form of knowing in any way comparable to the paradigm of visual perception. But what must be can be: From the phenomenalist’s critical standpoint, that determined by his constructive response to idealism, the whole substance of the practical knowledge which these activities give us cannot amount to more than would be gained from the conscious application of concepts to objects; for what that knowledge ‘amounts’ to depends upon the answer to the question of its meaning and validity which only the observing, inner self is in a position to give. That practical knowledge does not present itself as the ‘application of concepts to objects’ can only mean that the workings of our intellectual faculties do not lie open to view; imagination takes over, presenting this knowledge in a form in which its true nature remains hidden. Making sense of these complications would seem to present a rich field for ‘phenomenological’ investigation (5/72). But any such project must remain strictly ‘beside the point’ (5/75). The phenomenalist remains irrevocably committed to an ‘epistemology of the passive observer’, whatever the phenomenologist may choose to say in mitigation.


82. Just as the transcendental ego would remain a ‘thinking subject’, even if its experience was that of a passive observer in the literal sense, so ‘knowledge’ would remain essentially what it is even if its only source was visual perception. By virtue of that possibility, the ego becomes a passive observer and its knowledge becomes a knowledge derived solely from perception. All knowledge reduces to propositional knowledge; an application of concepts to objects whose sole function is to ‘anticipate’ possible perception. Our target is that illusory concept of ‘knowledge’. But we are not concerned to discredit the concepts of ‘visual perception’ or ‘propositional knowledge’. In the metaphysic of experience, visual perception and propositional knowledge become something other than the knowledge and perception which we know; for the phenomenalist project distorts the truth about their essential nature into falsehood.


83. In visual perception there is a subject who seems to see and an object seen. If the subject’s perception, say, of a seeming ‘F’, is veridical then, truistically, what the subject seems to see matches up with its object; there really is an F seen. Now the meaning of the passive observer’s ‘knowledge’ consists entirely in its function of ‘anticipating possible perception’. Its form therefore inherits the formal properties of perception; knowledge essentially involves a correspondence between how the subject represents things, and how they really are. Therein lies the fatal ‘distortion’. Let the passive observer have a true ‘representation’; a proposition which ‘corresponds’ with reality. What is it worth? For all he knows, the representation is false, does not correspond with reality. Knowing is something more than having a true belief which only happens not to be false, a representation which is correct only by accident. Then what more should the passive observer want it to be? One may think of judging as a game played against reality. Reality wins when the judgement ‘turns out false’; the subject wins by showing that reality can’t win. But now the mere fact that one can describe how the belief might, after all, ‘turn out false’, however certain one may be that such an event could never occur, means that the game can only end when the subject is defeated. Of course, the judgement that reality has ‘won’ may itself turn out to be false. But then reality does not need to ‘prove’ anything in order to win; its opponent merely gives up: The only thing the passive observer could regard as sufficient to convert mere true belief into knowledge is proof; winning the judgement game. And with empirical propositions, such ‘proof’ is not to be had.


84. Let us now suppose that the subject possesses a second representation, depicting the manner in which the first representation was obtained; its derivation from a ‘reliable’ causal route, at no point relying upon either faulty perceptual mechanisms or false assumptions; the ‘fact that the first representation is not true only by accident’. The ‘distinction’ between knowledge and mere true belief which this seeming ‘criterion’ suggests is a fake: We may now ask by what ‘manner’ this second representation was ‘obtained’; and the subject can reply only with a third representation. The regress is vicious because, in construing all knowledge as ‘representation’, the phenomenalist has constructed a device which immediately converts every seeming ‘answer’ into another ‘question’. No mere ‘representation’ can satisfy the passive observer’s desire to know that what he believes is indeed ‘knowledge’; for that would entail an ‘intrinsic guarantee of correspondence’, something which is ruled out by the very nature of applying a concept (3/29). But ‘representations’ are all the passive observer has. The phenomenalist is thus condemned to global scepticism.


85. Global scepticism is the possibility which Kant overlooked in his refutation of the problematic idealist’s scepticism with respect to the existence of an external world. The global sceptic ‘does not need to go so far as to put into question the very existence of an objective world, but only the possibility of gaining any particular knowledge of its states’ (5/49); while, on the other hand, going further than scepticism only with respect to the states of an external world’ (4/39). For the premise of Descartes’ ‘method of doubt’, that there is nothing in my present or past experience which ‘proves’ that it will not at some future time seem to me that I had only been dreaming (5/48), remains unassailable.


86. ‘Are we not in the same predicament as that of the passive observer? how can anyone prove that they really know anything?’ — The original question did not concern ‘proof’ but only the value of this state we call ‘knowledge’ (83). If all knowledge reduces to a pure ‘representation’ then, short of that proof which is ‘not to be had’, ‘knowledge’ cannot be knowledge, beliefs Can be true only ‘by accident’. In that case knowledge cannot be construed as ‘representation’. And we are entitled to that conclusion only if we first reject the phenomenalist’s ‘critical standpoint’ (81): The ‘value’ of knowledge can be grasped only by first rejecting the critical perspective of the ‘first person’. Knowledge is essentially something one shows in one’s actions; and the value of knowledge is in its value for action. By thus refusing to accept the reduction of knowledge to a representation we shall indeed have brought the thinking subject ‘back into contact with its object’: The pure idealist’s objection to the very notion of a ‘separation’ between subject and object (3/31) now appears only a confused perception of something which is indeed true.


87. Knowledge is ‘for the sake of action’: That is indeed not intended as a ‘phenomenological’ assertion. The experience of agency includes a place for the concept of a state called ‘knowledge’, where the experience of being a passive observer does not. But that is not the knowledge that the transcendental ego means when it asks itself: ‘What do I really know?’. This inner self merely observes that a certain ‘distinction’ is coined for a particular purpose; but that ‘purpose’ belongs only to the outer ‘self’ which it observes and not to its own self. The rejection of the ego illusion entails the rejection of that seeming ‘question’ which the semblance of an ‘inner self’ seems to ‘put to itself’.


88. The point of distinguishing knowledge from mere true belief lies in the practical, consequences of knowing; in the contribution made to our interpretation of a person’s actions by the attribution of knowledge. Judgement plays a different ‘game’ against reality (83): In practical reasoning I attempt to work out a strategy for achieving my goals which reality will not refute. We call practical reasoning sound when its success would not have been a matter of mere luck. Sound practical reasoning is based upon knowing how things are and not just possessing beliefs which accidentally turn out to be true. The concept of knowledge is thus fixed by our interest in assessing the soundness of practical reasoning; one should aim to base one’s reasoning upon knowledge. The ‘reliable causal routes’ (84) serve to identify the very distinction which matters to us.


89. Is that a refutation of scepticism? The question has no determinate answer. We may grant that what scepticism means is correct: The phenomenalist’s conception of knowledge is incoherent and should be rejected. (If the sceptic is prepared to get down to cases and challenge specific knowledge claims, that should be considered on its merits; but metaphysics is not required to take any interest in such a project.) However, the self-professed ‘global sceptic’ fails to think his position through to its logical conclusion: in refusing to see ‘another possibility’ he remains under the spell of the ego illusion.


10. Psychological explanation


90. In this chapter, we shall encounter a second ‘refutation of phenomenalism’; and not the last (17/161-3, 18/178). What could be the angle? If the first refutation (5/47-8) were less than conclusive then producing a second amounts at best to adding something to nothing; ‘at best’, because the admission that one argument cannot secure conviction without the corroboration of the other threatens to weaken their combined effect. Metaphysics cannot, in any case, tolerate less than conclusive arguments; for that suggests that we are searching only for the ‘most convincing’ among rival ‘metaphysical explanations’ (6/61). But nor does the second argument either establish a premise upon which the first relies, or clarify some aspect of the first argument which was left unclear. The rationale for a second, and subsequent refutations of phenomenalism is rather this: In each refutation a different aspect of the ego illusion is brought to light; and that serves our constructive purpose (4/35). The progressive expression of the illusion (1/10) simultaneously involves its progressive elimination. The manner of expressing the illusion in the form of ‘refutations’ may, however, be contrasted with the examination of consequences which we should have had to accept were an adequate refutation not available. For example, that a metaphysic of experience entails global scepticism (90/80-4) is unfortunate for the phenomenalist but in itself no argument for rejecting his position.


91. The passive observer ‘knows’ what he believes and desires a priori. Awareness of his employment of ‘representations’ (9/82-3) becomes a seemingly ‘infallible’ mode of access to his own cognitive states, immune from any empirical discovery concerning the world of his possible experience: I represent a state of affairs to myself; and that representation in turn consists in a ‘state of affairs’ whose properties are ‘immediately given’ to introspection. But I may be said to ‘know’ what my representation represents, to be capable of ‘recognizing’ when a representation of some possible future course of experience ‘matches’ the experience and when it ‘fails to match’, only in this sense: when it seems to me that it ‘matches’ then it matches. In a metaphysic of experience, the ‘private object’ is rejected (5/46), only to be replaced by the private content.


92. We may reformulate the refutation of phenomenalism as an attack upon the notion of a ‘private content’: The passive observer’s experience, by virtue of its having a determinate ‘structure’, directs the self to interpret it in a particular way (4/43). False belief consists in failure to match a given representation to reality. Conversely, frustrated desire consists in the ‘failure’ of reality to match a given representation. But let us now suggest that all his beliefs can be rendered ‘true’, all his desires ‘satisfied’, merely by changing the meanings of the terms in his ‘system of representation’. The passive observer’s ‘experience’ is indistinguishable from mere fantasy because there is nothing to distinguish ‘correct’ from ‘incorrect’ employment of his ‘private representations’ (6/58).


93. Our ‘second refutation’ prompts an illuminating misunderstanding: ‘Now I see the phenomenalist’s mistake: He has misconstrued perception of his own mental states as a form of ‘incorrigible access’ to a state of affairs; like the naive idealist’s ‘knowledge’ of his ‘mental redness’ (3/29, 31). If false judgement concerning the content of one’s mental states is to be possible then they must possess physical properties, available in principle to anyone’s inspection; they must be physical states of the brain.’ --suppose we open up a person’s skull in order to ‘inspect’ his brain states. Here is one: a ‘representation’ of a possible state of affairs. But which possible state of affairs does it represent? what is its ‘law of projection’ (2/25)? It cannot of course be another brain state for the same question then arises. The identification of the physical embodiment of a given mental state, supposing such a thing to be technically possible, presupposes the determination of its functional role in relation to the subject’s behaviour: For every hypothesis concerning a subject’s mental states we must be able to conceive of its making a difference to possible behaviour (6/58); and that means that possible behaviour must fix all the ‘facts’ concerning the subject’s mental states. The state of the subject’s brain could fix those very same facts only by being wholly and essentially the determinant of behaviour.


94. The subject’s behaviour manifests his beliefs, desires and intentions. If, per impossibile, we were able to form a conception of the essential, identifying features of a mental state independently of the determination of its functional role in relation to behaviour, then we should construe behaviour only as the combined ‘effect’ of several ‘independent’ causes, on the ‘hypothetico-deductive’ model of explanation. Beliefs, desires and intentions would constitute the ‘variables’ whose combination we posit as the simplest explanation for that given effect. Now this account is nonsensical; yet it seems easier to comprehend that counternecessaryfactual than the alternative which rejects its antecedent: Notwithstanding the ‘technical possibility’ of identifying the physical embodiments of mental states, beliefs, desires and intentions are really nothing ‘in themselves’. Their whole being consists in their role in explaining behaviour. But nor can we hope to define any mental state in terms of the behaviour which ‘manifests’ it, for mental states manifest themselves only in combination; and we can conceive of any given piece of behaviour ‘resulting’ from different ‘combinations’. Only these ‘combinations’ are not combinations of anything! Now the attempt to ‘objectify’ psychological states is itself a symptom of the ego illusion. Rejecting that illusion means resisting the temptation to ask what beliefs, desires and intentions are ‘in themselves’: To express one’s mental states, or to attribute mental states to others is an activity; our task is only to comprehend the nature of that activity.


95. The first step towards comprehension involves shaking free from the hypothetico-deductive model. Alien beings whose science was far in advance of ours might conceivably give up talk of our beliefs, desires and intentions and take to explaining our bodily movements in terms of physiological processes. From the point of view of the explanation and prediction of physical events psychological explanation is redundant. It indeed serves a certain heuristic purpose: Keeping track of explanation in the psychological mode will make it easier to find the physical ‘explanations’ which are all these beings are interested in. But admitting the possibility of psychological explanation carries no additional ‘theoretical commitments’: Positing beliefs, desires and intentions in order to explain behaviour does not entail the conceivability of an ‘alternative mode of access’ to those states as, for example, the explanation of reaction mechanisms in chemistry commits us to the independent existence of molecules, with the potential to manifest their states and doings in other ways besides what we observe as ‘chemical reactions’. (By contrast, when the atomic hypothesis was first put forward in order to ‘explain’ the fact that chemicals combined in constant proportions, there was not enough experimental evidence to decide its status as an ‘existence hypothesis’.)


96. Nevertheless, those alien beings would be missing something. They could predict all our bodily movements, but would not be in a position to evaluate them as actions; as something to be assessed as ‘appropriate, apt or successful’ (6/58); they would not see our bodily movements as those of ‘rational agents’. It would still be possible for them to ‘offer’ us what we should construe as ‘reasons for action’; but for them it would be only a meaningless barrage of noise, calculated to bring about various kinds of bodily movement. Of course, this fantasy only makes sense if we allow these beings a conception of one another’s beliefs, desires and intentions; otherwise we cannot see them as ‘rational agents’. Thus, the activity of attributing mental states is no mere ‘heuristic substitute’ for the theoretical prediction of bodily movements, but the essential activity of rational agents. Moreover, even with unlimited knowledge at one’s disposal, holding back from psychological explanation in a particular case means ceasing to perceive a ‘reality’ which is indeed there; a foregoing of possible explanation.


97. That account of psychological explanation presupposes a metaphysic of action. For the phenomenalist has only these options: First, one might attribute ‘private representations’ to other ‘transcendental egos’, on the basis of ‘induction from one’s own case’. Other persons then become Leibnizian monads; subjects of experience whose mental states transcend the surview of one another’s possible experience. From the point of view of a metaphysic of experience, that seeming ‘conception’ transcends the ‘bounds of sense’; for it entails the construction of representations (of another monad’s representations) whose own ‘meaning’ cannot be given by the conditions for their application to the possible experience of the transcendental subject which constructs them (5/45-6). Alternatively, the ‘meaning’ of psychological concepts must change from the first to the third person, where they no longer name ‘private representations’ but merely ‘explain behaviour’. ‘Dual meaning’ is in itself not an objection. However, on pain of failure to cohere with the use of psychological concepts to describe one’s own mental states, ‘behaviour’ must be conceived as guaranteeing a uniquely determined psychological description, in just the same way that the very ‘structure’ of experience must ‘determine’ its own interpretation (4/43); they must be definable in terms of possible behaviour (contra 94). The reason in this case is that, since embodiment as an agent constitutes a possible experience, and since the a priori (91) self-ascription of mental states cannot allow room for any question of ‘interpretation’, a one-to-one correspondence between the two modes of self-ascription qua ‘empirical’ and qua ‘transcendental’ subject could be preserved only by locating in behaviour a surrogate for the mental ‘states of affairs’ of a priori introspection.


11. Solipsism and self-consciousness


98. Within the metaphysic of experience, ‘What the solipsist means is quite correct; only it cannot be said...’ (Tractatus 5.62). Attempting to say what solipsism ‘means’ is the illusion of empirical idealism: ‘My own mind and its mental contents are all there really is; the only thing that exists in reality is me’ (3/28). The only way to render the ‘expression’ of the idealist’s metaphysical vision consistent entails the destruction of the very materials of expression; the destruction of the empirical idealist’s ‘me’ and ‘mine’ (4/38). There remains only ‘this...’ (4/42); and within the ‘this...’, the enduring subject of consciousness and the empirical reality which it perceives are equally real. But saying ‘this... is all there is’ is not saying anything. If ‘this...’ is everything then, in attempting to refer to the ‘this...’, one must refer to everything, and that is tantamount to referring to nothing; like a pointing gesture directed at every direction at once. This is this, and all there is is all there is: The ‘self of solipsism’ having ‘shrunk to a ‘point without extension’, there remains only ‘the reality co-ordinated with it’ (Tractatus 5.64).


99. However, the discovery of the ‘dual meaning’ of psychological concepts (10/97) complicates the picture. Wherever we find two meanings, we are entitled to distinguish between them in language, by the use of different words. If a belief’ means something different when! have it from a ‘belief’ which belongs to someone else, then I should be able to allude to that ‘difference’ by calling my own beliefs by a different name, say, ‘real beliefs’. One might protest that, given the way things stand in the actual world, no additional information is entailed by a belief’s being a ‘real’ belief, except simply that it is ‘mine’; for it so happens that the ‘possible experience’ of my ‘embodiment as an agent’ (10/97) is realized. But the protest overlooks the following point: Had I not been embodied as an agent, the attribution of ‘real’ belief would indeed have ‘conveyed information’, viz., that the existence of that psychological state had no behavioural consequences. Since a language satisfying the Tractatus criterion of ‘representational adequacy’ (4.04) must be capable of depicting every possible state of affairs, and consequently the possibility of my not being embodied as an agent, one must indeed be ‘saying something’ when one describes a belief as a ‘real belief’. From that it would appear to follow that what the solipsist means can indeed be said: Only my beliefs are real beliefs.


100. The solipsist is ‘all alone in the world’; but only by virtue of the existence of a possibility which is not in fact realized. His ‘aloneness’ is a metaphysical, rather than a physical fact: His body reduces to an ‘outer shell’, a mere object of his experience (5/45), while other persons become nothing but ‘outer shells’ with nothing ‘inside’ (10/97). This metaphor of a ‘purely inner’ self confronted by ‘purely outer’ selves finds its realization in the project of a metaphysic of experience. We should therefore expect that any phenomenological attack upon the solipsist’s seeming ‘predicament’ will necessarily fall short of its objective (8/73). The examination of one such attempt will, however, once more prove instructive; and in addition provides materials which we are able to put to our own use.


101. In Ch.3 of Individuals, Strawson presents an argument tantamount to a consolation for the phenomenalist who sees himself threatened by the solipsistic consequences of his metaphysic. Its ‘phenomenological’ intention is nowhere explicitly stated, but emerges as the only consistent interpretation. For the argument manifestly avoids addressing the ego illusion; and in the context of the dialectic ‘failing to address the ego illusion is to embrace it’ (4/42). Indeed, Strawson’s hypothesis, in Ch.2, of a ‘sound world’ accepts the coherence of the Kantian model of a ‘disembodied passive observer’. (One might note also the supplementary evidence provided by his interpretation of Kant in The Bounds of Sense; cf. 7/63.) I shall state the argument in my own terms. Its strategy takes a form analogous to Kant’s ‘refutation of idealism’: One way of expressing what the solipsist ‘means’ is to assert that one’s embodiment as a person is ‘only a contingent fact of experience’, while other persons are necessarily embodied (99). But if it could be shown that this apparent contingency’ is an illusion, the same thing would have been done for the problem of solipsism as Kant does for empirical idealism; for the empirical idealist is under the analogous illusion that it is a ‘mere contingency’ that ‘his’ experience has the character of ‘perception of an objective world’. We shall see how this approach does not aim to undermine the phenomenalist’s project, but only to expand its ‘terms of reference’ (8/75).


102. Now Kant argued that only if ‘experience’ has the character of ‘perception of an objective world’ could there be said to be a ‘subject of experience’, as the empirical idealist takes himself to be. Similarly, a necessary condition for self-consciousness as opposed to mere consciousness, for having the conception of one’s own ‘self’ as the subject of experience as opposed to merely being its transcendental ‘subject’, is that one should operate with psychological concepts which apply indifferently to the first and third person: It is not a ‘merely contingent’ fact that no additional information is conveyed by qualifying the terms ‘experience’, belief’ etc. with the term ‘real’, but a condition for being that ‘self-conscious’ subject which the solipsist takes himself to be. Consider a possible state of affairs in which ‘real’ belief differs from mere belief in having no behavioural consequences. The ‘redundant qualification’ in: ‘I have the real belief that...’ is no longer the term ‘real’ but rather the term ‘I’. Indeed, the term ‘I’ remains a ‘redundant qualification’ in a world where the subject has the experience of being embodied as an agent but fails to recognize that the existence of behavioural consequences of his own psychological states is no ‘merely contingent’ fact but carries the same necessity as the consequences of attributing those states to others.


103. Strawson’s argument shows that, contrary to what we first supposed, the verbal manoeuvre of introducing the term ‘real’ does not after all enable us to ‘say what the solipsist means’. The very possibility whose recognition appeared to justify the use of the new term does so only by simultaneously undermining its intended point: That failure to be embodied as an agent constitutes a possible experience indeed raises the question of. the ‘representational adequacy’ of language (99). But now we have discovered that the demand of adequacy must be satisfied in the following way: One omits reference to ‘I’ and simultaneously introduces the qualification ‘real...’ in just those worlds where that possibility is realized or where the subject ‘fails to recognize’ that his own embodiment is as ‘necessary’ as that of other subjects. As a consequence, the statement: ‘Only my beliefs are real beliefs’ reduces once more to a mere tautology.


104. True to his phenomenalism, Strawson regards becoming ‘disembodied’ as a ‘possible experience’ (Individuals P.115-6). There might still be a use for ‘I’, moreover, provided that I retained a memory of my former, embodied self. The constraint of ‘indifference’ in the consequences of applying psychological concepts to the first and third person (102) means, therefore that Strawson must allow that the disembodiment of other persons could also be an experience of mine: It is the difficulties which this possibility raises for our conception of the ‘necessity’ of the ‘behavioural consequences’ of psychological states, rather than, as Strawson suggests, the difficulty in specifying how long the mere memory of having once had a body could keep alive the spark of self-consciousness, which force us to impose a time-limit on the survival of a disembodied ‘I’. How one could possibly calculate it remains obscure. In any case, the possibility of disembodiment means that Strawson’s argument cannot be used to establish the supervenience of facts about mental states upon facts about the physical world, which is the least stringent form of materialism. The incoherence of ‘solipsism’ means only the rejection of the ‘mind-body’ problem as traditionally conceived, It is not merely a ‘matter of fact’ that I am an embodied person, as would be the case if I were an ‘immaterial self’ which only happened to ‘inhabit’ a ‘body’, but a necessary condition for my having the very concept of my own ‘self’. We should note, however, that the familiar, Cartesian form of that problem had already been rejected by Kant in the ‘Paralogisms of Pure Reason’ (Critique of Pure Reason, A341-405 and B399-432).


105. The striking paradox of Strawson’s argument is that the metaphysical vision of solipsism appears as a heightened form of self-consciousness. That is of course far from being an objection: For his dialectical point consists in the reminder that no-one who has ever thought about the problem of solipsism has been less than a self-conscious person; a necessary condition for being a true ‘solipsist’ is not experiencing its ‘metaphysical vision’. But beyond that, we should note that, for the purposes of the argument, the meaning of ‘self-conscious’ requires no further specification than this: The self- conscious ‘I’, as opposed to the merely conscious transcendental subject, is simply that which applies psychological predicates to itself in the ‘indifferent’ sense (102, 4). The problem of explicating the meaning of ‘self-conscious’ rests wholly on the account of the indifferent ‘necessity’ of ‘behavioural consequences’.


106. All we learn from the argument is that ‘what solipsism means’ cannot, after all, be said (103). But phenomenology leaves unharmed what it is that the solipsist means; that ‘the world is my world’ (Tractatus 5.62). Just as Kant’s ‘refutation’ only sharpens the ‘meaning’ of idealism (4/42), so Strawson’s argument amounts to no more than a sophisticated reformulation of what the solipsist ‘means’. For solipsism ‘makes itself manifest’ (Tractatus 5.62) in the acceptance of the necessity of the phenomenalist’s project: That my relation to other persons is that of a ‘purely inner’ self confronted by ‘purely outer’ selves cannot be said, but it can still be shown (100).


107. How do things stand in the metaphysic of action? Being a person does not reduce to a ‘feature of my experience’, nor even a ‘necessary feature’. My behaviour shows that I am a person (108). I may indeed imagine myself ‘disembodied’, but such a state of affairs cannot be coherently thought out. If someone were to relate to us an ‘experience of disembodiment’, we should not be able to make sense of the ‘hypothesis’ that what he describes really happened. For the fact that he ‘disappeared’ and ‘reappeared’ at the right times, that he seems to have ‘witnessed’ certain events while disembodied does not entail genuine disembodiment; these phenomena could equally be described as ‘going out of existence’ and ‘coming back into existence’ (i.e. as a duplicate of the original person), possessing ‘seeming memories’ which by accident turn out to describe something true. And because embodiment is the issue only because it makes agency possible, we cannot even allow that the person might have only metamorphosed into a passive, intelligent but invisible gas. For the gas might equally be described as the recording medium which served to convey information of the seemingly remembered events to the person at the moment of his ‘coming back into existence’. Now the experience of ‘disembodiment’ must be reinterpreted because it can be; and because it could be coherently taken at its face value only within a metaphysic of experience (8/78).


108. If there is a distinction to be drawn between the thinking subject who is only ‘, conscious’ and the subject who is ‘self-conscious’ it will be manifested in different modes of activity, different ways of being an agent. The logical schema for the conditions for the possibility of self-consciousness translates from the metaphysic of experience to the metaphysic of action (turning the tables on phenomenology; 8/72). Persons are agents who are self-conscious. To be a non-self-conscious agent is to fail to ever ascribe psychological attributes ‘indifferently’ (102) to oneself and to others; to be a person is to ascribe psychological attributes indifferently to oneself and to other indifferent ascribers of psychological attributes. (Persons are persons in mutual recognition.) since both persons and mere agents are ‘necessarily embodied’ (107), the crucial consequences of a person’s ‘indifferently ascribing psychological attributes’ must amount to something more than the recognition only of an ascription’s indifferent ‘necessary behavioural consequences’ (102-5; note the new sense which the rejection of phenomenalism gives to ‘necessary’) by the non-self-conscious agent, The meaning of a genuine, practical solipsism, of being a ‘non-self-conscious agent’, will depend upon the explication of that ‘something more’; as will the significance of the reminder that no philosopher has ever lacked the attributes of a person (105).


12. Mind and body


109. The rejection of the phenomenologist’s ‘possibility of disembodiment, (11/107) is another example of a necessity entailed by a metaphysic of action: The self is ‘necessarily embodied’. But we should note first that the whole sense of this ‘necessity’ consists in its rejection of the ‘illusion of a possibility’ (8/77). For all we know, there might be other possibilities of relation between the agent and his body which we have simply not foreseen (compare 6/60); it is only the imagined ‘experience of disembodiment, which we ‘refuse to take at face value’; just as the refusal to take at face value the imagined experience of spatially unrelated spaces (8/78) does not rule out unforeseen developments in physics. We are in no position to dictate the ‘necessary structure of reality’ in any sense which implies an a priori anticipation of the results of empirical inquiry. Secondly, the ‘necessity’ of embodiment may be called a ‘materialism’; but in no sense which implies an identity between a mental ‘inner’ and a physical or material ‘outer’. Any such reference to the concept of identity can only be motivated by the ego illusion; and is an attempt to state what would be a truth of transcendent metaphysics.


110. Materialism has sometimes been identified with the demand that the ‘whole of reality be accessible to science’. That slogan fails to distinguish a thesis which is certainly false from one which, although highly contentious, is not open to philosophical objection: We reject the thesis that all explanation can be reduced to physical explanation; the ‘science’ of psychology, for example, is not a ‘heuristic substitute’ for the physiological explanation of bodily movements (10/96), but retains an autonomy from physiology (the same might indeed prove true of the relation between physiology and physics). Autonomy in no way threatens the universality of the more ‘fundamental’ science, its understanding of its own chosen objects. By contrast, the so-called ‘pseudo-sciences’ of paraphysics and parapsychology challenge the very policy of seeking universality; implying that there is no guarantee that the objects of physics will not prove ultimately recalcitrant to any recognizably ‘physical’ investigation. The strength of one’s commitment to the accessibility of reality to science can be measured by the degree of apparent ‘success’ in the pseudo-sciences which would persuade one to abandon that policy. Insofar as a Cartesian ‘science’ of immaterial substances threatened the universality of physics, its rejection would simply be the consequence of our adopting the policy of a thoroughgoing scientific materialism.


111. The materialism entailed by a metaphysic of action remains neutral on the question of the validity of scientific materialism. By contrast, the whole point of the metaphysic of identity consists in its attempt to fill a perceived ‘lacuna’ in the framework of scientific materialism; the whole- hearted commitment to science ‘does not go far enough’. Let us suppose then that the science of psychology and the science of physiology have both advanced to the point where there is no longer any mystery about how they are able to deal in their different ways with the very same subject matter, viz., living persons. We should have finally ruled out the possibility of ‘immaterial substances’ whose ‘interactions’ with matter remain scientifically inexplicable. It appears that one could still raise the following question: Is consciousness in itself something which possesses a reality independent from all that is disclosed to the third-person point of view of the scientific investigator, the ‘accessibility’ of psychological phenomena to science being guaranteed only by non-causal, psychophysical laws? or is the relation in some sense an identity? To argue, on the assumption of scientific materialism, merely that events ‘under psychological descriptions’ must be identical with events ‘under physical descriptions’ (see, for example, Davidson ‘Mental Events’) appears to leave open the possibility of a parallel reality of non-interacting ‘Cartesian mental events’. In that case, there would remain something ‘inaccessible’, not in any sense which threatened science, but by virtue of the fact that the ‘very thing itself’ as it presents itself to me in consciousness could be grasped by science only ‘at a distance’, in the form of a ‘projection’ of its ‘image’ onto the material world; onto the human body, its behaviour and speech, and its brain and innards.


112. The one thing science could not explain would be the fact of parallelism and the existence of psychophysical laws. But that apparent ‘limit’ to the reach of science proves to be the condition for the very possibility of its ‘reaching’ psychological phenomena. It is for that reason the one thing which one might allow to remain exempt from the policy of not admitting ‘limits’ to science: Scientific materialism provides no grounds for a metaphysic of identity, because the denial of identity ‘cannot have any consequences for the practice of science’. Alternatively, one may still wish to argue for identity, on the basis of principles borrowed from science: Thus, the existence of both a ‘material’ and an ‘immaterial’ reality goes against the principle of Occam’s Razor; that one should posit as few hypothetical entities as is consistent with the (material?) phenomena. Alternatively, one might appeal to a principle of causality: If the material realm is ‘complete in itself’ with respect to the causal relations between its phenomena, so that the existence or non-existence of an ‘immaterial’ realm makes no difference to those relations, then to conceive of the bare addition, uncaused, of such an ‘immaterial’ realm together with ‘laws’ which ‘correlate’ the two realms is absurd. Now, the use of these ‘principles’ costs nothing; nothing turns upon a decision as to the ‘correctness’ or ‘incorrectness’ of their application to the question of identity except what we are to say about just that question and no other. So let us suppose that one philosopher regards their application as ‘correct’ and another regards it as ‘incorrect’: We have here merely an irresolvable ‘dispute’ between rival transcendent metaphysicians (2/14).


113. In the very same manner, the transcendent metaphysician ‘borrows’ a principle of induction in order to ground his belief in the ‘existence of other minds’, and thus ‘refute solipsism’. My reason for believing that the ‘immaterial realm’ reaches beyond the confines of my own consciousness is that, qua member of the material realm, I am no differently constituted from anyone else; what is true of my case is therefore true of others too. Now to this Wittgenstein jokingly replies: ‘How can I generalize from the one case so irresponsibly?’ (Philosophical Investigations I/293). The joke is in pretending to be a rival ‘transcendent metaphysician’ who ‘rejects’ the argument on the grounds that, in this case, ‘the inductive base is too narrow’.


114. If both identity and the denial of identity are transcendent metaphysics, how Can one avoid being a transcendent metaphysician? The question appears rhetorical; but the answer it immediately suggests is false: ‘Avoid asking transcendent metaphysical questions’. That is something the ‘anti-metaphysical’ philosopher might officially ‘avoid’; but the thought that avoidance can simply be a matter of ‘decision’ is self-deception. There is another way to escape the dilemma: to reject the illusion which gives rise to the illusion of a ‘question’.


115. There is something which presents itself to me. And there is also something which presents itself to the world. Now I ask: Are these things the same or different? Are those very states of my consciousness which I perceive from my inner viewpoint also in principle perceivable from the outer viewpoint of science? or is all an investigator could ever perceive only their ‘material projection’ (Ill)? Suppose that the identity is denied: There arises the further question whether what one could perceive of another person’s states of consciousness is the projection of something which is ‘inner’ for him or only a bare ‘outer’ (113). Suppose instead that the identity is granted: There still remains the question whether what one could perceive is an ‘inner’ or only a bare ‘outer’ semblance of a non-existent ‘inner’ (as, say, one may question whether what one seems to ‘see’ through the ‘window’ of a house is something ‘seen through a window’ or only a painted facade). The unobstructed ease with which one may interchange the notions of ‘inner’ and ‘projection of an inner’ shows that there is nothing to choose between ‘identity’ and ‘non-identity’; the ‘question of identity’ is not a question.


116. If there is ‘something which presents itself to me’ in any sense which raises the question of identity then that question ‘is not a question’; therefore nothing ‘presents’ itself to me in that sense. But suppose instead that I deny that the ‘something which presents itself to me’ raises the question of identity, on the grounds that the material and the immaterial are so utterly different that we ‘cannot conceive of what sense an assertion of identity would have’. That amounts to no more than the vehement assertion of non-identity. The question remains a question despite the fact that its answer appears a foregone conclusion.


117. That which ‘presents itself to me’ appears in a naive and a sophisticated form: The naive form appeared at the very first stage of the idealist’s ‘story’, at a point before the Cartesian ‘mental content’ expanded to engulf the whole of ‘outer reality’ (3/28). The rejection of the ‘question of identity’ which that raises becomes merely a stage in the internal development of idealism (4/42). However, there appears a second, ‘sophisticated’ form at the very point where the ego illusion is ‘constructively’ rejected by a metaphysic of action: One accepts the arguments, accepts the ‘incoherence’ of phenomenalism; but the ‘this‘ ‘obstinately refuses to go away’ (6/56). (That ‘position’ may be illuminatingly compared with the ‘refusal to either accept or renounce the phenomenalist metaphysic’ prompted by the issue of ‘things in themselves’; 5/50.) Being a partisan of either ‘identity’ or ‘non-identity’- it does not matter which-serves to express one’s continued recognition of the ‘this...’. The rejection of the seeming ‘question’ is one more support knocked away.


118. There remains a last refuge for the ‘this...’: as an a priori argument against the very possibility of a scientific materialism (110). Taking a stand on this would at least have real consequences, by contrast with the dispute between ‘identity’ and ‘non-identity’ conducted within the framework of scientific materialism, which merely evaporates into nothing. still, one remains discerning in one’s choice of pseudo-sciences: While being prepared to accept that no mystery in all of reality is equal to the mystery of the ‘this...’, one does not wish to commit intellectual suicide. But I shall leave the elaboration of the details of that metaphysic to the reader’s imagination.


13. Thought and language


119. ‘We are not analysing a phenomenon (e.g. thought) but a concept (e.g. that of thinking), and therefore the use of a word. So it may look as if what we were doing were Nominalism. Nominalists make the mistake of interpreting all words as names, and so of not really describing their use, but only, so to speak, giving a paper draft on such a description’ (Wittgenstein Philosophical Investigations I/383). — From the point of view of a metaphysics of meaning, the project of comprehending the possibility of thought, there at first appears justice in the charge of ‘Nominalism’. To attempt to reduce the dialectic to the investigation of ‘how certain words are used in the language’, or how, for some unknown reason, they end up getting ‘misused’, or alternatively, to recast the argument into an inquiry into the ‘form’ of a ‘theory of meaning’ with the intention of rendering the investigation ‘systematic’ (admittedly a project which Wittgenstein would never have undertaken) destroys the meaning of our discoveries concerning the relation between mind and reality. One may indeed set oneself the project of constructing a dialectic of language — we shall make a start on that — provided one treats it only as an extension of a metaphysical project which has already found its own independent reality: The investigation of language merely provides new, illuminating ways for the metaphysical illusion to manifest itself. However, it is not clear what Wittgenstein meant in his highly compressed remark. For the sake of dialectic and the self-understanding of our own project, it is worth seeing how much can be analysed out of it.


120. ‘We are not analysing a phenomenon... but a concept...’. — If one generalizes this insight (as was surely intended) one obtains a definition of ‘philosophy’ which traces back to its very beginnings. Plato might have said: ‘Knowledge of what thinking is in itself is not knowledge of the phenomenon of thought but knowledge of an eternal form. Phenomena are deceptive and transient; and cannot therefore yield real, philosophical knowledge.’ Ignoring the disproportionate valuation placed on philosophical knowledge, Plato’s thought may be taken a stage further: In analysing the use of a word (‘recollecting’ knowledge of an ‘eternal form’) we are not analysing linguistic phenomena. We are neither linguists nor grammarians. Then what are we doing? For both Wittgenstein and, arguably, the Plato of the early, ‘Socratic’ dialogues, the ‘analysis’ of a concept consists, not in its ‘definition’ — an entry in a ‘philosophical dictionary’, conceived on the model of a lexicographers dictionary entry, only somehow ‘deeper’ — but in a dialectic. The philosopher’s knowledge of a concept is attained through mastery of that dialectic (although Plato and Wittgenstein picture this ‘end state’ in different ways). A putative ‘definition’ of a concept is not its ‘analysis’ but only a contribution towards the dialectic; attempting to give definitions is only one of several ways, the ‘Socratic method’, of playing the analysis game.


121. Thought, the chosen example, is a phenomenon. And using the word ‘thought’ is using it to refer to that phenomenon. Both these assertions are consistent with Wittgenstein’s remark. (‘Phenomenon’ simply means here a ‘distinguishable feature of reality’.) But there are as many categories of phenomena as there are kinds of uses of the substantive. The category of phenomenon to which thought belongs is such that one can get to know what thought is ‘in itself’, what thought ‘really’ is, not by getting to know the phenomenon — we already know all the relevant facts — but only by working through the analysis of the concept. (The first thing one learns is that ‘thought’ is not a process concept but rather the concept of an activity; that of thinking: ‘We are not analysing... thought but thinking...’.) Generalizing, we may say that the concepts which are ‘philosophically problematic’ are just those which we do not ‘really’ know even though we may be intimately acquainted with the phenomena to which the concept-words refer; concepts which we cannot get to know ‘in themselves’ merely by getting to know or even ‘analysing’, their phenomena.


122. Wittgenstein and Plato do not disagree about what constitutes a ‘philosophical analysis’ of a concept. Neither thinks he is analysing the phenomenon to which the concept-word refers. (That is indeed a project which a ‘Nominalist’ might erroneously undertake.) Moreover, a Wittgensteinian scepticism concerning the possibility of a ‘complete’ analysis (cf. the dialectic of ‘family resemblance’ and ‘point’: Philosophical Investigations I/65-78 and 561-570) can be illuminatingly read into the inconclusiveness of the Socratic dialogues. But Plato and Wittgenstein have very different notions about what concepts ‘are in themselves’. Here it is Plato and not Wittgenstein who is the ‘Nominalist’ (according to the first part of Wittgenstein’s definition, rather than traditional philosophical classifications): Concept words are interpreted as names of eternal forms. That raises — and indeed is meant to raise — questions concerning our cognitive relation to the eternal forms, the explanatory relation between forms and phenomena, and the possibility of forming a proposition from the combination of concept-words. Wittgenstein circumvents these problems by identifying concepts with the use of concept-words: ‘We are... analysing... a concept... and therefore the use of a word.’ This project involves the refusal to accept a model of naming which leads us to ask the ‘wrong kinds of question’ about the philosophically problematic concepts; questions which lead to projects of transcendent metaphysics.


123. ‘Analysing a concept, not a phenomenon’ and ‘analysing a concept and therefore the use of a word’ each constitutes a rejection of a different possibility of ‘Nominalist’ illusion, of making ‘the mistake of interpreting all words as names’. Wittgenstein’s observation that ‘it may look as if what we were doing were Nominalism’ now appears doubly ironical. Indeed a third, and the only overt irony is that ‘Nominalism’ is in a sense trivially true: The easiest way to be a Nominalist is simply to assert that ‘thought’ is the name of thought etc. (‘... not really describing their use but only... giving a paper draft on such a description.’). By contrast, one might call Wittgenstein’s position a methodological nominalism.


124. What then is the real nature of concepts as they are in themselves? Methodological nominalism goes no further than to reject the Platonic concept of ‘eternal forms’, along with its transcendent metaphysics. It is not a theory of what concepts ‘really are’ but rather a refusal to answer the question: a refusal to say anything substantial about the concept of ‘concept’ except on the ground level of the dialectic of analysis. Indeed, on that level it is hardly ever necessary to employ the word ‘concept’; one talks merely of how this or that word is ‘used’. (‘Concept’ serves as a convenient means of abbreviation where ‘word’ won’t do: e.g., ‘What do these concepts have in common?’) But wasn’t the slogan: ‘Meaning is use’ intended to have more far-reaching consequences than that? Having rejected Platonism, agreeing only with its rejection of the identification of philosophy with any of our inquiries into phenomena, the most important element remains unaccounted for.


125. It is not by accident that Wittgenstein chooses the example of a psychological concept. We have seen how the phenomenalist’s transcendental construction of an ‘external world’ necessarily falls short of comprehending the reality of ‘inner’ thoughts and feelings. (‘The ‘private object’ is rejected, only to be replaced by the private content’; 10/91.) If the ‘inner’ is a ‘reality’ then ‘knowledge of the inner’ cannot be a priori; there must be a place for the defeasibility of judgement even here, for example, in showing by one’s actions that one has forgotten the meaning of a word. Talk of states of consciousness as ‘inner processes’ (121) helps camouflage the breakdown of phenomenalism; the ‘inner’ shores up its ‘reality’ by borrowing concepts of the ‘outer’. But rejecting the ‘inner process’ means nothing in the absence of a rejection of the phenomenalist’s illusion. (The phenomenalist might say: ‘It’s got to be an inner process; if that conflicts with usage then usage is wrong.’) Only a full-blooded sense of ‘meaning is use’ which transcends a simple ‘methodological nominalism’ suffices to reject the illusion: That sense is embodied in the dialectic of the ‘private language’ (6/58): We can find nothing in the meaning of any concept which cannot be located in all that suffices to give it a use in a public language, The ‘private’, superadded meaning is an illusion: That is Wittgenstein’s own contribution to the question: ‘What are concepts?’


126. This way of putting the question of the ‘relation between thought and language’ harmonizes with the dialectic of illusion. There is nothing in Wittgenstein’s remark or in its presuppositions to which we need object. But there remains another, less dialectical way of taking the remark which serves as the prolegomenon to the project of ‘constructing a theory of meaning’: We are to consider two models for the expression of thought in language. In the first, concepts are provided with ‘markers’ which we use for the communication of our thoughts: To speak is to ‘encode’ one’s concepts into words; understanding speech involves ‘decoding’ words into concepts. In the second, grasp of a concept is just grasp of the usage of a word; there is no ‘encoding’ or ‘decoding’ involved in communication. The first model entails a form of scepticism: I can never know whether another person really ‘understands’ what I am saying, or that I really ‘understand’ what he is saying, because I cannot be sure which of ‘his’ concepts he associates with each of ‘our’ words. Because that conclusion is obviously ‘unacceptable’, we must find a way to reject the model. Now, that model purported to explain something which its bare rejection leaves ‘unexplained’; how words are given their meanings: Each language user carries within himself a personal coding manual which translates the ‘language of thought’ into a public language. To justify the rejection it seems we must provide an alternative ‘theory’: Words get their meanings by being associated with rules for their use; just as in a game of chess the pieces derive their significance simply from the rules of the game. A ‘theory of meaning’ for concepts would somehow ‘explain the rules’, and thus account for the ‘significance’ of a concept.


127. The comparison of the two ‘models’ of communication at first appears to identify correctly the illusion of a ‘private language’. Of course that only means that we can read it as doing so; the illusion cannot be identified by any simple ‘sketch’, however suggestive. But then we encounter the first jarring note. We are dealing with rival ‘explanations’, alternative ‘theories of meaning’. Now whereas the only alternative to being under illusion is not being under it, introduction of the notion of ‘explanation’ raises the question whether a ‘theory’ is what we are seeking. It might indeed be; but the choice of alternatives is no longer exclusive. Once could take it in that way only by reading too much into the original sketch. The risk now is that, having failed to correctly locate the illusion of a ‘private, superadded meaning’, too much will now be needed to justify its rejection. Before even having resolved the question whether the analogy between the ‘rules’ of language and the ‘rules’ of a game is of the right kind to justify our envisaged project, we have already settled that the project ‘must be feasible’; there is ‘no other possibility’.


14. Does thought entail the possession of a language?


128. Thought and language mediate the relation between mind and reality. We know that their own relation must depend upon that mediating role. Yet the question of the validity of the phenomenalist’s metaphysic cannot be reduced to the simple form: ‘Does thought entail the possession of language?’ To represent the nature of the relation between thought and language as turning upon the acceptability of the ‘encoding/decoding’ model as an ‘explanation of how words get their meanings’ invites the unwarranted construal of the rejection of the ego illusion as the demand for a ‘theory of meaning’ (13/126-7); while the alternative project which the question immediately suggests, that of a phenomenological demonstration that ‘the experience of using a language is necessary for objective experience’ (or: ‘... necessary for F’, where F constitutes an ‘essential feature’ of a certain kind of ‘objective experience’; cf. 8/72, 5), proves acceptable to the phenomenalist. still, the ‘simple’ question seems reasonable, and we ought to be able to answer it.


129. Each of the rejected projects may be seen as an example of a misinterpretation of the ‘private language argument’ (13/125). The point of that argument is just the rejection of the ego illusion (6/58); the pretence of ‘taking seriously’ the possibility of a ‘private language’ which ‘underlies’ the use of a public language, and then deducing its incoherent consequences, is merely Wittgenstein’s own choice of dialectical strategy. By contrast, our way of expressing that same point pretends to ‘take seriously’ the possibility of existence as a ‘passive observer’, and then deduces its incoherent consequences. This second version does not even mention language but considers instead the question: ‘How is false belief possible?’ A false belief is necessarily something which ‘shows up’ in ‘inept, inappropriate or unsuccessful action’ (6/58). And what is thus ‘shown’ can be shown only to another person; for the question concerning the ‘objectivity’ of experience cannot be coherently construed as a question one could answer ‘for oneself’, with or without that other person’s ‘help’ (5/47-8, 50-3). It might now occur to the reader: Perhaps language is so far from being the question at issue in the dialectic of the ego illusion that being a language user is not necessary for thought!


130. Why can’t an animal have beliefs? We can conceive of the possibility of a being we thought was a non-language user (whether possessing animal or human form) turning out to ‘have beliefs’ by turning out to be a language user. That is not what the question means. Nor are we to consider the illusory notion that a ‘Cartesian ego’ might reside in the body of an animal. The question concerns rather the legitimacy of attributing beliefs to animals by way of ‘explaining’ their behaviour.


131. The question: ‘What is an explanation of an animal’s behaviour?’ is sometimes discussed as if it were perfectly clear what ‘explanation’ means in this context (by analogy with the psychological explanation of human behaviour) and that only the terms in which explanation should be couched are problematic; for example, whether we may attribute intentions, or only some less ‘loaded’ psychological states. The clarity decreases when one considers what ‘explaining human behaviour’ really involves. Understanding a person’s actions is only one component in a whole which includes reasoning with him, asking him to ‘explain’ himself, giving advice, criticizing, excusing, forgiving. In the light of this, one must then admit that ‘explaining’ animal behaviour is meant only as a notion of ‘explanation’ pared down to a useful device which, even though we lack God’s knowledge of the animal’s physiological states, enables some degree of prediction and control of its bodily movements. The form of such ‘explanation’ nevertheless ‘borrows’ at least some of its terms from the human case. It is as if this ‘device’ could be applied to human beings too, if only human life did not involve more than the detached, ‘prediction and control’ of one another’s bodily movements.


132. Note that it is consistent with this position to deny that a God’s physiological explanation of bodily movements would entail the more mundane psychological ‘explanations of behaviour’, which indeed fare no better in ‘predicting bodily movements’; in fact a great deal worse. The classification of all possibilities of bodily movement into what ‘interests’ us; the behaviour, involves the use of an intrinsically irreducible scheme of psychological concepts, the objectivity of which depends, as a systematic whole, upon the fact that we are able to train our eyes just on ‘what interests’, relaxing our focus on the meaningless complexities of bodily movement to the point of seeing only a sequence of ‘events’ (actions) with a connecting rationale; an ability which is indeed constitutive of our being rational agents (10/94-6).


133. The bare logical schema of the ‘irreducibility of psychological explanation to the physiological’ makes no mention of what especially distinguishes the explanation of human action. But that does not entail the coherence of an indifferent ‘pared down’ notion of psychological explanation which is able to find the necessary ‘connecting rationale’ in the ‘actions’ of both human beings and animals. One might, with equal justice, maintain that the ‘behaviour’ of animals is fundamentally inexplicable; that there is no real ‘explanation’ of the phenomena of animal life short of the physiological explanation of bodily movements, no ‘connecting rationale’ between their seeming ‘actions’. We merely make do with either teleological explanation of species’ behaviour (for example, why certain birds build their nests in a particular way), or with merely fanciful psychological assertions (‘the dog wants to eat and believes that its dinner is ready’) ‘justified’ on pragmatic grounds but strictly false, or else near-empty assertions that ‘action x was a response to the stimulus y’, whose generalization in the form of ‘stimulus-response’ definitions of psychological concepts amounts to pure equivocation. The argument for such a position is this: Cut out language and you cut out the very possibility of explanation in the psychological mode; for at the basis of psychological explanation lies the possibility of asking an agent to explain what he is doing.


134. One mistaken way of admitting the limitations to the ‘explanation’ of animal behaviour would be to allow that the attribution of beliefs is ‘fanciful’; but only because the linguistic expression of belief is too ‘fine- grained’ for whatever it is that the animal has; as if, lacking language, the animal is capable only of ‘vague beliefs’. This concept of ‘vague belief’ has been coined just to save a theory; and that does not suffice to justify its objectivity. (‘Having a vague belief’ is not the same thing as a person’s being vague about his beliefs; nor is it a belief which can be expressed only in vague language.) Indeed, the ‘difference’ between the assertion that animals can have only ‘vague beliefs’ and the assertion that all psychological statements about animals are ‘simply false’, though sometimes ‘useful’, becomes merely verbal: Animals do not have beliefs.


135. That conclusion appears a startling paradox. For we do ascribe ‘goals’ to animals, attribute the possession of ‘information’ concerning their environment. Surely we know what that means and what it is that it explains. But remember that the denial of the possibility of explanation was maintained only ‘with equal justice’. One might compare that denial with the claim that the phemomenalist’s ‘objective world’ is equally a ‘dream world’ and therefore neither (5/47), and the claim that the ‘identity’ between ‘consciousness’ and ‘the material’ is equally a ‘non-identity’ and therefore neither (12/112, 115-6). Now both persons and inanimate objects present themselves for ‘explanation’ as individuals; persons ‘explain themselves’ (133), while objects can be taken apart and examined. Animals fall into neither category: Fully understanding the workings of a machine enables us to predict all its movements. But with animals, the ‘machine’ is too complicated; the knowledge is beyond our powers. By contrast, understanding a person has a different point; persons remain unpredictable, but given any particular action there is something which counts, in a different sense, as ‘fully understanding’ it: Our understanding aims to match the person’s possible self-understanding; we explain him as he would, in the ideal case, ‘explain himself’. Animals are of course not ‘fully understood’ in this sense either. Nevertheless, we go through the motions of individual explanation, with our ‘pared-down’ vocabulary of ‘psychological’ concepts; then, oscillating between one notion of ‘individual explanation’ and the other, fall victim to the misconception that the individual is being ‘explained’; while in reality there is ‘neither explanation nor non-explanation’.


136. It is now time to explore the suspicion with which this chapter began: that the agent which the thinking subject ‘necessarily is’ is not necessarily an language user. In one of its forms, that suspicion has simply been removed: The abstract considerations appealed to in the refutation of the ‘passive observer’ model did not themselves mention language. The question whether the agent, qua genuine ‘thinking subject’, necessarily has a language was left undecided. But now we have decided it. However, there now appears a more serious form of the suspicion, which the previous considerations do nothing to allay: ‘The phenomenalist’s inability to answer the question: ‘How do you know that your ‘objective world’ is not merely a seemingly ‘objective’ world which you only seem to have been ‘determined’ to ‘construct’ on the basis of your seeming ‘experience’?’ is an inability to ‘explain’ the possibility of false belief’. And on that basis you reject phenomenalism. Your own response to the challenge of explaining false belief equates its possibility with the possibility of mistakes in the application of concepts ‘showing up’ in ‘inept, inappropriate or unsuccessful behaviour’. But how can a seeming ‘failure in action’ ever prove the existence of a false belief? Suppose the agent falls into a pit. And grant, for the sake of argument, that some independent criterion determines that this is not due to a loss of balance: Perhaps he meant to fall into the pit! We find this possibility conceivable for the sole reason that this is what the agent might say was his ‘real intention’. If false belief is possible, according to your supposed ‘criterion’ the agent’s statement of intention must not be allowed to undercut the possibility of our being certain that we have perceived a ‘failure of action’. There is only one way to secure that result: The facts of non-linguistic behaviour must alone suffice to determine a psychological interpretation. Whatever the agent says, as opposed to what he actually does, would then be, theoretically, irrelevant. If, for example, the agent did intend to fall into the pit then that could be proven on the basis of his other, non-linguistic behaviour.’ — In the light of the previous considerations, that conclusion could only be taken as a reductio ad absurdum of the project of a ‘metaphysic of action’, and consequently of its claim to have ‘refuted’ phenomenalism.


137. The reductio is motivated by a misinterpretation of the intention of the ‘sceptical’ question directed against the phenomenalist: It is not that there are two ‘possibilities’ which the phenomenalist does not know enough to be able to distinguish between. The phenomenalist’s notion of an ‘objective world’ is the notion of a ‘seemingly objective world’; and the ‘sceptical’ question serves merely as a reminder of that fact. However, having made the false start of mistakenly interpreting the question: ‘How is false belief possible?’ as a sceptical question demanding a ‘criterion’ which will ‘prove that false belief exists’, the reductio now applies it to behaviour and the answer is, unsurprisingly, found wanting: The addition of language would mean the defeasibility of any claim to have perceived a failure of action indicative of false belief; but then false belief cannot be proved. So language cannot after all be allowed to defeat the interpretation of behaviour. The fallacy here is in supposing that the question concerning the possibility of locating false belief in behaviour could be coherently taken in a ‘sceptical’ sense, rather than, as was intended, a means for directing attention to facts which the phenomenalist is unable to comprehend: What ‘makes experience objective’ can only be shown in action; and indeed such-and-such are the kinds of possibility (the ‘mistakes of action’) which ensure that there is in fact something to be shown. Now, the coherence of the sceptic’s claim, based upon the ‘conceivability of things turning out otherwise’, has been considered elsewhere and rejected (9/86-9). We need only point out that there is such a thing as covering up a mistake of action by asserting that one really intended to do what one in fact did. But there is also such a thing as being found out.


15. Concept and object


138. Where does the investigation of language begin? Ideally, it begins with a question which would put all other philosophical questions about language into question; a question one must be able to answer before any other. Yet that question need not be the one which first occurs to us or which we most want to answer. Thus, in the dialectic of idealism, the question raised against the ‘possibility of knowledge’ assumes a grasp of the nature of that truth which our knowing purports to be of; while the question raised against the possibility of a genuine, ‘objective truth’ in turn assumes that judgement has at least something objective to aim at, that, even if there is no ‘objective truth’, false belief is possible (2/17). An analogous order of priority appears in the dialectic of language: The question with the most obvious interest to a dialectic of illusion, how to determine whether a sentence is ‘meaningful’ (1/6) assumes that we know what meaning is; and that in turn assumes that we know how to identify that whose meaning and meaningfulness has been put into question. To begin: What is a sentence?


139. A sentence is the ‘complete expression of a thought’. The grammarian’s rules of sentence formation presuppose that competent speakers will generally agree over what constitutes a sentence, on the basis of intuitions about what it takes to ‘express’ a thought. The brute fact of agreement and the grammarian’s success in matching the grammatical rules with the linguistic facts suffice to justify the project of constructing a grammar for a given language. But grammar does not tell us what a sentence ‘really is’. We learn more about the phenomenon but remain ignorant of the concept (13/121). We do not really know what a sentence is any more than we really know what thought is; indeed, ignorance of the nature of the sentence is ignorance of the nature of thought. Should one then attempt first to analyse the concept of a thought? The previous considerations concerning the relation between thought and language suggest that analysing thought in abstracto would be a futile project; we can determine what thought really is only if that involves determining the real nature of the expression of thought in language. The question now becomes: What makes certain strings of words the kind of ‘complete entity’ apt for expressing a thought? The question is equivalent to: How is grammar possible?


140. Why is the question about sentences ‘philosophical’ while, by contrast, the question what makes certain strings of letters the kind of ‘complete entity’ apt for inclusion in the linguistic resources for the expression of thoughts is clearly not? In any given language, some strings of letters cannot be used to make words simply because they are unpronounceable. That observation prompts the claim that the parallel answer in the case of sentences: ‘some strings of words are unmeanable’ is, by contrast, philosophically problematic. But the reply overshoots the mark: We can construct unmeanable sentences too! To admit the immediate relevance of the question which that raises would trade the project of explaining sentencehood for the search for a ‘criterion of meaningfulness’; and whatever that search turns up cannot claim any authority if it cannot even say what a sentence is. Instead of attempting to explain its philosophical import, it will prove a better strategy to wait and see how the ‘interest and fruitfulness’ (2/17) of our question shows in its working out.


141. By attempting to construct a formal truth theory for a natural language, in the manner of a semantic interpretation of first-order predicate logic, one learns how a hypothetical speaker of a possible language resembling a fragment of, say, English might construct an indefinite number of new sentences from a given vocabulary and grammar: The language is only ‘possible’ because it is not quite English; only a ‘fragment’ because the truth-theory project is as yet incomplete; and the ‘construction’ of its sentences is, in certain cases, only an arbitrary choice from several competing modes of construction. Despite these limitations, the project retains an aspect of philosophical value. Looking at the formal truth theory one sees something that is not shown either by reading or listening to speech, nor even by studying a book of grammar. The ‘something’ that exists to be shown (and which cannot be said) is simply ‘what a sentence really is’: The essential elements of a sentence are exhibited, not merely as linguistic items but in their characteristic functioning (the name ‘a’ names a; the concept ‘F’ is true of x if and only if Fx, etc.). As the elements combine according to the rules of the grammar of the formal language, their functions combine also: Before one’s eye’s, sentences form, each showing the characteristic property which marks that particular sequence of words as ‘apt for expressing a thought’: for each sentence ‘P’, to be true if and only if P.


142. That analysis of the sentence is not a Socratic definition (13/120). The philosophical content of the truth theory ‘shows’ itself only to someone who knows what to look for. In that case it must be possible to teach someone to ‘see’. The essential dialectic remains to be worked through; and just because the truth theory does not say anything it cannot contribute anything to that dialectic. That is the point where one leaves formal truth theory. (If this critique is judged fair, and assuming that the ‘fragment’ is not so severely constricted as to put into question the coherence of its underlying concepts, then one must ask whether there is any point in attempting to ‘make the formal language as complete and as much like English as possible’, or to ‘resolve the claims of competing modes of construction’. The philosophical content of the truth theory is as easily ‘seen’ in the most primitive fragment of subject-predicate sentences as in the conjectured ‘complete theory’.)


143. Frege puts the question of the analysis of the sentence in the form: ‘How do the diverse parts of a proposition hold together?’ (‘On Concept and Object’ P.54-5). That suggests that there is a danger that the sentence will fall apart; that something holds sentences together. A superficial reading then confirms the suspicion that Frege is concerned merely to argue the claims of a new kind of ‘glue’: There are certain linguistic expressions which cannot subsist on their own but need certain other kinds of expression to ‘complete’ them, and thus sentences are formed; as if that explained anything! But the original suggestion is sound: In order to grasp the point of Frege’s answer we have to think ourselves into the false position which threatens to make sentences fall apart. In mathematics, when a functional expression is completed by an argument, the result stands for the value of the function for that argument. Now the beginner’s question: ‘How can putting something next to a number produce another number?’ is obviously wrong-headed (like the question: ‘What number is x?’). A functional expression is not the name of a ‘something’ but rather the sign for the ‘doing of something’, an operation: The function simply is the producing of a determinate ‘something’ for each something given. By contrast, the question: ‘How can putting meaningful words together produce a meaningful sentence?’ seems so far from being ‘wrongheaded’ as to be deeply problematic. But the diagnosis remains the same: Whereas names ‘give’ objects, concept words are only the signs for a certain kind of ‘operation’: in this case, the determining of a truth value for the ‘completion’ of the concept expression for each given object. To suppose that one can so much as refer to a concept by forming a name from the concept expression (e.g. ‘the concept horse’, ibid. P.46) is already a false philosophical. theory: The sentence falls apart into a mere collection of names. (One may illuminatingly compare the Tractatus theory of representation (2/18-26, 19/92): If the metaphysic of logical atomism is separated off, then the theories of representation and of concept and object reduce to the same insight seen from different dialectical points of view.)


144. The Fregean analogy between concept words and functional expressions appears vindicated in the construction of a formal truth theory; where, in the most primitive case, argument expressions (proper names) combine with functional expressions (concept words) to form sentences. One may now be tempted to say. What there is to be ‘seen’ in the construction of a formal truth theory is just the determinate expression of Frege’s (and Wittgenstein’s) insight into the distinction between concept and Object. However, with the rejection of the ‘false philosophical theory’ (143) there remains a metaphysical illusion to diagnose and reject: Frege’s ‘insight’ was not altogether new. The choice of the words ‘given’ and ‘operation’ should remind us of Kant’s distinction between ‘intuition’ and ‘concept’ (6/56). In asserting that distinction Kant meant to reject the false empiricist theory which attempts to reduce concepts to a ‘given’, constructed out of the materials of experience. If we wish to comprehend fully the concepts of ‘concept’ and ‘object’ we must determine how these concepts become ‘distorted’ (9/82) in a metaphysic of experience.


145. The distinction between concept and object is itself purely logical. In any given vocabulary, there remain different ways of ‘dividing up’ reality; different divisions into ‘concepts’ and ‘objects’ can nevertheless serve to convey the same information. For example, one can assert equivalently that Socrates (object) is wise (concept); or that the singleton set {Socrates} (object) is a wise-membered set (concept); or that wisdom (object) socratizes (concept); or indeed that reality itself (object) socrates-wises (concept). Then which are the real objects and the real concepts? Suppose one asks how language ‘gets to grips’ with reality: Language secures its ‘grip’ through the act of reference; naming is a precondition of being able to say anything about anything. (The full significance of this demand will emerge only towards the end of the dialectic; 32/318-22.) The problem now is that it appears we can choose which things are ‘named’, what the ‘objects’ are. But if the description can ‘grip’ reality at ‘different points’ and yet end up saying the same thing then there can be no inference from what is said to which parts of reality are ‘really gripped’: Language does not grip reality at all! Now both the metaphysic of experience and the metaphysic of action agree that asking for the ‘real’ objects and concepts in any sense other than the mapping of one language onto another is the symptom of an illusion; language does not ‘grip’ reality in that sense. The clue to their divergent interpretations of ‘concept’ and ‘object’ first appears in their differing accounts of why the question must be rejected.


146. Are the ‘real’ objects not simply given to the phenomenalist’s thinking subject through its faculty of sensibility (144)? Intuition as it is in itself does not contain any objects. Objects are ‘given’ only as the precipitate of bringing intuition under concepts; different ‘divisions’ of thought’s own system of representation into concepts will produce different objects. Only the rejected ‘private object’ of empirical idealism could claim an ‘absolute reality’, independent of concepts. Nor can I step outside the world of my possible experience and ‘compare’ my concepts of objects with objects as they are ‘in themselves’. That suggests the following interpretation of the rejection of phenomenalism by a metaphysic of action: Only for a thinking subject who is an agent can there be said to be real objects, existing independently of the subject’s choice of concepts; for example, the features of the agent’s immediate environment, the tools to be picked up and used, the obstacles to be avoided. We know that there is ‘something right’ about this assertion; only asserting it now places it at the wrong point in the dialectic. For it overlooks the fact that the interpretation of the subject’s actions necessarily takes place in a language; the description of the ‘real’ objects of the agent’s interactions with reality varies as before with the choice of a ‘division’ of reality into concepts and objects. Just as the transcendental ego cannot step outside its possible experience so we cannot step outside our language and ‘compare’ it ‘directly’ with reality.


147. In rejecting phenomenalism we do not recapture the ‘real’ objects but rather rescue the concept of an object from the distortions of the ego illusion. Indeed, the illusion concerning the nature of objects and that concerning the nature of concepts is one and the same: Consider the modal ambiguity in: ‘the necessity of knowing objects under concepts’. That can mean either: ‘The knowing of objects under concepts is necessary for knowledge’, or: ‘knowledge is necessarily the knowing of objects under concepts’. The phenomenalist intends the latter reading (9/80-1), unable to see any alternative mode of ‘knowledge’ except for the illusory empirical idealist conception of ‘direct contact between mind and its own internal objects’. But, as we have seen, the metaphysic of action provides an alternative possibility of ‘direct contact’, viz., the locus of interaction between an agent and his environment; a form of knowledge which is irreducibly practical (9/86). The phenomenalist’s falsification of the nature of ‘acquaintance with objects’ in that sense destroys the ‘reality’ of objects; and moreover entails the falsification of the concept of ‘concept’, of the role played by concepts in building up our map of reality. But we have yet to draw the consequences of the rejection of the phenomenalist’s illusion for the understanding of the relation between mind and its objects and the function of concepts (17/161-3, 18/178).


16. Sense and reference


148. The initial shock caused by the discovery that language does not ‘grip’ reality disappears when it becomes clear how things would ‘have to be’ to be otherwise: There are no ‘private objects’ and no possibility of comparing language ‘directly’ with reality ‘as it is in itself’ (15/145-6). But now the train of thought which starts from the recognition of the ‘purely logical’ construal of the distinction between ‘concept’ and ‘object’ reveals a far more disturbing consequence: In setting up the ‘ontological relativism’ of concept and object (this doctrine is part of what Quine means by ‘ontological relativity’, Ontological Relativity P.26-68; for the dialectical meaning of Quine’s ‘indeterminacy of reference’ cf. 17/162-3) we did not go so far as to put into question the prior determination of the language itself; the ‘division’ of reality by means of the determination of a lexicon of concept words. The possibilities which that raises appear devastating: If a question is raised against our own language, how shall we find the words to answer it?


149. Talk of different ‘divisions’ of reality suggests a relativism in the determination of language parallel to the relativism of ontology; that, further, the initial shock at the seemingly ‘sceptical’ challenge of this ‘linguistic’ relativism might be similarly supplanted by an ‘insight into the illusion which it rejects’. The parallel fails simply because we cannot give examples of the ‘alternative possibilities’ envisaged by linguistic relativism, We can only use the language that we understand; that suffices for the characterization of ‘alternative divisions’ into concept and object, but remains mute and unhelpful when one attempts to envisage what an ‘alternative determination of the lexicon’ might be. Consider an example of Kolakowski’s: the concept of ‘my ear and the moon’ (Marxist Humanism P.48). That employs concepts which we do understand to form a ‘conjunctive sorta1’; it is therefore only a parasitic concept and not the kind of ‘alternative possibility’ which linguistic relativism is meant to entail. If one imagines instead the ‘discovery’ of a genuine ‘alternative’, i.e., a word from another language untranslatable into our own, that seeming ‘possibility’ has not yet been distinguished in thought from a mere inability to find a suitable translation, owing to lack of sufficient data. The same applies to the particular case where speakers of the other language appear to employ a conjunctive sortal while refusing to recognize its parasitism on the conjoined concepts. We indeed seem to understand what the Kolakowski example ‘means’; but only by leaping ahead of understanding to something which the example itself can neither justify nor explain.


150. Ontological relativism’s advantage of expression over its linguistic counterpart might seem, at first, to render the latter less of a threat: One just throws the onus of proof over the linguistic relativist, who, far from being able to discharge it, cannot even say what he means! But that is a shallow response. We appear to be able to ‘comprehend the challenge’ even if we cannot express it. That means there is a dialectic to be worked through; a task which will have been completed only when the seemingly comprehended question is resolved. But if we take our task seriously then the tables are turned: The very fact that ontological relativism could be ‘expressed’ rendered it perfectly safe, without consequences for our linguistic practice. By contrast, in answering the challenge of linguistic relativism we must forego our naive, unreflective confidence in the very words we use.


151. Short of supplying an example, can one at least say what kind of thing an incomprehensible ‘alternative concept’ might be? If it is a concept then it has an extension; only this extension is determined in an incomprehensible way. Here is a clue: The notion of an undetermined extension is nonsensical; even picking objects at random ‘determines’ the extension of a concept, viz., that of being one of those objects (a ‘parasitic’ concept). Moreover, any example of ‘determination’ which we can give is necessarily comprehensible: a determination with a sense. The ‘alternative concept’ must therefore purport to have a ‘sense’; only it is one which we are incapable of comprehending. But that now raises the question: How do we know that what we find ‘incomprehensible’ is really ‘without sense’? And now the term ‘relativism’ itself proves badly chosen. For the possibility of an ‘alternative concept’, insofar as it is an alternative to a concept which we seem to understand — for only thus could our ‘understanding’ of our language be threatened — is nothing other than the possibility that both our seeming ‘understanding’ of our own concept and the seeming ‘incomprehensibility’ of its ‘alternative’ are wrong: The ‘alternative’ is no mere alternative. Linguistic relativism makes its point by dropping out of the picture.


152. That ‘seeming understanding’, by contrast with that of the ‘private linguist’, can be shared and communicated; the seemingly understood word has a use in a public language (as indeed has the word ‘private object’ among philosophers who share the illusion of a ‘private language’). But wasn’t the point of the slogan: ‘Meaning is use’ that if a word has a use then it has a meaning, a sense? (13/123-5). In that case (to continue in the spirit of the relativist’s challenge) the slogan is false. Any account of meaning which entails that ‘use cannot be wrong’ cannot be right. We may now be tempted to see a Platonic metaphysic (13/122) as the only alternative; knowing well that the price of such a move is an irresolvable scepticism about the possibility of ever knowing when a seeming ‘concept’ really corresponds to the ‘divisions of reality as it is in itself’. This dichotomy is indeed the way one first understands the ‘threat of linguistic relativism’. that to even go so far as to accept the mere possibility that ‘usage might be wrong’ entails that we do not ever really know whether we mean anything by what we say.


153. That interpretation of ‘meaning is use’ overlooks the following possibility: It was no part of the intention of the slogan to say what meaning is ‘in itself’, in the sense of giving something (‘linguistic usage’) which we may now talk about instead of talking about ‘meanings’; as if the ‘problematic’ concept of meaning required an ‘unproblematic’ proxy. Identifying meanings is the only way to talk intelligibly about ‘use’, unless one descends to the level of seeing language as a meaningless ‘barrage of noise’ (10/96). The target of the slogan is the illusions of Platonism and phenomenalism; and in rejecting those illusions the slogan redirects our attention to where meaning is to be found. One may remain within the spirit of that thought and still undertake the task of showing how we can know whether a word which we use ‘really has a meaning’.


154. The fact that there has existed agreement to date over the use of a word is a phenomenon which demands empirical explanation. There is some ability which we share which makes the agreement possible. The ability, once brought to light, might turn out to have no interest or value. Still, one could conceive of how it would form part of the meaning of a possible concept, even if the seeming ‘concept’ which now embodies it must be rejected. Now if the fact of agreement itself is not the kind of thing to be either justified or criticized, the only other possibility is that the use of a word, seen in the context which makes that use possible, must divide up into a plurality of abilities, into different ‘aspects’; and although each one of these aspects is necessarily immune to criticism, the aspects may nevertheless fail to cohere together when taken as a whole: The only room for the criticism of concepts can be in the upshot of combining different abilities together which, taken individually, are simply brute facts. But all this is highly abstract: There is yet no indication of what these ‘aspects’ or the ‘upshot’ of their combination might be.


155. Usage proves the existence of some kind of ‘sense’; of something half-understood, half-seemingly-understood. Let us say that the question whether a word is to be criticized for expressing an incoherent concept concerns the ‘objectivity’ of its seeming ‘sense’, of that purported ‘concept’. using a non-objective concept is comparable to using a name without a reference; something is presupposed which is false. Since there is no way one could represent the seeming ‘mental act’ of a person attempting to ‘use’ that concept except by employing that very concept and thus accepting its presupposition, failure of objectivity means a failure in the mental act itself; there is no thought to be ‘represented’. Being ‘objective’ is thus the concept of the being of a concept which corresponds to the concept of ‘being an object’. But whereas the existence of the named object is sufficient for the name’s being a genuine name, the existence of an extension, a shared ‘ability’ to pick out the objects to which the concept ‘applies’, does not guarantee the objectivity of the concept which determines it. Usage may ‘agree’ on the extension but still there remains something which it risks getting wrong. The manner of the ‘determination of an extension’, in the context within which the ‘determination’ takes place, reaches beyond that extension.


156. Frege introduced the distinction between the ‘sense’ and ‘reference’ of a sign ostensibly as the means of accounting for the potential informative value of identity statements (‘On sense and Reference’ P.56ff.). The sense of a sign is the ‘mode of presentation’ of its reference. Difference in the mode of presentation of a given reference will make a difference to the informative value of any proposition concerning that reference. This stands out in the case of identity statements, where there is clearly no other way in which the statement could be ‘informative’. We should not take Frege to be arguing for a distinction between sense and reference; as if, in the case of proper names, there is room for ‘dispute’ over the possibility of attributing a sense as well as a reference. Whatever the use of a name consists in determines its reference; the notion of an ‘undetermined’ reference is as nonsensical as that of an ‘undetermined’ extension of a concept (151). There can be disagreement only over the kind of sense that a proper name has. The context of the introduction of the sense-reference distinction rather determines the constraints upon an acceptable account of sense. In that case, whatever is shown by the example of identity between objects is shown equally in the case of statements of equivalence between concepts; even though there was never any doubt concerning whether concept words ‘possess a sense’. (Whether concepts may be said to be ‘in themselves’ the ‘incomplete entities’ referred to by the ‘incomplete’ expressions for concepts (15/143), distinguished from the extensions which concepts determine, need not be pursued here.)


157. It might seem only a ‘contingent’ feature of the manner in which human beings acquire knowledge, for example, the fact that we have to ‘acquire’ it, that words possess senses, rather than ‘transparently indicating’ their reference or extension in a way which would make it impossible to be ignorant about the truth of any statement of identity or equivalence. In the knowledge of an omniscient being statements of identity or equivalence would seem to contribute nothing. That thought is indeed compatible with the notion that sense does not ‘reach beyond’ the determination of an extension (155). But if the fact that different modes of presentation determine the same extension or identify the same object ‘contributes something to knowledge’ then cancelling out ‘opaque’ names and concept words in favour of a hypothetical ‘transparent’ language cancels out something which exists to be known, contrary to the hypothesis of omniscience. A being who ‘knew’ all of reality ‘transparently’ would know less than one who knew it in a form which indicated the systematic dependence of his knowledge on identities and equivalences. The whole point of a sense is to determine reference or extension; and yet, the very fact that different senses can determine the same reference or extension shows that a word’s having the sense that it has must contribute something more to knowledge than merely ‘determining an extension’. In that ‘something more’ lies the possibility — so far only a possibility — that a sense can determine an extension and still be wrong, still fail to be objective.


17. Name and object


158. Sense determines reference (16/151, 6). Sense is found in linguistic practice (16/153). Sense risks criticism, while the ‘agreement’ and ‘regularity’ manifested in the use of a word remain brute facts (16/154). The possession of a sense explains the contribution of a name or concept word to the informational content of a statement of identity or equivalence (16/156-7). One may be tempted to object that these assertions tell us only what sense does and not ‘what sense really is’. However, putting the question of the meaning of ‘sense’ in that socratic form threatens to commit the dialectical fallacy of ‘identifying a function with an object’ (generalizing from Frege’s argument; 15/143): Behind a ‘function’ in the widest sense, the potential to ‘do’ or ‘produce’ something determinate according to a given ‘input’, there mayor may not lie an identifiable ‘agent’, a subject with properties of its own to which the ‘performance’ of the function may be attributed. Admittedly, we lack an understanding of ‘sense’ for as long as all that sense does has not been gathered together in an intelligible way. But that requirement remains purely formal; we do not know what it is that we do not understand, and cannot know until we have found what we are seeking; so any attempt at a further, material specification of the requirement is likewise a ‘fallacy’.


159. The sense of a proper name has no other function than to determine its reference to an object. For that reason, failure of reference provides the only ground for criticism of the sense of a name (16/155), however much one might wish to object to the choice of name on aesthetic grounds, or for calling up unjustified associations. Whenever a name appears open to criticism of a more full-blooded kind, it is because the sortal concept with which the name is associated, the concept of the kind of object which the name refers to, itself fails to be objective; no such ‘objects’ can be identified. In that case, a fortiori, the name lacks a reference. The question of objectivity thus arises in an interesting way only for concepts (we shall return to that question in the next chapter): By contrast with the criticism and revision of concepts, it is no discovery that names can fail to be ‘objective’. Since the consequences of following up the ‘challenge’ of linguistic relativism (16/151) do not affect the use of proper names, that use has not yet been made to appear opaque to us. There must be other grounds to motivate an inquiry into the sense of proper names.


160. The problems of metaphysics concern the relation between mind and reality. And language serves to represent our thoughts about reality. Combining the two thoughts does not entail that metaphysics should be interested in every function of language, nor even every ‘representational’ function. Only the construal of the question of mind’s relation to reality as the demand for a globally adequate ‘theory of meaning’ (13/126-7; that ‘demand’ has yet to be investigated) would automatically guarantee that metaphysics will have ‘something to say’ about proper names. By contrast, in a ‘dialectic of language’ one can only wait and see how the metaphysical illusions ‘manifest’ themselves (13/119). We may allow that knowing the relation between a name and the object which it names in some sense entails ‘knowing more’ about the relation between mind and reality. But that sense is only interesting if, in attempting to describe the relation between a name and its object, one risks falling victim to a metaphysical illusion. In that case, fully ‘grasping’ that relation would consist in the ability to recognize and resist that particular illusion.


161. What does it take to make an object the object of my thought? If there was nothing in reality to distinguish my act of thinking about one object from thinking about some other object, that would be tantamount to saying that my thought was not ‘about’ any particular object. Then what does it take to make that distinction? It would suffice that I in some sense ‘knew’ which object I ‘meant’; let us for the moment consider only this case. In a metaphysic of experience, all knowledge reduces to the ‘anticipation of possible perception’ (9/83). Knowing an object entails being able to place it in one’s representation of the world of one’s possible experience. The transcendental ego must be able to represent to itself that which ‘distinguishes the act of thinking about one object from thinking about some other object’, for every question concerning the interpretation of its thought is a question which the ego can and must answer for itself (9/80-1). But now — the argument has been well rehearsed (5/47-8, 10/92) — if its seeming to the thinking subject that it has determined a particular object to be the object of its thought is that which ‘in reality’ determines it, then there are no objects of thought, no such thing as thinking ‘about’ anything, no difference between thinking and dreaming: The ‘reality’ which ‘makes’ an object the object of my thought cannot be ‘in me’, in the phenomenalist’s vicious sense.


162. That formulation of the ‘rejection of phenomenalism’ prompts an illuminating misunderstanding: ‘you have not shown that what makes an object the object of one’s thought cannot simply be one’s own identifying representation of that object, but only that there must exist the potential for public agreement over the relevant knowledge. The ability to communicate one’s thoughts to others shows that one knows which object one is thinking about.’ — The reaction expresses Frege’s demand for the ‘publicity’ of a name’s sense (‘On sense and Reference’ P.59-60). But even here the ego illusion threatens: Let us suppose that a ‘publicly understood’ name acquires its communicative import simply from a ‘harmony’ between ‘identifying representations’, manifested in ‘agreement’ over how to determine which object the name names. Now it sometimes happens that one comes to doubt whether one is using a name with the ‘same sense’ as someone else. When such a doubt arises one can request an identifying description of the named Object. But what if I find myself unable to recognize the Object I mean in the various descriptions which are presented to me? There remains demonstrative reference: My interlocutor conducts me to an object and I say. ‘yes, that’s the one I mean’. But now it might occur to me that the act of demonstration itself ‘may still have been misunderstood’. ‘understanding’ must not of course reduce to ‘having the same representation in our minds’, for the publicity of sense demands that ‘having the same representation’ be capable of being manifested in practice, on pain of never knowing whether or not demonstrative reference has ‘succeeded’. For that very reason, the present account forbids me to consider the possibility that demonstration and description might not jointly entail a unique reference. Nevertheless, it appears to me that I can still conceive of the ‘logical possibility’ of ‘understanding all acts of demonstration differently’. This is not the illusory ‘possibility’ of different ‘private representations’. Nor is it either an ‘ontological’ (15/145-6) or a ‘linguistic’ (16/148-51) relativism. It is rather a possibility which would be represented by a mapping of the lexicon of a given language onto a substantial permutation of that same lexicon, including the apparatus for identification and individuation, in such a manner as to frustrate all attempts to resolve the ‘difference in referential intentions’ by means of language or otherwise.


163. One is tempted to protest that the mind will only stretch so far, that some ‘possibilities’ are not really possible, are too ‘far fetched’ to worry about for practical purposes. However, the argument is not intended as a sceptical paradox, but aims to identify a fatal flaw in the ‘identifying representation’ model of reference. That model fails to incorporate an ‘essential ingredient’. Because of that omission it appears, first, that reference must be ‘logically guaranteed’; while, second, that language could be ‘fixed up’ in such a way that it would be logically impossible to determine whether or not referential intentions are in ‘agreement’: Taking Frege’s demand for the ‘publicity’ of sense too literally merely transforms the ‘passive observer’ of phenomenalism into a passive communicator. The whole point of communication becomes ‘communication’; either communication contains within itself the means to rule out the possibility of a failure in communication or no communication takes place. But we are agents and not passive communicators. We should not, however, expect the introduction of action to provide a new ‘criterion’ to ‘logically guarantee’ the success of communication where the act of communication itself fails. Rather, because false belief necessarily has potential consequences for action (6/58), a fully systematic ‘permutation’ of the lexicon of a language, such that no false belief could ever ‘turn out true’ or a true belief ‘turn out false’, cannot be ‘substantial’, must reduce to a notational variant. Communication nevertheless remains corrigible; one can conceive of how any given act of communication, or any finite series of such acts, might be ‘misinterpreted’ by means of an as yet undiscovered ‘substantial permutation’ which is not ‘fully systematic’. However, that does not rule out the possibility of knowing that communication has succeeded; the metaphysic of action has already refuted that form of sceptical argument (9/86-9). Communication is possible because its point, as with the concept of knowledge itself, lies in its consequences for action. But the behavioural facts provide no ‘logical guarantee’ against failure.


164. Now in what sense, if any, does reference to an object by means of a proper name entail possession of ‘identifying knowledge’? (161). It is first necessary to determine the point of proper names. Here is a familiar account: ‘The introduction of proper names into a language permits an individual to defer, in making reference to an object, to the stock of identifying knowledge possessed by the community. Speakers need not first agree upon an identifying description, provided that they possess a means of checking whether the identifying knowledge which either associates with a proper name or to which either defers in his own use of that name serves to identify the same object.’ --- Just because much of this picture appears a summary of the ‘facts’ of linguistic practice, it becomes hard to extract the ‘theory’. One should remember that the meaning of any such contribution to the dialectic may reside not in its face value but in the motivation for saying it at just that point in the argument. That motivation first appears when one considers the problem of empty reference.


165. If the object of a purported demonstrative reference does not exist then no thoughts are expressed concerning that reference. The phenomenalist applies this principle to his own case; the ‘objects’ in dreams and hallucinations are only seemingly ‘thought of’, to be eliminated as ‘don’t cares’ in the process of discerning ‘objective’ experience. For the existence of a thought concerning a seemingly ‘demonstrated’ object could be indifferent to the objective existence of the object only if the real ‘object of thought’ was a subjective, ‘private object’ (4/40, 5/46). The same principle applies to a name given to the object of a perceptual encounter. Names associated with only seeming ‘experience of an object’ have no meaning; the thinking subject cannot represent the thought of an object to himself by means of the use of a name if the presupposition of that name’s having a meaning fails. But a name may also be introduced as ‘shorthand’ for an identifying description; in that case the name remains available for the expression of thoughts whether it has a reference or not, for the words used in the description retain a meaning irrespective of its being uniquely satisfied.


166. How does the rejection of phenomenalism alter the picture? The crucial case is this. When I take up another person’s use of a name, does that involve my fixing its meaning as ‘equivalent to a description’, or is it to be treated in the same way as the names for objects with which I take myself to be acquainted? (‘no object, no thought’). The ‘community model’ of proper names (164) can be read in either way; but clearly it is intended to disarm the obvious objection to the first alternative, allowing that one can ‘use’ a name while ‘deferring’ to the community’s ‘knowledge’ of its equivalent, descriptive meaning. For the phenomenalist, that first alternative is indeed the only coherent possibility (understanding the ‘community model’ in the manner of a phenomenology of reference; 8/71-2). Another person’s perceptual encounters are nothing to me except as data of testimony from which I infer how things are in the world of objects of my possible experience; for whether the seeming ‘expression of a thought’ is genuinely so cannot depend upon a ‘question’ whose ‘answer’ transcends my possible experience (161). But we are not constrained by such a scruple; in the metaphysic of action, the interpretation of my referential intentions, and consequently the evaluation of their success or failure, is a task for a suitably informed interpreter of my behaviour. So what is the point of proper names?


167. We must first return to the question of what, in a metaphysic of action, ‘makes an object the object of one’s thought’. Knowing which Object I ‘mean’ is indeed sufficient; but this knowing now becomes a form of knowledge which I need only show in my actions, by my ability to find the object. But is identifying knowledge necessary? If the whole point of explanations of action lay in their use to predict an agent’s future behaviour (14/131), and if the attribution of thoughts remains fixed by its role in explaining action (10/94), then the following possibility could not be coherently described: that a subject’s thoughts should be ‘about’ one rather than the other of, say, a pair of identical twins whom the subject would be unable to distinguish between as the ‘intended’ and ‘unintended’ reference of his thoughts. But we do not merely ‘predict’ behaviour; we evaluate it, and one of the consequences of evaluation is our distinguishing those true beliefs which succeed in being knowledge: Suppose now that our subject possesses true beliefs whose source is the subject’s past encounter with one of the identical twins. The very same beliefs may have arisen through a ‘reliable causal route’ in the one case and an ‘unreliable’ route in the other (9/88). Even though the subject does not in any sense ‘know’ whom he ‘means’, we know; the twin meant is the twin whom the subject met, for which one he met matters in the evaluation of his actions. Now that does not mean that being the ‘causal source of one’s beliefs’ is sufficient to determine an object as the object of one’s thoughts. It can be; but in the absence of the sharp criterion provided by ‘identifying knowledge’, necessary and sufficient conditions cannot be given. Each case must be examined on its merits, bearing in mind the point of attributing thoughts about a particular object.


168. With the introduction of proper names, the above account may be extended one stage further: It is not necessary to ever have encountered the object of one’s thoughts. It suffices that one has taken up the use of a proper name for that object. The point of identifying a source of the subject’s beliefs as an object of those beliefs of which the subject has no identifying knowledge depends, in this case, upon the possibility that the use of a given name can serve to transmit either knowledge 2E true beliefs which fail to be knowledge. If possession of identifying knowledge is thus not necessary for competence in the use of a proper name, then there can be no philosophical grounds for resistance to the idea that whether a name has a sense, and consequently can be used in the expression of thoughts, may depend upon the existence of a purportedly named ‘object’ which the subject need never have encountered.


169. Here is a game I might play: to write down every proper name from my personal vocabulary with whose bearer I am not acquainted, and decide upon an ‘identifying description’ for which the name is now to be treated as shorthand. If an interpreter of my behaviour were to find cause to take this ‘intention’ seriously then utterances of mine containing bearerless names from the list could now be regarded as genuine expressions of thought (165) rather than only ‘expressions’ of the fact that I believed that the utterances were apt for ‘expressing thoughts’. One might describe such behaviour as ‘opting out’ of the use of our common language in favour of a specially constructed ‘private idiolect’. Now to ‘opt out’ would mean retreating to a position of ‘methodological’ solipsism analogous to the solipsism of the passive observer, but a solipsism shown through one’s actions rather than privately ‘meant’ (11/106; contrast the ‘practical solipsism’ of the ‘mere agent, 11/108); accepting the phenomenalist’s ‘scruple’ (166) without the phenomenalism. The only conceivable motive for such an action — short of a radical breakdown in the linguistic community — is the determination to make one’s ‘theory of proper names’ true at all costs.


170. Herein lies a complete theory of proper names: We encounter objects (161, 167) and give them names. The use of these names is taken up by other members of the linguistic community, through the process of receiving information about those objects (168). To fully understand these assertions entails the ability to ‘recognize and resist’ the ego illusion (160). That proper names do not function as ‘equivalent to descriptions’ (169) is a matter of our intention; or, rather, the absence of an intention. Speaking a ‘common language’ has great utility. But there are limits. In a less than perfect world one cannot always trust what purports to be ‘information concerning an object’; names are potentially ambiguous, leading to a confusion of sources; the source may become temporarily lost, or else cease to be a genuine ‘source’ of information, reducing to the nominal subject of a fixed set of beliefs. One cannot be certain, therefore, when one unreflectively ‘takes up’ the use of a proper name, that such an action will have its intended significance.


18. What are concepts?


171. Concepts are nothing ‘in themselves’ (15/143). They are operations of thinking and speaking associated with the use of concept words (16/153); the producing of a truth value according to a given input but not the ‘agents’ which ‘do’ the producing (17/158); the determination of an extension which nevertheless in some manner ‘reaches beyond’ that extension (16/155). To this we may now add: Just because the sense of concepts is problematic in a manner in which the sense of names is not (17/159), understanding thought and language as they are ‘in themselves’ revolves around an understanding of concepts. If philosophy has anything to do with language, then it has at least to do with comprehending the concept of ‘concept’.


172. How does language work? To the ‘theorist of meaning’ (13/126-7), talk of the ‘workings’ of language suggests the thought of a complex mechanism; an instrument which I use in order to communicate without knowing its inner construction. A second thought is this: My ignorance is surely not the ignorance of someone who, say, uses a calculating machine without knowing ‘how it works’. Language is not a thing; it is a mode of human activity. Knowledge of the workings of language is self-knowledge. So one attempts to combine the two thoughts: It is as if my unreflective linguistic competence results from the covert operation of a mechanism inside my own mind. But how does this mechanism itself work? Language is not the phenomenalist’s ineffable ‘code’, an ‘association’ between public words and ‘private’ representations (13/126, 17/162) but only a game played according to rules. The speaker is not normally aware of these rules; but they are nevertheless ‘in’ the speaker’s mind, guiding his unreflective linguistic practice. The mechanism is thus constituted by implicit knowledge of rules. — The picture provides a powerful impetus to philosophical theorizing about language; but it is fundamentally mistaken.


173. According to the ‘mechanical’ picture of language, concepts would after all acquire an ‘agent’ to which their operation in thought and speech could be attributed (17/158): not the agent who uses concepts, the subject who thinks and speaks — for it is not up to the language user to determine how concepts will ‘operate’ — but rather the rules of the language game, internalized in each speaker’s mind. Despite one’s recognition of the ‘fundamental difference between concepts and objects’, one may nevertheless be tempted to fallaciously identify the functions which concepts embody with hypothetical ‘objects’ which ‘perform’ those functions. In that way, false ‘theorizing’ about language acquires a further impetus.


174. The reader should note that the foregoing is offered, not as a ‘diagnosis of metaphysical illusion’ (although the illusion may indeed contribute; 17/162-3) but only the explanation of an error. The purpose of a dialectic of illusion is to remind the reader of his own metaphysical illusions, to bring the illusions into consciousness (1/6, 3/27) in order to exhibit their illusory nature. Even after the illusions have been ‘rejected’, they remain in view, a continuing source of ‘temptation’ (5/54, 6/56). By contrast, the explanation of philosophical error remains purely hypothetical until directed at a particular individual, or group of individuals; and then stands to be refuted by a plain denial. Appeal to hypothesis is psychology, not metaphysics; such a departure from the dialectic can be justified only as a last resort, faced with having to deal with the undialectical reality of some particular opponent’s position: in the present instance, Michael Dummett’s philosophy of language (1/13).


175. A ‘theory of meaning’ would make explicit the speaker’s ‘implicit’ grasp of the sense of the concept words in his linguistic repertoire. That means stating what this knowledge ‘amounts to’ in terms which do not employ the concept whose sense is being explained. Otherwise, one could give the following vacuous ‘explanation’ of the ‘rule’ for employing the concept ‘F...‘: The concept ‘F...‘ may be applied to all and only those objects which are F (cf. Dummett ‘What is a Theory of Meaning?’ P.108-9). The ‘theory’ has to explain what all the ‘rules’ really are. By the same token, our present inability even to describe the correct form of an ‘explanation of the rules’ shows that we do not as yet comprehend the essential nature of concepts. But that presents a deep mystery: What stands in the way of ‘comprehension’? We are no longer shackled by the illusion of a ‘private language’ (13/126). And it is not simply that the rules are ‘very complicated’; for our task is not to actually ‘explain’ all the rules but only to describe the correct ‘form’ of an explanation. So the deficiency in our knowledge is not one which will be put right just by hard work; if we only knew how to carry out the task, the work would be unnecessary; but since we don’t know even where to start, no amount of ‘work’ will get the project underway. — That is no argument against the coherence of the project. However, the idea that, despite our complete ignorance, we know enough to be able to assert that the project ‘must be feasible’, ought to raise one’s suspicions. And it encourages one to search hard for an overlooked possibility.


176. What is the point of concepts? The question seems absurd. For we surely do not envisage, as in the case of proper names (17/164-8), a ‘point of view’ from which one might contemplate the ‘introduction’ of concept words into a language! But consider this more pertinent formulation: We have identified our object (s) and now we ‘apply a concept’; and the result of that action is either ‘correct’ or ‘incorrect’, ‘true’ or ‘false’. But what follows from that? So what? — The question seems impossibly general I When the description or explanation of something is incomplete, then ‘so what?’ can serve to prompt further explanation. But after a certain point, the question is just perverse. Still, metaphysics is ‘perverse’; and intuition tells us that, in the present context, the question is getting at something. Now, if a question seems ‘impossibly general’ it can sometimes mean that a reformulation of the question will provide its own answer. Call any reason which justifies applying a concept a ‘ground’ (in the case of perceptual concepts, the ground for applying ‘F...’ can simply be its seeming that an object is F); call any further judgment or action which the concept’s truly applying can serve to justify a ‘consequence’. The point of concepts is simply and exclusively this, that applying a concept necessarily involves a potential transition from grounds to consequences. The ‘so what?’ question always applies (only when we run out of things to say there nevertheless remain things to do). This does not yet tell us ‘what concepts are’. But we have yet to determine what would suffice to answer that question. Asking for the ‘point’ of concepts only provides the angle. (I do not distinguish here between ‘conventional’ and ‘non-conventional’ grounds and consequences. Moreover, the ‘reasons’ consequences embody exclude all that does not serve to justify a belief or an action, and consequently contribute towards the evaluation of an agent’s behaviour (6/58). Contrast Dummett Frege Philosophy of Language P.362, 396-7, 454-5.)


177. The only concepts for which there was no potential for inference from grounds to consequences would be those of a private language (6/58). Consider the concept ‘red’. That an object is red can be a reason for believing it to have properties which we have discovered to be associated with redness in objects of that kind; to have a surface whose properties explain why it appears red; to be that object (the one I saw earlier) rather than some other. If it were only a contingent fact that an object’s being called ‘red’ had ‘consequences’ then possession of the concept ‘red’ could be exercised even in a world where ‘colours’ were not correlated with any physical properties or events, and moreover changed frequently and unpredictably, thus making identification by colour impossible. In such a world any seeming ‘agreement’ over the ‘use’ of the word ‘red’ would be an inexplicable phenomenon, for otherwise whatever explained it (radio transmissions from Mars, say) could be taken as that of which ‘red’ was the phenomenal manifestation. To suppose, on the contrary, that this ‘agreement’ would still be an agreement in judgements is just to suppose the existence of a ‘private colour language’ whose use ‘underlies’ the public ‘manifestations’ of ‘agreement’ in our judgements about colour.


178. The phenomena1ist, in rejecting the idealist’s ‘private object’ in favour of a ‘private content’ (10/91), merely rejects the idealist’s ‘bare recognition’ of secondary qualities while retaining an equally incoherent ‘bare recognition’ of primary qualities: The application of concepts of secondary qualities now has ‘consequences’ for the determination of how things stand in a ‘spatial reality’ which transcends the mind’s own contents. But the transition from grounds to consequences terminates with the construction of a ‘representation’ of this spatial reality. The ‘correctness’ or ‘incorrectness’ of a representation answers to nothing except other, similar ‘judgements’ that reality has been ‘correctly’ or ‘incorrectly’ represented. suppose then that instead of erratic ‘manifestations’ of seeming ‘colours’, there appear in our hypothetical world spatially ‘located’, enduring phenomenal objects, unconnected with any physical processes. Once again, the ‘agreement’ over the use of concepts of ‘primary’ qualities of ‘phenomenal shape’, ‘phenomenal location’ (to be distinguished from the genuinely ‘primary’ qualities of physical objects; 8/76) reduces to an ‘unexplained phenomenon’. To suppose that this would remain an ‘agreement in judgements’ is to embrace the phenomenalist’s illusion that the subject who ‘represents’ reality is essentially a disembodied observer; that the ‘representation’ of reality terminates in thin air. But what then is a representation? There is no question of giving here a pragmatic definition of ‘representation’ (a ‘pragmatic theory of truth’). Having diagnosed the phenomenalist’s illusion, we have said all that is necessary, or indeed possible to say concerning the ‘practical import’ of ‘constructing a representation’.


179. We have seen how the possibility of the criticism of concepts entails that the use of concept words ‘divides up’ into different ‘aspects’, themselves ‘brute facts’ but the ‘upshot’ of whose combination may indeed be criticised (16/154); and that the ‘objectivity’ of concepts is a ‘presupposition’ of their use (16/155). We have seen how the very fact that statements of identity or equivalence can be informative shows that there is something more to ‘sense’ than the brute fact that, for each name or concept word, a reference or an extension is ‘determined’ (16/156-7). In the case of proper names (17/168-70), the mode of determination reduces simply to the manner in which information is obtained and transmitted from a given ‘source’ by means of a proper name. The only thing we have to say about this ‘manner’ is to recognize that the attribution to a subject of thoughts concerning a given object may be motivated in part by the way that object serves as a source of knowledge for our subject; and therefore that the ‘line of transmission’ of information involving a particular name of the object, rather than another name of that same object, may make a difference to the evaluation of true beliefs as knowledge, just as the fact that it is indeed that object, rather than a different but qualitatively identical one (17/167-8). With concepts, by contrast, we know that there is something more to ‘sense’ than simply the ‘source’ of information, and the ‘manner’ of our approach to that source; for the manner of the ‘determination of an extension’ reaches beyond that extension (16/155, 17/159). Only now, with the assembly of all the necessary materials, are we in a position to say what that ‘something more’ is.


180. Every concept has its own point. It is not an ‘aspect’ of use but rather the point of the combination of aspects I the combination of ‘grounds’ and ‘consequences’ (176). Only the point can be criticized; but only the use can be changed. Before a concept word is coined, its place in the language already exists; the point of a concept predates the use of that concept, in the point of the activities surrounding it. When a concept word is rejected, it is in recognition of the fact that its place in the language never existed; the concept never had a point, although it seemed to. The point is the rationale for associating grounds and consequences with a particular concept word. In some cases, the point is just obvious (e.g. the concept of a chair); in others, its discovery can constitute a philosophic advance. A question concerning the point of a concept can be as sharp, or as unclear, as any question concerning the purpose of a human activity; asking for the point does not always have a point. And just as with different activities, there are limits to what can be said in order to explain the point; when explanation has come to an end, either you see the point or you don’t. — Now, if a philosopher claims to have discerned a third feature in between the manifest usage and the point of a concept, the ‘rules’ which ‘govern’ the use of the concept word, we may justifiably demand the point of that notion. It turns out that positing ‘rules’ has no work left to do, contributes nothing to our understanding of ‘how language works’.


181. Because this reason for applying a concept, that consequence which applying the concept is a reason for are not regarded as ‘requiring further justification’, we save ourselves the trouble of having to justify the inference which they embody simply by our use of that’ concept word. Criticizing the point of a concept is criticizing the validity of potential inferences: Now this account is highly abstract; the logical schema has to fit an enormous variety in the kind of point that concepts can have. (In many cases, no particular inference is in question; it is as if reality ‘provides’ the concept and we only ‘discover’ its potential for inference. But that reality ‘provides’ it, is corrigible by the discovery that no ‘inferences’ are to be found; cf. 177, 8.) However, the schema serves as a means for translating talk of ‘rules’ for a concept’s use, ‘criteria’ which constitute ‘part of the meaning’ into ‘regularities of usage’ and ‘rationale’. For example, our use of a term for a certain ‘character trait’ may embody an unacceptable conception of morality, the use of a seeming ‘natural kind’ term a false scientific theory. It may be possible to salvage the point of the concept word by merely rejecting certain grounds or consequences, and thus preserving the validity of the potential inferences which it embodies. Alternatively, the concept is rejected altogether. Of course, there is no sharp cut-off point between the two cases; and the currency of a particular word may survive the ‘rejection of the concept’. Moreover, the fact of ‘linguistic change’ is not necessarily evidence of such ratiocination; concepts may also ‘evolve’ through a process of ‘natural selection’. ‘Rules’ come into existence — we construct or alter definitions, distinguish and combine meanings — only when reflection upon the point of our concepts obliges us to invent them.


182. The speaker competent in the use of a concept may not possess the ability or the knowledge to explain its point. This can seem like a paradox, If the speaker does not know what the point is then what has point got to do with usage? That the usage turns out to ‘fit’ the point would seem to be a mere accident if usage were not guided by knowledge of the point. If, on the other hand, usage is only like the technique, say, of an athlete, which adapts to the laws of physiology and physics, but is incapable of being fully guided by knowledge of those laws, then the ‘point’ of a concept cannot perform its intended role in the criticism and revision of linguistic practice. But that overlooks the following possibility, Our linguistic actions display a rationale which is accessible to the reflective observer of our behaviour. The observer sees the point of our proceeding in that way rather than in this way; while we proceed in that way because we see the point which proceeding in that way has, even if we cannot say what the point is. Talk of an inexpressible ‘seeing’ simply means that the observer knows something about our practice which we do not yet know; while this knowledge remains knowledge of the point of that very practice, rather than merely some practice which resembles it in certain respects. But we are capable of becoming observers and critics of our own practice. When that happens, we gain a knowledge of the point to which we may appeal in determining whether a particular feature of linguistic practice is to be retained or rejected.


183. The sense of a concept word is its point. To put into question the objectivity of a concept is to ask whether it has a point, or only the ‘semblance of a ‘point’. Knowing the sense of a concept word is competence in its use informed by a perception of its point; but does not entail the ability to explain the point: These assertions, seen in their dialectical context, may be regarded as contributions to a ‘philosophy of language’; but in no sense a ‘theory of meaning’. They constitute the ‘possibility’ which the meaning theorist has ‘overlooked’ (175); their claim to validity consists in the challenge to the theorist to find something more to say. But a theory of meaning has not been shown to be impossible; and it is difficult to see how that could be shown: The rejected ‘possibility’ which one does not comprehend becomes an indefinitely extended field of half-perceived ‘possibilities’; one can only work through the field, refuting each new ‘possibility’ as it turns up, never certain that all have been exhausted. Still, there remains an untried tactic. The idea of a theory of concepts as a systematic theory of rules for use begins with the rejection of ‘any theory which entails that usage cannot be wrong’ (16/152). We have provided our ‘alternative’ account of how the criticism and revision of linguistic practice is possible (179-81). We should now ask: Does the ‘theory of rules’ satisfy its own basic requirement?


184. If an ability to see the point of a concept (179) is not to be presupposed in the meaning-theoretical ‘account’ of its sense — that means. if the ‘statement of what knowledge of its sense amounts to’ (175) is debarred from containing an expression for that very concept — then its sense must be captured by rules whose understanding and communication does not presuppose a prior grasp of the concept. But now suppose that the concept is wrong and needs to be changed. We are not allowed to appeal to an ability to ‘see the point’ in order to justify the ‘new’ concept which replaces it. However, the failure of ‘coherence’ between ‘criteria’ and ‘consequences’ does not by itself indicate anything except the fact that some inference or inferences have turned out invalid; it does not say which aspect of the use of the concept is to be changed. If our decision on how to proceed is not a matter of pure indifference that can only be because we possess a second-order ‘rule’ which determines how coherence is to be restored. This second-order rule derives from a second-order concept of criticism of concepts of the type under examination. But suppose that this second-order critical concept is ‘wrong and needs to be changed’. We must posit a third-order ‘rule’; and so on. One could go on for ever ‘discovering’ (inventing) new ‘rules’ and never get anywhere. For if there is so much as a ‘logical’ possibility of a question arising which needs a ‘rule’ to answer it then we must be able to give the rule. But if ‘giving a rule’ always introduces the possibility of another question then a decisive answer to the first question can never be given. Instead, we have merely a potentially infinite series of ‘hypothetical’ answers; a logical absurdity.


19. What is truth?


185. Reality cannot be reduced to appearance: that assertion encapsulates the entire dialectic of idealism. The ‘external reality’ which the transcendental ego discerns within the ‘this...’ (5/46) is appearance, not reality; a ‘dream world’ (5/47-8). Every attempt to express the idealist’s seeming ‘insight’ in the form of a philosophical project is just the vain substitution of a construct of appearances for a reality which has been put forever out of reach by the idealist’s ego illusion. Of course, if there were no reality then there would be no appearance: If false belief is impossible, that does not mean that beliefs cannot but be true but rather that there is no thought of any kind. The ‘dream world’ is not even a dream; for dreams are something which occur in the real world. We have seen how the phenomenological introduction of the experience of ‘publicity’ (11/101-3, 14/128, 17/162-3, 4, 6, 18/178) makes no difference to the reduction of reality to appearance. For example, the ‘seeming discovery’ of a species of ‘phenomenal objects’ can and must be taken ‘at face value’ (cf. 8/78, 11/107) by the phenomenalist; for all the objects of the transcendental ego’s experience are already phenomenal objects. Consider then a phenomenological enrichment’ of the ego’s experience; the ego now experiences itself as a person among other persons, in a world of objects concerning which there is intersubjective agreement. All this seeming agreement’ amounts to is what ‘agreement in judgements’ would mean if phenomenal’ objects appeared to appear’ in our own world, The ‘correctness’ or incorrectness’ of such judgements would be constituted solely by coherence with other judgements about these seeming objects. It would first seem to us that we were ‘aiming our thoughts at reality’; but when the true nature of this ‘reality’ became known it could only be regarded as a collective hallucination, In reality, our seeming ‘thoughts’ were ‘aimed’ only at coherence with other, similar, seeming ‘thoughts’. Now this furnishes a striking model for a certain theory of truth: If, per impossibile, reality could be reduced to appearance and still be ‘reality’, then judgements could aim at the appearance of a ‘reality’ which was in reality only their coherence with other judgements about that ‘appearance’ and still be ‘aiming at reality’ In a reality constituted by its own appearance, coherence itself becomes the ‘reality’ at which judgements aim.


186. Judgements may be said to aim at reality only if false judgement ‘has practical consequences’. But merely saying that leaves phenomenalism unscathed; for in the absence of a refutation of the illusion that I possess ‘something’ whose existence sets me the task of determining the objectivity of my own experience, a ‘this…‘ whose ‘structure’ must indeed suffice to determine objectivity (6/56-7), the assertion reduces to a phenomenological critique of the coherence theory of truth, replacing it by a ‘pragmatic’ theory which ‘incorporates the recognition that judgement has practical consequences’ (a second ‘pragmatic theory’; 18/178). Indeed, so long as ‘practical consequences’ is understood phenomenologically, we have merely introduced more variety in the judgements with which a given judgement aims to ‘cohere’ (9/81-2). Of course, all we have to go on in determining the truth is our judgement; and what judgement gives is only an appearance, how we take reality to be. But judgement aims at reality, not appearance; reality, not appearance makes a judgement true or false. Now the sole point of this assertion is its rejection of the phenomenalist’s conception of ‘truth’ and ‘reality’. Taken out of its dialectical context it becomes a mere truism; an assertion which no-one, not even the phenomenalist, would wish to deny. But even within its proper context, it does not tell us what truth is. Having worked through the dialectic of idealism, we remain ignorant of that ‘reality’ which ‘cannot be reduced to appearance’, of how reality ‘makes judgements true or false’. If there remains more to be discovered about the nature of truth, it can only show itself in a new dialectic, in the rejection of a second illusion.


187. Whether a proposition is true is something which reality determines and at the same time something which, in a different sense, we determine. We determine that reality determines that a proposition is true. And reality determines which propositions we ought to determine as true. These two different uses of the word ‘determine’ are nevertheless equally justified: They are the two senses in which the ‘value’ of a ‘function’ is ‘determined’ for a given ‘argument’. Reality is the function of determining an ‘output’ of truth or falsity for a given propositional input; and in our capacity as knowers, our role is to determine what value the function has for that given input. There is much to say about what our ‘determining’ amounts to, about all that goes on between input and output. One might therefore put a parallel question to reality: How does the function which reality is determine the truth or falsity of judgements? and moreover what makes the ‘truth’ and ‘falsity’ which ‘reality determines’ the very same ‘truth’ and ‘falsity’ which ‘we determine that reality determines’? Now instead of seeing two questions here we should regard the second as setting a constraint upon the acceptability of answers to the first: The truth and falsity which we judge is the very same truth and falsity which reality makes true or false. For what judgement aims to judge is nothing other than reality; and judgement succeeds in that intention. If it did not do both those things then judgement would not be judgement and reality would not be reality.


188. Our kind of determining is ‘investigating and finding out’. Our finding out finds out what it is that reality has determined: that certain propositions are true. The truth of propositions is ‘up to reality’ and not ‘up to us’; we investigate in order that reality should tell us which propositions are true. The following picture now suggests itself: The truth of propositions is realtity’s product; something made by reality. Investigating and finding out ‘unearths’ the product and brings it to light. Of course, taking the picture literally would lead to immediate absurdity, What we investigate is none other than reality itself; everything that reality ‘produces’ is itself reality and, if it were not, then investigating that would not be investigating reality. Still, one can accept the obvious fact of the matter and remain impressed by the metaphor of reality ‘making propositions true’. The metaphor expresses our discovery that there is something to be discovered; that there is something to be understood which we do not yet understand; a metaphysical problem about: the way reality produces its ‘product’, about what it is that makes propositions true.


189. The truth which reality produces is an abstract, not a physical product. A physical product would be, for example, roots one digs out of the soil. And one can imagine being in a primitive society where truths are ‘determined’, say, about a member’s loyalty, by inspecting dug-up roots. But then the truth about his loyalty is not the roots, nor in the roots, but what they mean. Determining the truth is not obtaining some thing from reality; although, in order to determine the truth we must indeed obtain things from reality and investigate their properties. Now what the things which we obtain from reality ‘mean’ is not other things in reality; for if what the roots ‘mean’ is the, loyalty then the truth about the roots themselves would be those very roots, which is absurd. What the roots mean is rather something which stands in a certain ‘relation’ with reality, in the sense of not being simply ‘in’ reality while yet remaining ‘connected’ with reality in some way: Determining what the roots mean is determining the content of a proposition about reality. Truth appears as the abstract product of that ‘relation’, the result of reality ‘being the way the proposition says it is’. However, that still does not tell us how reality makes propositions true. The question we must now ask is: What kind of ‘relation’ is that which truth is a product of?


190. What are ‘propositions’ for? what is ‘truth’ for? what kind of ‘relation’ do we want truth to be? Consider: I can look out of the window and see that it is raining. In that way I come to know that it is raining. But suppose instead someone else looks; and then tells me: ‘It is raining’. As a result of hearing those words I acquire the very same state of knowledge: The words represent a possible state of reality in which it rains; and that representation is indeed correct. The truth relation thus appears as a ‘correctness of representation of reality’. One may say: A true proposition corresponds with reality. And this is the point we have reached: The kind of ‘determining’ which is the function of reality is to simply be itself; a term in a relation of ‘correspondence’. In order to ‘make’ propositions true reality does not have to do anything!


191. To describe the truth relation simply as the ‘correspondence of a representation with reality’ seems, at first, as if it would have constituted a highly significant piece of information; if it had only been true. It also seems that it cannot be true. How, for example, can a ‘representation’, a mere picture, express the pastness of a ‘proposition about the past?’ A representation is something one can hold up against its object for comparison. But the past is gone. To say that a past-tense proposition ‘represents’ what its associated present-tense proposition ‘would have represented at the time’ merely employs another past-tense proposition, concerning ‘what it was possible to represent’; that ‘explanation’ is a vicious regress. The objection holds quite generally: If there is any information conveyed by the assertion that a given form ‘correctly represents’ reality other than that it is a true proposition (say, that it ‘represents’ in the way pictures represent) then we shall necessarily require, or need to assume, information concerning the exact specification of what is represented and how its representation is to be taken, which cannot be derived merely by examining the representation itself or the circumstances of its production. Thus, we say that a post-card is a ‘correct representation’ of the way the beach looked during the two weeks last summer when the weather was fine, provided that one allows for the over-intense colour-rendering. The post-card becomes an object referred to by a proposition describing how the beach looked.


192. However, as soon as the objection against truth as ‘correctness of representation’ is generalized, it becomes clear that, far from being ‘highly significant but false’, that characterization of truth must be true and cannot be informative. For what the ‘objection’ shows is that language is the one universal system of representation. In Tractarian terminology: The ‘pictorial form’ of language is ‘logical form’. All other kinds of representation may be incorporated into a piece of discourse; but nothing besides language, that system of representation whose pictorial form is not confined to any form other than logical form, can represent wholly and unambiguously. For the possibilities of reality transcend the possibilities which can be represented by any given pictorial form just in the case where the disambiguating specification of the manner in which representations of that form are to be ‘taken’ entails the transcendence of the confines of that form by means of linguistic representation. Truth, an essentially linguistic notion, is ‘correctness of representation’ in the only universal sense.


193. Language defines the conditions for the ‘correctness’ of all non-universal forms of representation. In practice, explicit definition may prove impossibly difficult; but that does not prevent giving instead an ad hoc explanation of how a given representation is to be taken (e.g. the post- card). The difference between ‘correct’ and ‘true’ is that there remains more to say about how the ‘correct representation’ is to be taken; and indeed the ‘more’ that remains to be said explains what ‘correctness’ for that representation is. By contrast, asserting a proposition is not like holding up a representation to an object; ‘correct’ if taken in one way, ‘incorrect’ if taken in another. Just because the conditions for the truth or falsity of the proposition are fixed by its intrinsic meaning, asserting a proposition is not producing a string of words whose interpretation waits upon an explanation of how we are to take what the words represent: The proposition is asserted as true. The truth of a proposition therefore tells us no more than what the proposition itself tells. For that reason, the ‘definition’ of truth becomes trivially easy: A proposition ‘P’ is true if and only if P. At the same time, just because there remains ‘something to say’ only about how reality makes non-linguistic representations correct or incorrect, the question of what truth itself is remains necessarily without answer. For if the question could be answered then it would not be truth but rather some other form of ‘correctness of representation’.


194. There is nothing to say about ‘how reality makes propositions true’ because there is nothing left to say. Of course, recognition of that fact does not simply ‘dissolve’ what appears as a deep metaphysical problem. However, one should remember that ‘metaphysical problems’ are not something given. Our problem’ is a problem which we do not know how to formulate in a coherent question, an unease which does not yet possess a determinate expression. What is ‘given’ is only the feeling of this unease. That can only mean one thing: The expression’ of the metaphysical problem of truth will turn out not to be the expression of a coherent question but only the expression of a metaphysical illusion.


195. The only way to determine the nature of this ‘truth illusion’ is to work through its dialectic. But we may assemble a suggestive ‘clue’ from the materials already provided. One’s first reaction to being told that the fact that ‘snow is white’ is made true by its being the case that snow is white, ‘grass is green’ is made true by its being the case that grass is green etc. is all there is to say about ‘how reality makes propositions true’ might be to object that ‘the generality of the question is not satisfied by merely giving specific instances’. That cannot be right, because the generality can be fixed up by asserting instead: ‘For all P, the proposition that P is made true by its being the case that p’. What the objection means is that one needs to be told what its being the case that P in relation to the proposition that P, its being the case that R in relation to the proposition that R etc. have in common. But it appears that that can be ‘fixed up’ too: What they have in common is a ‘relation’ between a linguistically defined entity and an ‘its-being-the-case-that...’. Give this second ‘term’ a name if you like (‘fact’); the only property it possesses in its own right, that which remains constant’ through ‘changing propositional content’, is to be ‘that...’. The ‘fixing up’ now becomes manifestly unintelligible. To that kind of question one cannot give an intelligible answer. And if saying ‘that...’ in the appropriate tone of voice seems to ‘answer’ some ‘question’ that is because this seeming expression’ of the seemingly ‘inexpressible’ is in reality the name of the illusion which prompts both question and answer.


20. The meaning of ‘realism’


196. The dialectic of the truth illusion is the dialectic of realism. The word ‘realism’ has of course many meanings, but the meaning of ‘realism’ intended here is what a certain would-be transcendent metaphysician whom we shall call the ‘realist’ means to mean by his meaningless utterances; the seeming ‘expression’ of his ‘metaphysical vision’. As with the dialectic of idealism, the problem of ‘adequate dialectical expression’ of a meaningless ‘meaning’ constitutes a major theme and reference point (1/9, 10). And as before, the dialectic is overlaid by the undialectical rejection of the meaning theorist’s false account of what the rejection of the illusion entails; in this case, the idea of a theory of meaning for propositions. Parallels with the dialectic of idealism will suggest a connection between the ego and truth illusions; this will be followed up. But one should note that it would be a mistake to force a ‘similarity’ in the structure and strategy of the two arguments: First, a seeming ‘similarity’ or ‘difference’ in surface structure can mislead as to the underlying dialectical form; for example, heuristic considerations may dictate the choice of dissimilar strategies for the expression of similar ‘underlying forms’; that depends upon the contingent facts of human psychology. Second, contingent historical facts influence the selection of particular strands of dialectic from all its possible elaborations; for example, realism, unlike idealism, does not have its ‘Kant’. The necessity of ‘undialectical’ departures from the dialectic in response to a particular opponent, of a heuristically motivated choice of expression and historically motivated selection shows only that a work of metaphysics is written at a time and a place: For that reason, the ideal of remaining true only to the inner voice of its own dialectical necessity, renouncing all ‘influences’ and the need to communicate, is a self-deception. But the dialectic loses none of its necessity for recognizing that fact.


197. We begin with a very ‘undialectical’ motivation for embarking on a critique of the truth illusion: I find myself wanting to say (speaking for the idealist in me) that ‘after all is said and done, what idealism means is still true’. I accept all the ‘arguments’; but the plain fact stares me in the face: This is my consciousness, my ego resides here, looking through these eyes (6/56, 7). In a metaphysic of action I cannot even ‘mean’ that metaphysical fact which distinguishes me qua subject of experience from all other subjects of experience; I cannot find myself. I can no longer pretend that these words ‘mean’ anything; they are just what I feel compelled to say. Now the irony of this predicament is that the compulsion appears only when one attempts to ‘do metaphysics’. Even a metaphysics which sets itself the goal of rejecting the illusion can do so only by bringing it into consciousness so that it can exercise its ‘effects’. So why not simply give up metaphysics, refuse to give the illusion the opportunity to manifest itself? I cannot think of a satisfactory answer to this question, one which is not at some level question-begging. (For example, if one speaks of the ‘subliminal’ or ‘unconscious’ influence of the illusion which a metaphysician’s self-knowledge could at least identify, there remains the question whether there is not some other way of destroying this ‘influence’.) But, as luck would have it, a second illusion appears to occupy my thoughts. For as long as I am distracted by this second illusion, the ‘influence’ of the first fades away.


198. As in the case of pure idealism (3/28), the truth illusion first presents itself as a ‘discovery’; a way of seeing things which never occurred to me before: ‘I am holding an ancient bronze vase. There are many things which one could discover about its history; for example, who owned it, what it was used for, when and how it was made. But the concrete detail is lost forever. I could invent an indefinite number of stories featuring this vase anyone of which ‘could be true’. But now it dawns upon me that I possess an extraordinary power: Just by entertaining one of these possible histories I might, unwittingly, think something which is in fact true. What I previously took to be a truism, that ‘true’ does not mean the same as ‘can be known to be true’, now seems to possess a deep metaphysical significance. Thought, unfettered by time or the limitations upon one’s practical and intellectual powers, does not stop short at what can be known but penetrates to the very heart of reality. This ‘extraordinary power’, which no spatial or temporal distance can defeat, might equally be attributed to reality itself, by virtue of the fact that in order to make my thoughts true all reality has to do is simply be the way my thoughts represent it as being’ (19/190).


199. Discounting the metaphor, there remain only the truisms that truth transcends our ability to know the truth, and that a thought is true when reality is the way the thought represents it as being. The project of realism is to discover a non-metaphorical expression of its ‘vision’. And for each new ‘expression’ which the realist ‘discovers’ we have to find a means of showing that it cannot be used to ‘mean’ what the realist intends it to ‘mean’. Thus, the realist might say: ‘understanding fiction involves ‘having thoughts’, representing possible states of affairs to myself. But when I speculate about the history of the vase I am doing more than that: I am aiming my thoughts at reality. And for all anybody could ever know, I might accidentally succeed in representing how things really are.’ — We must ask: What does ‘success’ mean here? The only explanation which the realist has so far given turns out to be trivial: One ‘succeeds’ in representing how things are when the way one represents them is the way they really are.


200. Let us call questions which we are unable to get into a position to answer ‘undecidable’. It is sometimes difficult to be certain when a given question really is undecidable; but one can give reasons for believing that some questions are indeed so. (In the case of historical propositions, for example, ‘evidence’ can be obliterated.) Now someone might maintain that a belief founded upon the mere enumeration and rejection of all the possibilities we can think of cannot amount to knowledge; or else that it remains hypothetical on the condition that no relevant possibility has been overlooked. In the case of a proposition which we ‘believe’ to be individually ‘undecidable’, it is not clear whether one can say that the possibilities of its decision have been merely ‘enumerated’ or genuinely ‘exhausted’. But we may bracket off the uncertainty over the status of a ‘belief’ founded upon enumeration and rejection of possibilities and from its application to the putative concept of ‘individual undecidability’. It suffices for the objectivity of that concept that we can conceive of a possible world in which there were genuinely ‘undecidable’ propositions. The realist’s ‘deep truth’ might now be reformulated as follows. ‘At least some undecidable questions have answers’. Taking this in the spirit in which it is intended, we should not attempt to raise difficulties for the notion that some propositions are undecidable. Instead, we now point out that the notion of ‘having’ an answer is no more informative than that of ‘success’ in representing how things are. I can ‘succeed’ in representing the truth by means of undecidable propositions because their associated propositional questions ‘have’ answers which ‘determine’ that some of them are ‘in fact true’.


201. Suppose now that, for the sake of dialectic, one denies that undecidable questions ever ‘have’ answers. The realist will argue: ‘you must accept that the proposition which serves as the content for a given propositional question is either true or false; there is no other possibility. Therefore, the answer will consist either in that proposition or its negation; one of them must be true.’ — We reply that the proposition is indeed ‘either true or false’ but that there is no ‘answer’ to the question whether it is true or whether it is false; it is not ‘determinate’ whether the ‘determinate state of affairs’ which it represents obtains or not. The realist retorts: ‘But if you admit that a proposition represents a determinate state of affairs, that means that in understanding a proposition one forms a conception of how things would stand if it was true. In that case some of the states of affairs represented by undecidable propositions actually obtain, while others do not: If one can conceive of how things would stand if a given undecidable proposition was true that means that the proposition can be true. It follows that some undecidable propositions must be true — We simply deny that there ‘exists’ an ‘answer’ to the question which undecidable propositions are true. Now the meaning of either... or... and of the ‘conception of how things would stand if...' will be followed up in later chapters (Chs.26, 24). But that is not called for at this stage of the argument. For either our denials have a significant content’ or not: If they have then the realist has yet to provide adequate grounds for his assertions, besides the fact that they are prompted by what, for all he knows, is only the semblance of a ‘metaphysical vision’; if they have not then neither have the realist’s assertions.


202. One ought now to be tempted to protest that this ‘dispute’ is only a typical example of the nonsense philosophers get up to when they discuss metaphysics. One party only thinks he means’ something while his opponent only thinks he is denying’ something. In addition, both parties only think that what the one ‘means’ the other denies. But in truth they are both just playing with words: There must indeed be occasions where illusory or wrong- headed ‘disputes’ of this nature occur. In that case, we must show why the dialectic of realism is not an example of such a dispute. However, looking further into the objection, there appears a disguised version of the fundamental problem of the possibility of a dialectic of illusion: How can we know what the realist ‘means’, and consequently what it is that we mean to ‘deny’, if realism really means something meaningless? and what kind of ‘meaning’ can be attributed to that seeming ‘denial’?


203. The dispute would be pointless if realism were not an illusion, and if there were no means of providing that illusion with a ‘dialectical expression’. But of course that only labels the problem. The dialectic has yet to be worked through, and can be judged only as a whole. The story of the realist’s discovery’ (198) is not the expression’ of realism; and the ‘denial’ of what the realist ‘asserts’ is not its rejection. Neither stands up on its own. only when the dialectic is ‘worked through’ do they acquire a significant meaning; and that meaning can then be characterized only as their ‘contribution to the dialectic’.


204. Bearing in mind the fact that every contribution to the expression of the illusion is a contribution to its refutation, and vice versa, our strategy will be as follows: First, the ‘meaning’ of realism will be drawn out, through the realist’s account of an objective world (Ch.2l). That account, far from ‘explaining’ or ‘justifying’ realism, ends up supplying a powerful motive for its rejection. Only then shall we attempt a proper ‘refutation’ (Chs.22, 23).


21. The idea of an objective world


205. The realist in me wants to say: ‘The very idea of an ‘objective world’ presupposes the truth of realism; and the existence of such a world is itself a presupposition of rational discourse. If reality were not so constituted as to contain answers to our questions, irrespective of our being in a position to discover those answers, then there would be no objective target for our thoughts to aim at. Truth and falsity would cease to matter; and language, divorced from its function of ‘representing reality’, would lose all its meaning.’ — The objection simply asserts that ‘if realism were rejected, reality would be reduced to appearance’ (19/185). As it stands, it is merely rhetoric, and not an ‘account of the objectivity of the objective world’. However, from our ‘biased’ viewpoint, we can construct the missing argument: One of the symptoms of the truth illusion will manifest itself as the realist’s picture of an ‘anti-realist reality’. In this picture, ‘reality’ retains the metaphysical status which the realist accorded it; only it loses its ‘solidity’, and becomes instead a ‘bare skeleton’ surrounded by ‘nothingness’. ‘Anti-realism’ becomes a metaphysical position defined within a realist metaphysic. And this ‘position’ reduces reality in the most drastic way: That ‘reality’ which appears to us, that which remains within the bounds of our capacity to acquire knowledge, is the only reality; all the rest is reduced to nothing. Of course, we have no right to protest that this is not what the ‘rejection of realism’ means, prior to our articulation of an alternative ‘picture’. But it had better not be a picture of a reality ‘reduced to appearance’.


206. The rejection of realism is not the denial that ‘truth transcends verification’, . However certain we may be of a proposition’s truth, it is possible that the proposition is nevertheless false. In the case of mathematics it is not obvious how proofs previously taken as valid might ‘turn out to be invalid’. But mathematical practice is founded upon the additional empirical assumption that we can rely upon our ability to spot ‘mistakes’. Note that ‘possible’ must not be understood here in the sense of ‘for all we know, it may turn out that...’. If we know anything at all, such ‘epistemic possibilities’ are not always possible. But what is not ‘possible’ in the epistemic sense remains conceivable; a possible state of affairs which we can represent to ourselves in ‘language without thereby casting doubt on the possibility of knowledge (9/86-9). If, by contrast, our forming the judgment that a proposition was true were to constitute its ‘truth’ then reality would have been ‘reduced to appearance’ (19/186). There can be no sense of ‘verified’, understood as what we take ourselves to have ‘achieved’ as the result of our investigations, which logically entails the ‘truth’ of a proposition.


207. We have already seen how this ‘corrigibility’ of verification leads to global scepticism within the metaphysic of experience (9/83-5). But the realist finds himself in the same predicament. Judgement is said to sometimes ‘succeed’ and sometimes ‘fail’ to get hold of the truth. The possibility of knowledge is the possibility of knowing when judgement has achieved its aim. For the realist, however, the sense in which a question ‘has’ an answer cannot depend upon whether or not we are able to ‘determine’ that answer. For it is not as if reality ‘has’, in the ‘realist sense’, the answers only to undecidable questions. The ‘real’ answer is always ‘determined’ in the same way by reality, no matter what the question; and the point of judgement is to discover that ‘real answer’. But all judgement can ‘succeed’ in doing is simply being true; the intention of judging, to ‘discover the truth’, never ‘succeeds’. For our ‘determining’ the truth of a proposition must always remain provisional upon whether reality itself ‘determines’ that the proposition is true. Nothing could count as an answer to that question because every ‘answer’ is only our’ judgement’; in which case the very same question arises. The realist is thus condemned to global scepticism.


208. In the absence of an ‘adequate expression’ of the truth illusion, the ‘argument’ that ‘realism entails scepticism’ could not be understood. The connection must remain problematic until the dialectic has been ‘worked through’. But now I want to ask a question which, perhaps, cannot be resolved. Is the realist’s ‘scepticism’ on the same level as the ‘global scepticism’ entailed by a metaphysic of experience (9/83-5)7 or should it be compared with the ‘scepticism with respect to an external world’ which makes way for pure idealism (4/36-7, 9)7 If the latter, then we should expect a line of argument which proceeds from an ‘impure’ realism, rejecting its analogue of the naive idealist’s ‘external world’ scepticism, to a ‘pure’ realism, vulnerable at worst only to an analogue of the phenomenalist’s ‘global’ scepticism (3/28, 31). The sceptical argument, as presented, could then be read as leading to either of the two ‘analogues’. The elaboration of a ‘possible dialectic’ will be necessarily speculative.


209. Consider the assertion: ‘The real answer to any question is always determined by reality’ (207). One might suppose its ‘truth’ to consist in the fact that reality ‘divides’ itself into Platonic ‘forms’: The ‘real answer’ is a ‘relation between forms’, each form being in its own way ‘related’ to phenomena (13/122). But we have seen how this leads to an ‘irresolvable scepticism about the possibility of ever knowing when a seeming ‘concept’ really corresponds to the divisions of reality as it is in itself’ (16/152). The logical extension of this thought is that, far from knowing which propositions are true, we do not even know what our words mean. But if we do now know that then we cannot know what talk of ‘Platonic forms’ means. This is not comparable to the supposed ‘knock-down’ argument against scepticism, that the sceptic cannot claim to ‘know’ that scepticism is true, If the only thing I know is that scepticism is true then I know nothing else; but if I do not even know that then I know nothing at all; either way, I at least know nothing else. By contrast, throwing doubt upon the meaning of ‘Platonic form’ attacks the source of scepticism concerning meanings, If I know what ‘Platonic form’ means then I do not know what any word means; therefore I do not know what ‘Platonic form’ means. The ‘pure’ realist thus rejects the Platonist conception of how reality ‘determines’ answers to our questions; while not going so far as to reject the thought that the ‘answers’ which we are able to ‘obtain’ do not suffice to ‘determine how things stand in reality’. This ‘pure reality’ does not ‘divide up into forms’: Reality is not the agent which ‘determines answers’ in that crude sense; rather, it simply is the function of determining answers to questions (17/158). But now we are deprived of the words to express why this ‘determining function’ renders knowledge problematic; what, in the way of ‘determining answers’ reality can ‘do’ which prevents us from determining which answers reality determines. The ‘reality’ to which we attribute this ‘power’ is not a describable ‘something’, but neither is it nothing. It is simply ‘that...’ (19/195).


210. If realism were to be rejected, in what sense could our thoughts be said to ‘depict a verification-transcendent reality’? ‘Reality’ is the source of the pragmatic constraints upon what may be taken to be a ‘procedure for investigating reality’. That does not mean that truth is ‘whatever works in practice’. The validity of a putative ‘answer’ to a question is determined not by its individual ‘practical consequences’ (18/178) but through its location within a system of questions and answers. The pragmatic constraints upon the operation of that system are holistic. Now this would not satisfy someone who wanted to know ‘what it is for an ‘answer’ to be valid’. The rejection of realism is the rejection of a simple, though far from trivial, answer to the question. ‘What is truth?’. ‘P’ is true if and only if ‘that...’: namely, P’. That might be interpreted as the rejection of the ‘simple’ answer in favour of one which is ‘very complicated’; an account of all the ‘investigative activities’ which take the place of the ‘that...’. But the ‘very complicated’ answer would amount to a criterion of truth; and that is ruled out by the corrigibility of verification (206). (At most, we can aim for a second-order ‘methodology of methodologies’; 24/243-4.) The answer to the question: ‘What is truth?’ turns out to be simple and trivial. The question whether ‘P’ is true is the question whether P. The difficulty lies in the effort required not to misunderstand that simple answer; for that entails recognizing and resisting the realist’s illusion.


211. The realist finds our account of ‘verification’ abhorrent: ‘Imagine the world of Orwell’s 1984 where historical records are destroyed and ‘rewritten’. You maintain that the answer to a question is the product of ‘verification’; whatever the evidence ‘tells’ us is the ‘truth’. If we give up the idea that the truth of our beliefs is the product of how reality is in itself and not of mere ‘evidence’ then there is nothing to stop us believing what we want to believe; if we don’t like the facts then we can change the ‘evidence’ and our beliefs at the same time.’ — We may note that the realist, in placing his ‘reality’ beyond the range of ‘verification’, ends up in this very predicament. But that does not meet his challenge. In the world of 1984, the ‘methodology’ of deluded ‘historians’ is indeed’ justified by practice’; for they are the ones who survive! There must be a deeper connection between truth and ‘practice’ than that conceived by such a crude ‘pragmatism’. And we have already given an account of it: The very possibility of belief, whether true or false, involves the existence of a subject who acts in order to satisfy his desires and risks failure (6/58). If he could believe whatever he desired to be the case then there would be no such ‘risk’; he could not be said to act in such a way as to warrant the ascription of ‘false beliefs’ and consequently could not be said to possess ‘beliefs’ at all (8/63).


212. I can deceive myself into thinking that I am investigating the truth when I am merely creating a fantasy world. But the recognition that my propositional attitude is prompted merely by ‘evidence’ which I have made up is recognition that the attitude is not belief. But might not the fact that the proposition was useful for me to believe be taken as the ‘evidence’ for its truth? If I were to lower my sights and genuinely consider only the question whether the proposition was useful to believe, the answer to that question would not itself depend upon what it was useful to believe concerning ‘what it is useful to believe’. To equate the question of truth with the question of what it is useful to believe leads to vicious regress.


213. Our methods of investigating reality provide us with the answers to our questions; there is no repository of ‘real answers’ beyond the range of our ‘earth-bound’ perception and intellect. The ‘correctness’ of our answers is a function, not of ‘agreement’ with an unknowable ‘reality’, but of that same practice which determines what we take to be a ‘correct answer’. But now it might occur to us to ask whether the ‘practice’ might have been different from the one with which we are familiar. Consider the practice of ‘divination’ by means of witchcraft in a primitive society. If their practice ‘succeeds’ in its own way does that not’ justify’ their ‘methods of investigation’? No: It is not sufficient that ‘practice’ performs some useful purpose. No doubt the society could not ‘do without’ witchcraft. But that might only mean that witchcraft, though ‘useful’ to believe in, is nevertheless false. However, we may bracket off the question whether there is any empirical evidence that truth is ‘relative’ and consider only the question whether we can conceive of a possible world where it might have been so: That reality is completely ‘accessible’ to science is our commitment, rather than a mere fact which we ‘discover’ (12/110). But one can conceive of circumstances which would render that commitment no longer tenable. Science, or the ‘scientific method’, would then become what is in its own eyes a ‘pseudo- science’; a seeming ‘procedure’ for ‘investigating’ reality which lacks a coherent account of its own possibility. In such a world, rival ‘sciences’ might compete for the ‘explanation’ of the very same phenomena: How one set about ‘investigating the truth’ would then depend on one’s allegiance.


214. The realist will not allow that this is a coherent description: ‘We may convince ourselves that ‘truth is relative’, but we would be wrong. For what is true is not ‘relative’ to what we take to be true; and could not, therefore, inherit the ‘relativity’ attributed to that ‘taking to be true’.’ — Let us, for the sake of argument, accept that point and ask instead: Why could realist ‘truth’ not itself be ‘relative’; not by ‘inheriting’ the relativity of ‘taking to be true’ but in itself? Taking that thought a stage further I Once ‘real truth’ is cut loose from ‘taking to be true’, we may suppose ‘real truth’ to break up into an infinite number of ‘relative truths’! The realist may reply. ‘Reality is reality; and a proposition either does, or does not, serve to represent reality. If it does, then the question whether or not it agrees with reality is not relative but absolute.’ — That is not an argument but only what the realist has got to say. But the words themselves mean nothing; they are merely the symptoms of the realist’s illusion.


215. The sole ‘meaning’ of ‘realism’ that emerges from the realist’s account of the ‘objective world’ consists in an argument for global scepticism. But that unhappy consequence does not refute the doctrine (10/90). Moreover, our failure to grasp what the realist means by a question’s ‘having’ an answer remains a failure to grasp the nature of his illusion. We have still to refute realism; and indeed discover what it is that our refutation ‘refutes’.


22. The refutation of realism (1)


216. ‘I aim my thought at reality and reality either makes it true or makes it false. The question of its truth is purely a matter of the content of the thought and of how things stand in reality. It has nothing to do with my relation to how things stand; whether or not I could ever come to know the answer to the question is irrelevant: The question must therefore have an answer’ (20/198-9). — It seems, to the realist in me, that I know exactly what I mean by the therefore...; and furthermore that there could not be room for dispute. And yet there appears an opponent who just as confidently contradicts me (20/201). I want to say: Either he just refuses to see or else he doesn’t really mean what he says. But that means the sole ground for my assertion is something that I ‘see’ and cannot imagine anyone not ‘seeing’; an incorrigible metaphysical intuition (2/14).


217. There are no contradictions in reality. We cannot call the dispute a ‘paradox’ and leave it at that. This question at least must have an answer; there must exist a continuation of the dialectic, only we have yet to discover it. If realism is indeed a metaphysical illusion then the dispute, as presented, can be taken in two radically different ways: First, both the realist and the ‘opponent’ who rejects realism may be ‘transcendent metaphysicians’; like the ‘mind-body identity’ theorist and his opponent (12/112). In that case, the real ‘illusion’ embraces both realism and its rejection; but then we have no clue as to what that illusion might be. Alternatively, the ‘denial’ of realism might be understood in the same way as the challenge to the phenomenalist. the ‘assertion’ that the phenomenalist’s ‘objective world’ is only a ‘dream world’ (5/47-8). However, if it is to be so understood then there is a vital component missing; for it does not yet amount to a ‘refutation of realism’. That we do not yet possess the means to decide between these two interpretations means that the dispute itself does not yet have a determinate place in the dialectic. It does not indicate a way forward. Let us then consider instead strategies for ‘resolving metaphysical disputes’.


218. The seeming ‘question’ of mind-body identity turned out, on examination, not to be a question; the very same consequences followed either from identity or non-identity, there was ‘nothing to choose’ between the two (12/115-6). Furthermore, we were able to identify the illusion which gives rise to the appearance of ‘thesis’ and ‘antithesis’; that there is ‘something which presents itself to me’, the ‘this...’ of idealism (12/117). By contrast, a considerable ‘consequence’ at least appears to follow from the resolution of the question of realism; whether we know anything at all about that ‘reality’ at which our thoughts aim (21/107). The mind-body question must therefore be the wrong paradigm. What about the alternative? We have seen how the metaphysic of experience is in reality a metaphysic of pure idealism (4/42, 5/46-7, 7/66-7); and that, with the rejection of phenomenalism, the only consistent ‘expression’ of pure idealism is after all only pure silence (3/31-3). The idealist is therefore powerless to oppose our rejection of his ‘terms of reference’ (3/34, 6/55), our ‘constructive’ refutation (6/58). But the realist is far from being reduced to silence: Whereas the idealist’s recognition of the sole reality of the ‘this...’ entailed the rejection of all the apparatus of conceptual thought, the realist’s refusal to say anything about reality in respect of its capacity to ‘determine’ answers to our questions except that it is ‘that...’ is precisely intended to save our grasp of the meaning of our words from the consequences of the Platonist’s ‘impure’ realism (21/209).


219. If the realist is not first ‘reduced to silence’ then nothing we choose to say in ‘opposition’ to realism will carry any weight. If, for example, we were to construct an account of what it is for a question to ‘have’ an answer (‘either we know the answer, or we know where to find the answer, or...’) the realist would simply reject it; that is not what he means by a question’s ‘having’ an answer. Nor could we expect to get away with ‘explaining the illusion’; the purported ‘explanation’ would serve only as a psychological hypothesis (18/174) which the realist remains free to reject. We might have indeed found ourselves in the very same predicament with idealism. Had we allowed the idealist the first ‘description’ of his ‘new experience’ (3/28) we should have been unable to refute his ‘position’. He would have continued to assert, while we could only deny, that the only ‘real’ things are his own ‘mental contents’. But when we examined the meaning of the idealist’s ‘description’ it turned out that the idealist must deny himself the means to ‘express’ his idealism (3/29-30). If the same strategy applies to realism, then the first, seeming expression’ of the realist’s discovery’ (20/198-9) must be shown to have a different ‘meaning’ from that which the realist ‘means’ to attribute to it; that there are indeed no words to express what the realist ‘means’.


220. Previously, when the realist put forward the generalization: ‘At least some undecidable questions have answers’ in order to explain how it is possible to ‘succeed’ in representing reality by means of an undecidable proposition, our objection was that the concept of ‘having’ an answer is no more informative in this context than that of the ‘success’ of a representation (20/200). As used by the realist, their connection is truistic rather than mutually explanatory; both amount to no more than a bare assertion of realism, unable to defeat its bare denial (20/201). Our strategy now will be to investigate the meaning of ‘some undecidable propositions are true’; and to demonstrate that it cannot be used to mean what the realist intends it to ‘mean’. Since the whole point of the argument is to show that the realist does not ‘mean’ anything, the very idea of such a strategy is problematic: For it is not as if the meaning of ‘some undecidable propositions are true’ and the ‘meaning of realism’ may be compared’ with one another, and the purported ‘equivalence’ shown to fail. But rather than attempt to prove the coherence of its strategy, we should allow the argument to speak for itself.


221. The use of the predicate ‘ true’ does not assert a relation of ‘correctness of representation’ between the proposition of which it is predicated and reality. For the assertion of the truth of a proposition asserts no more than what the proposition itself asserts. By contrast with the act of showing, someone a picture, the point of assertion takes care of how that act is to be taken: The asserted proposition means its truth (19/193). But if the assertion that ‘P’ is true adds nothing to the assertion that P, the predicate ‘... is true’ appears redundant, Why not simply assert that P7 If ‘P’ is in fact a whole book then signaling one’s agreement with everything it asserts by ‘What the book says is true’ saves a lot of time. However, if instead we are only sure that some of the book is true, but are not sufficiently confident to identify the relevant parts, then ‘some of what the book says is true’ is the only thing we are in a position to say. That is the best case that can be made for the non-redundancy of ‘ true’. But even here the predicate is eliminable: ‘For some P, (in what follows, ‘EP’) the book says that P. And P’. One might say that the point of the predicate’... is true’ is such as to determine its equivalence in ‘representational power’ to quantification over propositions. We may therefore replace the realist’s assertion, ‘some undecidable propositions are true’ by its equivalent’ ‘(EP)(P is undecidable and P)’.


222. To know what ‘(Ex)(Fx)’ means entails knowing what it is for ‘F ‘ to be true of a given object: Since the truth of ‘Fa’ may be said to verify (in a logical, rather than an epistemological sense) the proposition ‘(Ex)(Fx)’, we may assert that understanding that proposition entails knowing what would count as a verifying instance. That means knowing of some/object, first, that it makes sense to call it ‘F’ and that, second, if an object were indeed F then (Ex)(Fx) would be true. Moreover, to know that it makes sense to call an object F entails being able to understand the assertion that it is F. Our charge against the realist is that, in the case of ‘(EP)(P is undecidable and P)’ there is E such ‘understanding’. The grounds for that charge can be given in a single sentence; the rest of the argument consists in replies to objections. Thus: A necessary condition for recognizing an assertion as an assertion of a verifying instance for ‘(EP)(P is undecidable and P)’ is to understand that the proposition which takes the place of the variable ‘P’ is asserted to be true; but that is in itself sufficient to disqualify the assertion as an assertion of a verifying instance.


223. The requirement of ‘knowing what counts as a verifying instance’ may first appear too strict: ‘A person cannot express recognition of a verifying instance for ‘(EP)(P has never been written or uttered)’ without contradicting himself; yet surely we know what that proposition means I, — That is because the very act of asserting the putative instance materially alters the state of affairs which that instance was intended to describe. However, the following proposition counts as an adequate surrogate for a verifying instance. ‘If things were as they are now but for my assertion of this proposition then A (a proposition which has never before been written or uttered) would never have been written or uttered’. The truth of this counterfactual is constituted by the fact that A has never been previously written or uttered; together with the fact that the assertion of the counterfactual was not a necessary condition for the previous non-writing and non-utterance of A. (The point of this second condition is that the surrogate is adequate only if it does not materially alter the state of affairs which the verifying instance for which it acts as the surrogate was meant to describe.)


224. The realist first attempts to sidestep our ‘refutation’ by offering a proof of the disputed proposition: ‘suppose that A is undecidable. By the account of ‘decidability’ (20/200), not-A is also undecidable. From ‘A or not-A’ we may infer that (A is undecidable and A) or (not-A is undecidable and not-A). Either way, (EP)(P is undecidable and P).’ — We should note first that a proof of a proposition can never substitute for an understanding of what the proposition means. If we do not understand what has been ‘proved’ then we do not understand the proof itself. That does not mean that the proof is invalid; only that we o not yet know how to take it. Of course, if one holds a ‘theory of meaning’ according to which knowing the meaning of a proposition is knowing how to prove it (in some way or other), then one must take the argument as a reductio of the ‘verifying instance’ requirement. However, for the realist to take the argument in that way would be self-destructive. Now the crucial premise in the realist’s proof is ‘A or not-A’; whatever uncertainty we may have about the meaning of ‘(EP)(P is undecidable and P)’ reflects back on that premise. So far, we have no grounds for supposing either proposition to have any meaning; except for the brute fact of the proof itself. But if it is a fact then it ‘means’ something. Let us then say what we can about the ‘meaning’ of premise and conclusion, even at the risk of saying something senseless or paradoxical: The existence of a verifying instance for ‘(Ex)(Fx)’ would be the existence of an answer to the question. ‘Which x is F?’. If it is impossible for ‘(EP)(P is undecidable and P)’ to have verifying instances, while nevertheless remaining in some sense true, then there is no answer to the question: ‘Which undecidable propositions are true?’ and consequently no answer to the question whether A or not-A (20/201).


225. It will strengthen our grip on the argument to see why it does not apply to the proposition: ‘There exist objects in the universe which we shall never be able to encounter’. The proposition is indeed not the consequence of a problematic ‘proof’ but an unchallenged scientific fact. But that does not excuse us from giving an account of how we are able to understand the existential quantifier. However, that is no problem. In order to assert of an object that it constitutes a verifying instance it is necessary to identify that object. But we can secure identifying reference to an object without encountering it or even being in a position to encounter it, by the use of predicates of spatio-temporal location. By contrast, there is no such resort in the case of true propositions; their truth must first be decided. The realist can only assert, for some undecidable A, that a verifying instance for ‘(EP)(P is undecidable and P)’ would be ‘whichever of A or not-A is true’. But we now see how that move is futile. For that same reason it is useless to argue that, since unencounterable objects are, were they to be encountered, potential ‘grounds’ for the truth of undecidable propositions, it follows that propositions which satisfy the generalization. ‘ (EG)(EP)(P is undecidable and had object G been encounterable P would have been decidable and it would have been the case that P)’ are true undecidable propositions. For if there is no answer to the question: ‘Which undecidable propositions are true?’ then there is no answer to the question: ‘Which undecidable propositions would have been decidably true had their grounds been encounterable?’.


226. The realist now resorts to counterexamples: ‘you surely do not wish to deny that there are decidable truths which we do not know or even those which, as a matter of contingent fact, we shall never know. How then do you account for the meaning of: ‘(EP)(P is decidable but undecided and P)’, ‘(EP)(P is decidable but will never be decided and P)’?’ — In the case of the former proposition we have the ‘verifying surrogate’ (223): ‘If things were as they are now but for my exercise of a capacity to determine the truth of A, then A would have been decidable but undecided and it would have been the case that A’. Similarly, for the latter we have: ‘If things were as they are now but for my present and possibly future exercise of a capacity to determine the truth of A then A would have been decidable but never to have been decided and it would have been the case that A’. These surrogates are known to be true as a consequence of one’s knowledge of the truth of the proposition A. Their truth is constituted by the existence of that ground by means of which one was able to determine the truth of A, together with the fact that the exercise of that capacity to determine its truth was not a necessary condition for the existence of the ground (the difference between dropping a jug in order to determine whether it is breakable and flipping a coin in order to ‘determine’ whether it will come up heads). Now the point of the concept of a ‘verifying surrogate’ is to discount any disarrangement of the reality at which one aims existential generalizations caused by actions which asserting an example of a verifying instance necessarily entails. That rules out the following counterfactual as a ‘verifying surrogate’ for ‘(EP)(P is undecidable and P)’. ‘If things were as they are now but for my being able to get into a position to determine the truth of A then A would have been undecidable and it would have been the case that A’. For to suppose that one might not have possessed the ability to get in a position to determine the truth of a proposition is to suppose that reality itself might have been different in certain respects from the way it in fact is, and not merely that one might have chosen to act differently, other things remaining as they are. (Whether possession of the ability might itself have been a consequence of one’s actions is of course irrelevant; for those actions would then have gone beyond what the act of ‘asserting a verifying instance necessarily entails’.) Whether that ‘difference’ makes any difference is precisely the question at issue.


227. Note that the truth of counterfactuals of the form just now rejected as verifying surrogates for ‘(EP)(P is undecidable and P)’ is not in question. However, divorced from that role, they cannot be taken as expressions of what the realist ‘means’. To merely conceive of a possible world in which a question to which I have an answer would have been undecidable but still have the same answer is not the same as asserting that a question which is undecidable nevertheless ‘has an answer’. The difference between the mere conception and the assertion could be thought ‘not to make any difference’ only if ‘(EP)(P is undecidable and P)’ meant what the realist intended it to ‘mean’.


228. The realist’s last resort is to shake our confidence in the ‘rejection of realism’, and is not an objection to the argument itself. ‘The purported verifying surrogates for ‘(EP)(P is decidable but undecided and P)’, ‘(EP)(P is decidable but will never be decided and P)’ are inadequate. For one may accept their truth while denying that questions have answers until the moment when they are decided. You surely do not wish to take your ‘rejection of realism’ that far.’ — But of course we do. Indeed, E questions, whether ‘decided’ or otherwise ‘have’ answers in the realist’s sense. In that rejected sense, a question’s being ‘decided’ by us does not ‘decide’ the answer which it really has, which reality ‘gives’ it (21/207). In place of that rejected sense, we offer the ‘surrogates’ as a possible illuminating account of what ‘having an answer’ means. But if they seem too trivial to be illuminating it doesn’t matter. Their meaning consists in what is implied by the fact that they are all we are prepared to offer; in their contribution to a dialectic which rejects the thought that ‘having an answer’ means something more: It is the dialectic that counts.


23. The refutation of realism (2)


229. Undermining the intended ‘meaning’ of ‘(EP)(P is undecidable and P)’ undermines all connected ‘expressions’ of realism: the possibility of ‘success’ in representing reality by means of an undecidable proposition (20/199), that undecidable questions ‘have’ answers (20/200), that either a proposition or its negation is ‘made’ true by reality (20/201, 21/224); indeed, that thought and reality possess an ‘extraordinary power’ in their combining to ‘determine’ the truth or falsity of undecidable propositions (20/198). But we are still only at the stage of the dialectic of realism corresponding to the non-constructive refutation of idealism (4/35): Denying the realist the means to express his realism does not destroy the ‘that...’; any more than reducing the idealist to silence destroys the ‘this...’ (3/31). A purified ‘pure realism’ corresponding to this ‘pure’ idealism will go beyond the mere rejection of the Platonist’s ‘impure’ realism (21/209) and now accept the futility of all attempts to say what realism means: Not only is it impossible to ‘say anything about reality in respect of its capacity to determine answers to our questions except that it is ‘that...’ (22/218), we cannot even intelligibly ‘ascribe’ any such ‘capacity’ to the ‘that...’. We can only mean it. When I aim my thought at reality my eyes remain fixed upon the ‘that...’. That is how the act of expressing a proposition comes to mean that the proposition is true.


230. In one important respect, the silence of the idealist and the realist still appear to differ: Whereas the pure idealist’s silence was total, the rejection of the very possibility of ‘applying a concept’, the realist has been silenced only in his attempt to say what realism ‘means’. However, the thought that the realist can continue to make judgements about reality proves a mere self-deception: If we can never know whether a judgement is really true (2l/207), then this very act of ‘aiming’ a thought at reality loses its point. Either our judging does not concern what is really true, in which case it is not judging about reality; or else we can say what we like without risk of reality ‘proving us wrong’; in either case we are not judging. Now in attempting to continue to address the metaphysic of idealism in the face of the idealist’s silence, we were fortunate to encounter its attempted ‘embodiment’ in the metaphysic of phenomenalism; recognition of the sole reality of the ‘this...’ becomes a metaphysical project (4/42, 5/46, 7/66-7). There appears no parallel ‘project’ for realism. But we should remember that the whole point of the project of phenomenalism was only to express a model of the relation between mind and reality (5/45, 51-2). The constructive refutation of idealism (6/58) presupposes the rejection of the model of a ‘passive observer’, the phenomenalist’s ‘transcendental ego’ and the ‘world’ of its ‘possible experience’, as the embodiment of the illusory ‘this...’ (5/55-8). The constructive refutation of realism must therefore identify the realist’s ‘false model’.


231. When we previously looked to the rejection of naive idealism for a strategy with which to reduce the realist to silence (22/219) a crucial component was overlooked: In analysing the purported ‘description’ of the idealist’s ‘new experience’ (3/28) we discovered that the idealist had illicitly borrowed the notion of ‘applying a concept’; that the ‘consistent’ idealist must reject the use of concepts (3/29-30). We have not yet attempted to locate the ‘illicitly borrowed notion’ in the expression of the realist’s ‘discovery of the amazing power of thought’ (20/198-9). It is not obvious how this would leave us better placed than we are already, how it could amount to more than just another strategy for denying the realist the means to express his realism. but this second refutation’ might at least indicate the way forward.


232. For the very same reason that one can succeed in representing reality by means of a proposition only because its associated propositional question has an answer (20/200), so a question may be said to ‘have’ an answer only if one may be said to ‘succeed’ or ‘fail’ in the attempt to correctly represent reality by means of a proposition which purports to answer it. But one can ‘succeed’ in performing some action only if it makes sense to say that one intends to perform that action. The realist must therefore be able to make sense of his own ‘intention’ to ‘represent reality by means of the undecidable proposition that...’. A necessary condition for ‘making sense’ is being able to distinguish that intention from a different intention. But consider now the manner in which fiction ‘represents’ reality. Fiction is not ‘aimed at reality’ but only ‘supposed’ (20/199). One may conjecture that a proposition is true or only entertain the supposition that it ‘might be’ or ‘might have been’ true. (In the case of propositions recognized as ‘undecidable’, the realist will admit that judgement is, strictly speaking, out of the question; it is only when one follows through the consequences of realism that one discovers that all’ judgements’ are only ‘conjectures’, 230.) Conjecture is a move one can play in the game of ‘investigating reality’, distinguished from the move of ‘entertaining a supposition’ by its consequences. The point of conjecture is the commitment which one accepts when one makes a conjecture; for example, to follow certain lines of investigation, to choose certain questions to put to reality, to interpret evidence in a particular way, Our charge against realism may, once again, be put in a single sentence (22/222), Thus: The claim that one’s thinking or saying or writing down a proposition which one recognizes as ‘undecidable’ is ‘meant as a conjecture’ is otiose.


233. If the realist asserts that he is not entertaining his undecidable proposition as a mere ‘supposition’ but ‘aiming it at reality’, we simply reply: so what? There is nothing the realist can say or do, and no ‘mental act’ that he can ‘perform’, which will suffice to turn ‘supposition’ into ‘conjecture’; for in the very expression of the seemingly ‘conjectured’ proposition, the recognition of ‘undecidability’, the realist rejects that ‘commitment’, which alone distinguishes the act of conjecture from mere supposition. Everything and nothing is the ‘right course of action’ to undertake on the basis of that seeming ‘conjecture’, We are not concerned here with psychological speculation about the kinds of mental act one can or cannot coherently perform: As a matter of contingent fact, social reality is so constituted that the attempt to realize certain ‘intentions’ regarding the attitudes or behaviour of other persons is self-frustrating. One finds that out from experience; but having found that out, one continues to ‘desire’ to bring about the impossible. By contrast, the realist’s ‘intention’ to ‘conjecture’ the truth of his undecidable proposition is logically self- refuting. The ‘distinction’ between different kinds of mental act to which the realist appeals is robbed of its very content by the use to which he would see it put.


234. But the illusion remains that one can ‘mean’ one’s ‘representing’ of reality as a ‘conjecture’ and not a ‘mere supposition’. It seems, to the realist in me, that all I have to do is to ‘turn my mind’ towards the proposition in a particular way; not to hold it at arms length but really mean it: That semblance of ‘really meaning’ is my vision of the ‘that...’ (229). Now some things Can be ‘seen’ and ‘shown’ but not put into words. But the supposed ‘psychic fact’ of my necessarily silent ‘meaning’ is not of that kind. For its. ‘existence’ is constituted solely by my seeming to ‘believe’ that it exists; my desire to ‘mean’ the proposition in that ‘particular way’ is constituted solely by my seeming to ‘desire’ it; an idealist’s ‘private object’.


235. The realist may now protest: ‘surely the proposition itself is made true or made false by reality. I do not have to do anything to ‘make it aim at reality’.’ — There is a sense in which what the realist says is quite correct. When we discover a new truth we do not suppose that the truth only ‘became’ a truth when our judgement was ‘aimed at reality’. The proposition which the judgement expresses was already true. For it is not as if we have first to breathe life into each expression of a new thought before it may be said to possess a truth value. But that is only because it inherits its life from the life of the language itself; something which is possible only when the inherited ‘life’ is a life which it is possible for the proposition to have. The bare propositional sign does not have a ‘life of its own’; its life consists in its potential use in the expression of propositional attitudes. To suppose otherwise is to treat the concept of a ‘representation’ as self- explanatory; to suppose that the possibility of a false ‘representation’ can be understood independently of the possibility of false belief (2/25-6). Thus, if the realist’s attempt to use a particular propositional sign in order to express his seeming ‘propositional attitude’ proves self-refuting then the ‘proposition itself’ cannot have the ‘life’ which the realist attributes to it.


236. This second ‘refutation of realism’ uncovers something which previously remained hidden: the realist’s model of the relation between thought and reality. Whereas for the idealist the very notion of ‘structured experience’ ‘is the notion of subject and object, of mind and reality’, the ‘substance in which mind and reality inhere as necessary correlates’ (6/61), for the realist, the idea that thoughts are thought by a subject does not even enter the picture. The proposition has a ‘life of its own’, independently of its potential to serve as the content of a subject’s thoughts. Now as it stands this ‘description of a model’ is either a metaphor or sheer nonsense: One might, fancifully ‘construct a metaphysic’ according to which each proposition was a ‘substance’ in its own right, a Leibnizian monad whose sole property is to ‘think’ the thought which it represents. However, there is an alternative, more familiar way of expressing the realist’s model. The subject who essentially is the thinker of thoughts is not a subject who ‘discovers’ truths, who’ judges’ or ‘conjectures’. For all thoughts are essentially God’s own thoughts. Still, this necessarily remains only a ‘picture’. Indeed, we shall see how the existence of an omniscient being would not, after all, ‘make realism true’ (Ch.25).


237. Thoughts are necessarily thought by a subject; and the subject who thinks is necessarily an agent. The thoughts which aim themselves at the ‘that...’ do not exist: The destruction of the ‘that...’ is its disappearance within a metaphysic of action (compare 6/56-7). Does that now serve as a ‘constructive’ refutation of realism or are we still obliged to ‘construct a definition of what it is for a question to ‘have’ an answer’ (22/219, 228)? A complete set of necessary and sufficient conditions might well prove very difficult to construct. Worse, the very idea of such a ‘definition’ is misconceived. For the thought behind that project is this: The realist believed that reality was ‘solid’, whereas we have discovered that it is really ‘only a skeleton’; all that would have constituted the answers to undecidable questions has been ‘cut away’. Having rejected realism, we must now determine just how much of reality is left. But that account could only be motivated by the metaphysic of realism; a residual clinging to the truth illusion. For such an ‘anti-realism’ is merely the incoherent description of what the rejection of realism means to someone who holds the realist ‘position’; realism defines the ‘position’ of its rival (21/205). The alternative is to refuse to treat the concept of ‘having’ an answer as a ‘metaphysical problem’. Having rejected the illusion which makes that concept become problematic, all that remains to say is that there is a time and a place for talk of questions ‘having’ answers; in certain contexts such talk has a point which it lacks in others. Of course, we often talk without any ‘point’.


238. There remains the task of refuting the realist’s ‘global scepticism’ (21/207). In order to raise the sceptic’s seeming ‘question’ whether a given belief is ‘really’ true it is not enough to entertain the ‘supposition’ of its not being true; for that is merely the recognition of the ‘corrigibility of verification’ (21/206). The sceptic’s ‘question’ must be aimed at reality. But the sense in which the sceptic ‘intends’ his ‘question’ — it is, by definition, ‘undecidable’ — rules out that possibility (232-3). As in the case of phenomenalism, what the self-professed ‘global sceptic’ means turns out to be ‘quite correct’; only he ‘fails to think his position through to its logical conclusion’ (9/89); in maintaining that the ‘real’ truth exists but cannot be known he remains under the spell of the truth illusion.


24. What is knowing a proposition’s truth-conditions?


239. The rejection of realism is the rejection of a metaphysical illusion. That realism is to be rejected is a proposition of metaphysics whose truth consists solely in the adequate expression of the truth illusion through its functional role within the dialectic which refutes it (1/10); it is not the description of a ‘metaphysical vision’ of an ‘anti-realist reality’ (21/205, 23/237). In order to perceive that truth the dialectic must indeed be ‘worked through’; for the bare ‘expression’ of realism and its bare ‘rejection’ are equally nonsensical (20/201-3). But the essential core of the dialectic (22/222, 23/232), passes by one’s eyes so quickly that one may fail to see what the ‘refutations’ really add to ‘bare rejection’. One is tempted to protest that there must be more positive things to say about such a momentous discovery; that we should not rest content with oblique expression but explain what it is that makes the ‘refutations’ the expression of a metaphysical truth; not simply assert that ‘the truth consists in the refutation’, And so the meaning theorist will argue: The impossibility of transcendent metaphysics means that we still cannot hope for a direct route to explanation. Nevertheless, the ‘positive’ meaning of the rejection of realism can be approached indirectly, via an ‘analysis of the workings of language’ (13/126-7, 18/172); an account of the ‘form’ of an ‘anti-realist theory of meaning’ (Dummett Frege. Philosophy of Language P.671; for Dummett’s most complete account of meaning-theoretical anti-realism see ‘What is a Theory of Meaning? (II)’). — This is not Dummett’s argument; only a ‘psychological hypothesis’ (18/174). It explains only the conviction that such a theory ‘must be possible’. We still have to describe the ‘overlooked possibility’ (18/175).


240. How does our grasp of the meanings of words provide us with a knowledge of the meanings of propositions in which they occur? — Language represents our knowledge of reality. In order to obtain that knowledge we have to investigate reality, determine the truth or falsity of propositions. But how do we know that the ‘results’ of a particular ‘investigation’ serve to genuinely verify or falsify a given proposition? We do not normally stop to ask; nevertheless, the fact that we follow certain procedures of investigation must in some sense be ‘accountable to’ the meanings of propositions which they aim to verify or falsify. If we could arbitrarily choose whichever ‘procedures of investigation’ we liked then truth would reduce to a mere convention; or, rather, there would be no such thing as ‘truth’. But at the same time there cannot be anything to the meaning of a proposition to which its use cannot be seen as genuinely ‘accountable’. Now our grasp of the meaning of a concept word can at most only be made to yield, through a knowledge of the grounds for its application, a knowledge of ‘procedures’ for investigating the truth of propositions in which it occurs which one could use only in especially favourable circumstances; for a proposition may concern a time other than now, a region of space other than here, may involve a plurality of objects which cannot be directly surveyed. Our grasp of that concept can at most tell us only what an ‘ideally placed or empowered observer’ would be able to ‘observe’ or ‘determine’. The question is: What relevance has that purported ‘knowledge’ for the procedures of investigation we actually use?


241. To assert that knowledge of the meaning of a proposition is ‘knowledge of the conditions under which it would be true’ is just to deny that anything needs to be ‘added’ to our grasp of concepts in order to ‘account for’ the procedures we actually employ to investigate the truth of propositions. Having explained the nature of concepts we have nothing more to say about the meaning of ‘meaning’: To know what an ‘F’ is, to know what ‘G’ means entails knowing, subject to category restrictions determined by the concepts ‘F’ and ‘G’, what it is for an F to be G, knowing the ‘conditions’ under which ‘some F is G’, or ‘All F’s are G’, or ‘some F was G’ would be true. But paradoxically it can seem as if this ‘denial’ asserts too much. The meaning theorist argues: Knowledge of ‘truth conditions’ at most shows, how an ideal observer would determine the truth of a proposition. Our actual procedures of investigation are supposedly all ‘accountable’ to this ‘conception’. But one can ‘know the truth conditions’ of propositions whose truth value one cannot be certain of being able to get in a position to determine. We have the ‘conception’ of truth conditions but do not always know what to do with it; it is then not fully manifested in linguistic practice, remaining just a ‘picture’ in our minds. But it is not as if one were positing an ‘unobservable’ cause in order to explain an ‘observable’ effect (10/93-4). The notion of ‘meaning’ is exhausted by its role in ‘accounting’ for a proposition’s use. We must therefore reject the theory which identifies meaning with that seeming ‘conception’; and that means refusing to allow that ‘stating truth conditions’ can in any form sufficiently ‘account’ for the facts of language use. We have to construct instead an account of the form of an ‘anti.-realist’ theory of meaning; to show how use is ‘accountable’ to a fully manifested knowledge of canonical modes of truth-determination, systematically based upon the manner in which propositions are constructed from their constituent elements. Now, as in the case of a theory of concepts (18/175), we do not yet know what ‘form’ an ‘anti-realist theory of meaning’ will take. However, nothing less would suffice for a knowledge of how language works.


242. Everything turns upon the sense in which use must be seen to be accountable to meaning. We have seen that, in the case of a ‘theory of concepts’, that role is adequately performed by the notion of a concept’s ‘point’; what use is accountable to is not a ‘rule’ which the competent speaker ‘implicitly knows’ (18/180-2). In general, that there exists a ‘point’ or ‘rationale’ behind our linguistic activities is all that accounts for the ‘accountability’ of their unreflective performance. Only in cases of difficulty or dispute do we formulate ‘rules’ on the basis of what we are able to discover by reflection upon that point. But there are limits to such reflection. Explanation of the point may simply ‘come to an end’ short of resolving the question (18/180). That in no way threatens the ‘empirical content’ of the notion of ‘point’; the point is fully ‘manifested’ in use. The correct conclusion to draw from the impossibility of substituting a fully explicable ‘knowledge of rules’ for a knowledge of point which cannot be reduced to something conveyable in words to someone who lacks the ability to see the point is this: There are limits to the potential audience to whom the meaning is ‘fully manifested’. — What now constitutes the ‘rationale’ behind those procedures of investigation which are not so ‘accounted for’ by the point of our concepts, that is to say, where the point at most tells us only how an ideal observer would determine the truth of a proposition, and not how we are to proceed to investigate it? What in general ‘accounts for’ the validity of what we take to be ‘procedures of investigation’?


243. According to a strict sense of ‘accounting for’ nothing less than a ‘criterion of truth’ would satisfy that request; in that case the request is necessarily unclassifiable (21/210). We must lower our sights: What is a ‘valid procedure of investigation’ in sociological or historical inquiry? in physics? in astronomy? in a court of law? Such a universal ‘methodology of methodologies’ might well prove immensely difficult to construct. But perhaps we can confine ourselves to the piecemeal philosophical criticism of the methodologies of particular fields of inquiry; we need not decide the question here. In any case, the bare appeal to an ‘ideal observer’ would accomplish very little; it is absurd to suppose that the ‘truth conditions’ account of meaning could take upon itself such a task. Even if the conception of an ideal observer could in some sense be said to ‘account for’ inferences to other places or times, or concerning unsurveyable generality, that constitutes only a small fraction of the problem of ‘accounting for’ the validity of procedures of investigation, once reformulated in terms of their rationale, rather than the mythical ‘content’ of the investigator’s knowledge of ‘meaning rules’.


244. The ‘facts of language use’ include all the various fields of inquiry in which we use language to represent reality. To acknowledge that an account of the content of the competent speaker’s linguistic knowledge can do no more than ‘state truth conditions’ means acknowledging that whatever sense philosophy can make of these facts should be located elsewhere than in a ‘theory of meaning’. It does not entail setting our conception of the ‘ideal observer’ a task which it could not possibly perform. Of course, the meaning theorist does not suppose that an ‘anti-realist theory of meaning’ would ‘account’ for the validity of any but a small core of ‘verification’ or ‘falsification’ rules, leaving the ‘methodology’ which leads us to the point of performing the crucial verification or falsification of a given proposition to take care of itself. But remember that it is the validity of procedures of investigation which is at issue, and not merely the utility of one valid procedure as compared with another. Having raised the question of what must be added to the meaning rules for concepts in order to ‘account’ for those judgements of truth and falsity which we are in a position to make, one cannot stop half-way. The idea that one can draw a line around the crucial ‘test’ of a proposition, so that the choice of ‘method’ does not affect our judgements of truth and falsity but only the efficiency with which we are able to get in a position to make those judgements, is a fiction. It could only have been motivated by the attempt t? locate within linguistic practice a surrogate for the rejected. conception of an ideal observer; there is no other reason for believing it.


245. How could the notion that knowledge of the meaning of a proposition is knowledge of its truth conditions have ever been identified with realism? There are two questions here. why the realist takes the notion of an undecidable proposition’s having truth conditions as an expression of realism; and why the meaning theorist takes the rejection of a truth-conditions account of meaning as the rejection of realism. To deal with the latter first. We have to consider the following counternecessaryfactual: What would ‘knowing truth conditions’ have been if ‘aiming thoughts at reality’ meant what the realist intended it to ‘mean’? Simply by virtue of my ability to represent to myself the possible state of affairs which a proposition describes, I possess a conception which licenses the assertion that ‘some undecidable propositions are true’, in the sense in which the realist ‘means’ it. Only there is nothing I can do to show that I have this knowledge. Someone else, whose linguistic practice does not differ from mine, may, for all I know, not have the ‘knowledge’ that I have. Then what ‘knowledge of meaning’ does this other person have? not knowledge of truth conditions but rather of the rules for the game of asserting and refuting, verifying and falsifying. Now the rejection of the inexpressible, incorrigible knowledge which I only seem to ‘have’ means that what the other person has I also have. The seeming ‘reality’ which the seeming ‘truth conditions’ of propositions ‘represent’ does not exist. Thought does not aim to ‘represent’ reality but only to be a correct move in the game. We may, however, talk of a ‘reality’ which is not simply equivalent to what we know: When one considers how things stand on the supposition that a proposition is true one does not mean to consider the different supposition that the proposition has been verified. But the only notion of truth licensed by our practice of asserting conditionals is that of the knowably true; only when the antecedent is knowable is there a possibility that the consequent will be detached. To suppose that the antecedent describes an unknowable possibility is a ‘conception’ which cannot be ‘manifested’ in the language game. — Now this ‘reality’ of the knowably true is just the realist’s picture of the ‘anti-realist’ conception of an ‘anti-realist reality’ (239).


246. Our refusal to follow the meaning theorist in rejecting the equivalence between meaning and truth conditions prompts the realist to reply. ‘Reality is what would constitute the answers to all determinate questions, what either satisfies or fails to satisfy the truth conditions of any given proposition, provided only that it has a determinate sense. The possibility of satisfaction does not depend upon our being able to get into a position to determine whether or not truth conditions are satisfied. But if undecidable questions do not have answers, if there can be no answer to the question which undecidable propositions have their truth conditions satisfied then reality itself must be incomplete. And that is absurd. Reality is not like the plot of a story; there can be no details ‘left out’.’ — We can only retort. The thought of how things stand in reality is indeed the thought of something whose details are all ‘completely filled in’. But the question which of all the possible ‘encyclopaedic representations’ of a ‘completely filled in’ reality represents the way things ‘really’ are cannot itself be said to ‘have’ an answer (20/201). Now that blank refusal is not yet the ‘refutation of realism’; the dialectic must first be worked through. The realist must be brought to see the ineffectiveness of his appeal to ‘truth conditions’ in attempting to convey what he ‘means’; that there is nothing to ‘convey’ except the symptoms of his illusion.


25. God and realism


247. Forced into a corner, the realist eventually admits to the most revealing expression of his metaphysical vision: ‘However limited our own powers of knowledge and perception, that truth which we shall never be in a position to know remains the truth because God knows it’ (23/236). — The rejection of this realist vision does not of course entail the opposite principle; that the truth mysteriously ‘ceases to be the truth’ when we are unable to get in a position to know it. That would amount to an empirical hypothesis which could indeed only be understood by invoking a deity’ that ‘reality’ is not what we take it to be but only the ‘back-drop’ of a cosmic film set, with just enough detail to match our capacity for knowledge. In order to prevent our ever finding out ‘the truth’, and to keep the ‘detail’ down to the necessary minimum, there would have to be someone manipulating the backdrop ‘behind the scenes’; a ‘benevolent deceiver’. The ‘empirical content’ of the hypothesis is that God might one day slip up. But this is not meant as a coherent description: The supposition that the truth might have ‘ceased to be the truth’ is the antecedent of a counternecessaryfactual.


248. The thought that what is really true is what is known by God to be true, however gratuitous and indefensible it may appear to the opponent of transcendent metaphysics, should not be viewed simply as the last resort of a crumbling ‘theory’, nor, alternatively, only a metaphysician’s idle conceit. It is a fundamental religious principle (‘I am the truth’); and also a common blasphemy. We regularly appeal to ‘what God, at least, knows’ in order to back up our claims to veracity; or, to underline the universal ignorance on a particular topic, to ‘what God only knows’. And we do this in the same unselfconscious and unreflective way that a child might appeal to the authority of its parents. (Only, paradoxically, our appeal remains unanswered, save in the rare and questionable circumstances of a ‘religious revelation’.) One may draw two ‘immediate conclusions from this apparently familiar association of realism with the deity: The association constitutes strong circumstantial evidence for an identification of one source of the religious impulse with the source of the truth illusion. Furthermore: The very familiarity of the association may well have blinded us to its fundamental illogic.


249. Whatever realism may finally turn out to be, it cannot be identified with that same-named ‘position’ in the ‘dispute’ concerning whether knowledge of the meaning of a proposition may be represented as knowledge of its ‘truth conditions’. We have seen how the meaning theorist’s project of ‘explaining’ what it is to grasp the meaning of a proposition appears only when one is first persuaded that there is such a thing as a theory of meaning for concepts: The mythical schema of the ‘rules for use’ of concept words would leave undetermined their ‘semantic role’ in those propositions whose truth value we are unable to get into a position to decide ‘directly’, in accordance with the ‘canonical modes of truth-determination’ associated with the ‘rules’ for their constituent concepts (24/240). ‘Realism’ is then understood as denying that more ‘rules for use’ are needed to ‘explain’ how propositions in general derive their meaning from the meanings of their constituent concepts: A ‘conception’ of how the ‘rules for use’ of concepts would be directly applied by hypothetical beings more fortunately situated or with greater perceptual or intellectual powers than ourselves suffices to cover the so far ‘undetermined’ cases where our own powers give out (24/241). ‘Anti- realist’ opponents of this ‘position’ are meant to reject the theoretical employment of that ‘conception’ on the grounds of its lack of empirical content when applied to cases where the truth of a proposition cannot be decided by us at all, thereby committing themselves to search for ‘rules of verification’ or ‘rules of falsification’ for propositions which will combine with the ‘rules’ for concepts to ‘determine in an empirically satisfactory way’ the ‘meaning’ of any proposition, irrespective of whether we are able to decide its, truth value (24/241). There appears room here for detailed argument over which kinds of ‘hypothetical beings’ may be invoked by the ‘realist’ (to be taken seriously only if one is a ‘realist’); in particular, over the legitimacy of the ‘realist’s’ use of a ‘conception’ of how God would understand our language. However, the ‘anti-realist’ is far from being entitled to reject the theoretical employment of that conception with a clear conscience. For the ideal of a ‘theory of meaning’ which explains ‘all the rules of the language game’, whose form the meaning theorist aims to comprehend, reveals itself ultimately as a theory whose actual construction only God could complete; a ‘conception’ of God’s understanding of our understanding of language’.


250. The envisaged ‘argument’ over the legitimacy of the realist’s appeal to the conception of ‘what a deity would understand’ proves futile. Resistance to the anti-realist project of constructing a ‘theory of propositions’ does not involve the appeal to hypothetical beings of any sort; instead, it refuses to take the anti-realist’s first step of admitting the coherence of the ideal of a ‘schema of the rules for the use of concepts’ (24/242-4). But this ‘resistance to anti-realism’ is not realism, not a manifestation of the truth illusion. It indeed amounts to a ‘defence of a truth-conditions account of meaning’. Refusal to take even the ‘first step’ is a fortiori a refusal to take the second ‘step’ of attempting to construct an ‘alternative theory’. However, the representation of the semantic articulation of propositions by means of a recursive definition of their truth conditions (15/141) is not an ‘expression’ of what the realist ‘means’ (24/246). We thus rule out one way of connecting realism with the appeal to ‘what only God can know’; if there is a path towards the philosophical demonstration of the existence of God it is not through meaning theory.


251. In a dialectic of illusion, the grounds for rejecting realism reduce to this: It is impossible to express what realism means to ‘mean’ (22/222-8, 23/232-5): not that some position successfully identified as ‘realism’ entails ‘consequences’ which are false; nor that it is impossible to ‘justify’ realism; nor that some other ‘position’ is ‘preferable’ on some other grounds. Argument over the ‘onus of proof’ or over the ‘weighing of claims’ does not even get the chance to get started. We do not allow realism the opportunity to enter the arena before it has identified itself; which it cannot do. Thus, when the realist ‘invokes the conception of a deity’ it is not as the ‘ground’ for, or ‘presupposition’ or ‘consequence’ of realism. It is an attempt to give some substance to the assertion that, when a question does not ‘have’ an answer in the non-metaphysical sense, when we do not know the answer, or know how to find it out, or that there is a genuine possibility of an answer being found out, or when the answer does not in any sense ‘exist’ (perhaps mislaid, perhaps known by some unknown person)(NB 23/237), when it is none of these things the question nevertheless ‘has’ an answer in the metaphysical sense: no matter what the question, provided only that it is a meaningful question, its answer is there in reality (20/200). It is this ‘metaphysical sense’ which we refuse to ‘understand’. It is not as if only when a question ‘has an answer’ in the non-metaphysical sense do we allow that it ‘has an answer’ in the metaphysical sense; as if, e.g., the fact that some hermit has discovered a proof of Fermat’s Last Theorem but is keeping it to himself entails that the proof is ‘there in reality’, and otherwise ‘wouldn’t have been there’, in any sense other than simply that ‘some hermit has discovered ‘. We reject the ‘metaphysical sense’ of ‘having an answer’ altogether (22/228). Now, instead of saying merely that the answer to every question ‘must be there in reality’, the realist may be tempted to embellish his assertion: to assert that ‘reality gives the question its answer’, or that ‘reality knows the answer’. But to ‘give’ or to ‘know’ are things only a person can do. The concluding step could hardly be called an ‘inference’: Reality is a person.


252. Let us call the person which reality is ‘God’, leaving aside the question of its possession of God’s moral attributes (that might indeed be required for a ‘moral realism’). We have seen how the realist’s ‘model’ ‘of the thought as having a ‘life of its own’ finds expression in the idea that all thoughts are ‘essentially God’s thoughts’ (23/236): When anyone thinks, it is God thinking. But God also completely encompasses the ‘reality’ at which God’s thoughts aim. When God’s knowledge is subtracted from the nature of reality as it is in itself there is no remainder. This picture should remind us of the naive idealist’s conception of ‘reality’ as exhausted by ‘my own mental contents’ (3/28). But the comparison reveals no mere ‘analogy’: As pictures, the ‘reality’ of idealism and the ‘reality’ of God’s self-knowledge are indistinguishable. Just as the pure idealist was obliged to forswear the use of concepts, so that all that can be said about the idealist’s ‘reality’ is that it is ‘this...’ (3/31), so also in describing’ the ‘content’ of God’s all-pervasive knowledge’ we just ‘forswear the use of concepts’; for the deity, appearance and reality fuse together in a ‘this...’ As finite beings, unable to conceive of the nature of God’s self-knowledge, His ‘this...’ for us becomes ‘that...’ (21/209).


253. The question before us is not whether God exists or what it means to assert that God exists. Let us assume that God exists and we know what that means; or, at least, ‘comprehend its incomprehensibility’. Our question is simply whether the assertion that God knows’ the truth or ‘gives’ questions their answers succeeds in expressing what the realist means to ‘mean’; that the answer to undecidable questions ‘must be there in reality’. Let ‘P?’ stand for a question which no-one will ever be able to answer. Now God exists; God knows whether P. But so what? what is that to us? We have merely discovered another question which no-one will ever be able to answer: ‘Is it the case that God knows that P?’. Now if God knows that P, then P; and if, instead, God knows that not-P, then not-Po One may nevertheless assert, by way of a ‘first contribution to the dialectic’ (20/201, 24/246), that ‘the question whether God knows that P, rather than not-P, does not have an answer’. ‘But surely God knows the answer to whether God knows that P! — Certainly. If God knows that God knows that P, then God knows that P; and if, instead, God knows that God knows that not-P, then God knows that not-Po say if you like that ‘the question whether P must have an answer’ simply means that ‘God knows the answer’. That sense of ‘having an answer’ does not by itself suffice to yield a point to the ‘conjecture that P’ where previously it had none I to show that there is something which distinguishes the intention to say something true by means of ‘P’ from merely considering what it is that ‘P’ says (23/232-3). And that is the essence of what the realist is attempting to ‘mean’: The conjecture ‘has a point’ because the answer is ‘there in reality’. To give the conjecture a point we must suppose in addition that God not only knows the answer but hears the conjecture and judges its truth value. But if being’ judged’ means anything here then we must be able to look forward to receiving a token of judgement. Yet that very possibility is closed off by the supposition that ‘no-one will ever be able to answer the question’: In receiving the token of God’s judgement we finally get to know the answer!


254. The thought of making unwarranted conjectures simply for the sake of being awarded points by God is a comic fantasy. If someone seriously gave that as an explanation of what he was doing we should call him insane. Still, we should understand the point of his activity; that is to say, we should not only perceive it as an intelligible sequence of bodily movements recognizable as ‘actions’ but also grasp what the activity was meant to accomplish. And if, instead, a person took some attenuated sense of ‘comfort’ in the thought that there is a God who is able to determine the answers to questions which we shall never be able to answer, then that is intelligible also. But the realist in us wants God to do more than either of these things: We want God to make realism true. That ‘desire’ is unintelligible.


26. The law of excluded middle


255. In the eyes of the realist, the assertion: ‘P or not-P’ presents an impression of meaning what realism ‘means’ (22/224). When the proposition ‘P’ is undecidable, his assertion of the law of excluded middle is intended to express what it is for the question whether P to ‘have’ an answer (20/200). Now in claiming that ‘P or not-P’ does not mean what the realist intends it to ‘mean’ but rather ‘something else’ we appear to be caught in a vicious trap: If we cannot express what it is that the law does not mean, then saying: ‘This is what it does mean...’ is not giving its supposedly ‘different’ meaning; for there is no ‘realist meaning’ from which it differs, which this ‘different’ meaning excludes as a possible meaning (22/220). In that case the realist could translate our account of the meaning into a form which does not threaten realism. By virtue of its meaning just what we assert it to mean, the law ‘acquires’ its additional, inexpressible ‘realist meaning’. The only way out of this trap is to give an account of the meaning of ‘P or not-P’ which essentially embodies components from the dialectic of realism; so the account could not be understood if the dialectic itself was-not grasped and the correct moral drawn. This is not an ‘account of the meaning’ in the ordinary sense, but rather a contribution to the dialectic.


256. We should first note that ‘P or not-P’ is a logical truth only when ‘not-P’ is taken as the fully external negation of ‘P’ (expressed in terms of the customary sense of ‘not ‘: ‘Either not-P or ‘P’ fails to express a thought’). The apparent power of the law of excluded middle to ‘prove’ substantial conclusions in the absence of additional, empirical premisses arises from the use of negation in the ‘customary sense’, a negation which is not ‘fully external’. For example, from the premise: ‘Deidre either was or was not a brave witch’ one can prove that Deidre existed, that there is such a thing as being a witch, and that, in the absence of any occasion which would have tested Deidre’s bravery, there is still something that the ‘unactualized disposition towards bravery’ consisted in, a determinate ‘state’ of, or ‘fact’ about something, perhaps Deidre’s brain, embodying the potential to exercise the disposition. The disjunction of ‘Deidre was a brave witch’ with its fully external negation does not entail any of these things. Aside from mathematics (Ch.27), the only use for arguments which purport to derive substantial conclusions from the law of excluded middle by means of ‘or-elimination’ is in sophistry. It is an attempt to disguise the fact that the sense of ‘P or not-P’ which entails these consequences is not a logical truth but rather a powerful device for extracting the presuppositions of a proposition’s having a sense, or of a state of affairs’ being regarded as so constituted that the question whether a particular concept applies may be meaningfully raised.


257. ‘A proposition is either true or its negation is true; one or other has got to be true, cannot fail to be true.’ — That sounds like. If I shoot in every direction I cannot fail to hit the target. But truth is not that kind of ‘target’; the intention to ‘aim a conjecture at reality’ simply in the ‘hope’ of ‘hitting the truth’ is undermined by the recognition of that proposition’s undecidability (23/232-4). That means. The application of the law of excluded middle to an undecidable proposition is not ‘made true’ by the metaphysical fact that the question whether that proposition is true ‘has’ an answer (22/224), for the realist’s ‘metaphysical fact’ is only an illusion. What then makes such instantiations of the principle true? At the risk of ‘saying something senseless or paradoxical’, we can only say that absolutely nothing ‘makes’ them true. They are simply, barely true.


258. The first step seems obvious: We must explain the meaning of ‘or’: ‘P or Q’ is true if and only if either ‘P’ is true or ‘Q’ is true. That is a ‘circular’ explanation in the sense that it would not communicate the meaning of ‘or’ to someone who did not already know it. But it is not trivial; for it rejects the idea that any more is needed for the truth of ‘P or Q’ than simply the truth of ‘P’ or the truth of ‘Q’. Accepting the fact that one cannot informatively state the meaning of ‘or’, we still wish to go beyond the ‘circular account’, for that clearly does not tell us what we did not know already. Suppose one asks instead: ‘What determines whether any given item of evidence is to count as verifying an empirical proposition of the form. ‘P or Q’?’ — The question serves merely as a roundabout means of once again asking for the meaning of ‘or’. The assertion of ‘P or Q’ commits one to just these possibilities: P, Q, P and Q; in order to justify that assertion the evidential ground must rule out the possibility that neither ‘P’ nor ‘Q’ is true. That is no less ‘circular’ than the previous account; nevertheless, it adds a new element, for it introduces the fundamental logical concept of a possibility in parallel with the concept of negation, the rejection of a possibility. It might seem that there is a further question one could ask. ‘What kind of state of affairs does ‘P or Q’ represent?’ — What this question has in mind is something indeterminately ‘half-way’ between the possibility represented by ‘P’ and that represented by ‘Q’; and that is an absurdity. ‘P or Q’ represents only a range of possibilities; and that yields another ‘fundamental logical concept’. (In quantum mechanics, the momentum of a particle is described as being ‘indeterminately a or b or c or... ish’; but then ‘a’, ‘b’, ‘c’ are not themselves regarded as representing individual ‘possibilities’.) ‘But if there is no single state of affairs represented by ‘P or Q’, what is it that connects up with the evidential ground for that proposition and makes it evidence for that very proposition?’ — The material ground of the evidence must indeed be conceived as a non-disjunctive ‘state of affairs’; the evidential ‘trace’ is of such a kind as could only have been caused directly or indirectly either by the state of affairs represented by ‘P’ or that represented by ‘Q’. That is how the evidence is able to justify the assertion: ‘P or Q’.


259. The meaning of ‘or’ is purely logical; I do not assert the existence of any kind of ‘causal connection’ between ‘P’ and ‘Q’ when I assert ‘P or Q’. (That is the point of the original, ‘circular’ account.) But the question why there is a point in having such a ‘logical notion’ in the language is far from being a question of simple logic; it is rather a question of transcendental logic (Critique of Pure Reason B92-116, known as the ‘metaphysical deduction of the categories’; what follows departs completely from the substance of Kant’s account, and reflects only that part of its spirit which survives the rejection of phenomenalism). The utility of deductive inference, the use we have for the logical notions, reflects our ‘cognitive predicament’. The questions which we put to empirical reality all have the general form: ‘How do things stand in the light of this evidence?’ As the result of our investigations, certain possibilities are excluded while others remain open. (From the point of view of transcendental logic it is irrelevant that we do not in practice take into account all the possibilities.) Questions become increasingly more specific as more possibilities are excluded. Beings for whom, by contrast, the acquisition of knowledge did not involve ‘narrowing down’ a ‘range of possibilities’, who obtained all their knowledge simply by asking non-compound propositional questions and putting the answers ‘in the archives’ would have no essential use for logic (a counternecessaryfactual). For example, the assertion: ‘P or Q’ would always be grounded upon either knowledge that P or knowledge that Q; there would be no opportunity for deducing that P from the subsequent rejection of the possibility that Q. The same applies to the ‘range of possibilities’ represented by propositions of the form: ‘some F’s are G’; one would always know which F’s were G. Now the use of ‘not’ to reject possibilities is co-ordinate with the use of ‘or’ and ‘some’; and together, ‘or’, ‘not’ and ‘some’ serve to define all the Classical logical notions. (A parallel explanation could have been given for ‘if then ‘, ‘not’, ‘all’; but it would have been less perspicuous from the point of view of transcendental logic.)


260. A propositional sign represents the possibility or range of possibilities of its being true, or the rejection of a possibility or range of possibilities; its truth conditions. Having truth conditions is what qualifies a form of expression for potential use in the ‘knowledge game’ (as indeed in all the other language games where linguistic acts have a propositional content). We may think of propositions as counters and their assertion or conjecture as the introduction of these counters into the game in progress. Taking hold of a counter is ‘considering what the proposition says’; while introducing it into the knowledge game is ‘intending to say something true’ (23/232-3, 25/253). We represent reality by the use of propositions in the knowledge game; the proposition itself ‘represents’ only its truth conditions (23/235). When a proposition ‘P’ occurs within a compound ‘P... Q’, its role in the knowledge game is merely its contribution to the truth conditions of ‘P... Q’; when one asserts ‘P... Q’ one does not intend to ‘say something true’ by means of ‘P’ but only by means of ‘P... Q’. Now the question: ‘Is this possibility realized?’ only makes sense within the knowledge game; the counter must be taken out of the box and put on the board. However, some counters, the propositions which we discover to be undecidable, have to be withdrawn from the game; they remain in the box, their power to represent reality reduced to the empty ‘representation of truth conditions’. Thus, a disjunctive proposition ‘P or Q’ may have a use in the knowledge game, say, because we have evidence which decides its truth, while its undecidable disjuncts ‘P’, ‘Q’ are themselves condemned to remain in the box. A disjunction may be true despite the fact that there is no ‘answer’ to the question which of its disjuncts is true.


261. Evidence narrows down a range of possibilities. The assertion of a disjunction rules out the possibilities that its evidential grounds ‘rule out’ (258). But what about the disjunction: ‘P or not-P’? There is no ‘third possibility’ to be ruled out and consequently no need of grounds for ‘ruling’ anything out. Now we have discovered two senses in which there may be said to ‘exist’ something to ‘make’ a disjunctive proposition true: that the question associated with at least one disjunct has an affirmative answer; or that there exist grounds for the assertion of the disjunction itself, but possibly ‘no answer to the question which disjunct is true’. But the disjunction of an undecidable proposition with its negation is not ‘made true’ in either sense; it is simply, barely true’ (257).


262. We have seen how, if an or-elimination argument from a non-mathematical premise: ‘P or not-P’ is used in order to derive a ‘substantial conclusion’, that premise is not the logical ‘law of excluded middle’ (256). But has the law any legitimate use in extra-mathematical deductive practice? When one talks of the ‘logic’ of a given argument one means the interpretation of that piece of discourse by means of a system of formal logic; an interpretation which shows the argument to be valid solely by virtue of its form. That entails reorganizing what is given’, supplying the missing ‘steps’ etc. Now whenever it becomes necessary in an argument to underline the fact that there are ‘no other possibilities’ besides its being the case that P and its being the case that not-P, to reject any premise which implies that neither a proposition nor its (external) negation are true, the form of the argument may be adequately interpreted by means of Intuitionist logic, as depending only upon the logical principle: ‘not-not- (P or not-P). In Intuitionist logic, that principle is weaker than ‘P or not-P’, the latter not being a logical truth. Thus, nothing would count as showing that the ‘law of excluded middle’ of Classical logic is ever used in an informal argument. The fact that a person leaves out the double-negation from the expression of the Intuitionist principle proves nothing, for in our informal deductive practice many things are ‘left out’. The assertion that ‘nothing would count as showing... should not of course be understood in a sceptical sense; for it is not as if the difference between the use of Classical and Intuitionist logic could be a matter of a hidden, private ‘intention’. Classical logic simply does not have application to extra-mathematical practice.


263. The sense of the proposition. ‘P or not-P’ in which it is simply, barely true’ (261) is not what the realist means. In his eyes, ‘P or not-P’ could only be true by being ‘made’ true by the ‘existence of an answer to the question whether P’. Now if, per impossibile, ‘P or not-P’ could be used to mean what the realist intended it to ‘mean’, then the difference between the use of Classical and Intuitionist logic would be a matter of ‘hidden, private intention’ (262), In the absence of any other coherent ‘expression’ of realism, the ‘interpretation’ of an informal argument by means of a system of formal logic would then become a necessarily unverifiable hypothesis.


264. Let us suppose that the realist argues as follows: ‘you agree that ‘P’ is either true or not true. If ‘P’ is true then the question whether P has an answer. If ‘P’ is not true then the question whether P again has an answer. Therefore, in either case, the question whether P has an answer.’ — The inference indeed licenses the assertion: ‘The undecidable question whether P has an answer’. But that does not suffice to ‘convey’ what the realist ‘means’. For we may reply (in the spirit of 22/224): ‘There is no answer to the question which answer the question whether P has.’ If the realist now uses the same form of argument to prove. ‘There is an answer to the question which answer the question whether P has’, we reply: ‘There is no answer to the question which answer the question. ‘Which answer does the question whether P have?’ has’. (Compare the regress generated by the realist’s: ‘God knows the answer to the question whether P; 25/253.) But this potentially infinite series of ‘denials’ is not yet the ‘refutation of realism’ (24/246); the dialectic must once again be ‘worked through’.


27. Intuitionist logic and mathematics


265. In a dialectic of illusion there are no metaphysical positions: To be an idealist or a realist consists only in the disposition to manifest the symptoms of a metaphysical illusion; and to be deceived by the ego or truth illusions is to be ‘deceived’ into ‘believing’ nothing. — To the meaning theorist, that conclusion appears intolerably paradoxical: The content of our true metaphysical vision cannot be reduced to the rejection of a false metaphysical vision. For if the seeming ‘content’ of that false vision is in reality ‘nothing’, then the ‘content’ of the true vision cannot even be described as ‘negative’, since it does not negate anything. This argument is indeed not fallacious but rather a restatement of the fundamental problem of’ a dialectic of illusion; the demand for a justification of its concept of ‘adequate expression’ (1/9, 10). However, the meaning theorist sees only one possibility: The ‘content’ of a metaphysical vision can only be realized in its account of the form of a theory of meaning; the ‘truth’ or ‘falsity’ of the vision is the correctness or incorrectness of that account. But that raises the question: What was the content of the metaphysical vision before it found its meaning-theoretical ‘realization’? It was still ‘something’ rather than ‘nothing’. Now if the real objection to a dialectic of illusion is not merely that it fails to engage in the necessary constructive work (24/239) but rather that its concept of ‘adequate expression’ is incoherent in itself, then the pre-meaning-theoretical vision also demands its ‘realization’. To the meaning theorist, this appears as the logic accepted in linguistic practice: A theory of meaning justifies that linguistic practice whose logic reflects its own metaphysical vision. It simultaneously rejects any practice which reflects what, according to that vision, is a false vision; a practice which requires for its ‘justification’ an incorrect account of the form of a theory of meaning.


266. To ask whether the practice of deducing propositions by means of logic is justified carries the breathtaking, Promethean, insane and terrifying resonances of a revisionary metaphysics expanded to the limits of its ambitions. Like the question: ‘Is time real, or only an illusion?’ (28/283, 30/304) it attacks the very foundation of what we take to be ‘reality’, threatening to destroy the meaning of all thought and all knowledge. This ‘scepticism’ appears far deeper and more devastating than the ‘epistemological’ kind. One imagines that one could live with the discovery that we do not really know anything; but not with the realization that, since logic is itself illogical, there is no difference between a ‘valid’ or an ‘invalid’ argument; or that, since time itself is unreal, there is no such thing as witnessing an ‘event’ or performing an ‘action’. — Now it turns out that the meaning theorist does not intend his question mark against the validity of logic to be taken in this ‘sceptical’ sense. His notion of ‘justification’ carries only the more restricted sense of ‘determining whether all the logical inferences which we sanction are faithful to the meanings of the logical notions’. But that presupposes that a dialectic has already been worked through: The metaphysical illusion whose partial perception gives rise to the ‘sceptical’ question against logic (compare 9/89, 23/238) must first be ‘adequately expressed’ and ‘refuted’. (It is the illusion which Wittgenstein diagnosed as the super-physical ‘hardness’ of the logical ‘must’; Remarks on the Foundations of Mathematics 1/113-42.) Only then could the project of ‘justifying’ logic in the more restricted sense have any genuine interest. In that case, the justifying ‘theory of meaning’ could not claim to have supplanted the earlier dialectic but only (at most) to supplement it.


267. The meaning theorist’s project of ‘justifying logic’ would not have been undertaken were it not for the discovery of two remarkable facts. The first is this: One would have thought that logic must be everywhere the same; there is only one logic, besides which there remains only illogic. But in mathematics we discover two competing ‘logics’: The so-called ‘Intuitionist’ mathematicians refuse to accept that the truth of a proposition follows from a proof of the absurdity of its denial. If we look at their practice we find a different kind of ‘logic’ from the ‘Classical’ logic which once held the field unchallenged. Intuitionist logic does not include amongst its laws the ‘law of excluded middle’ of Classical logic. Yet the ‘logic’ is indeed logical and it adequately describes the Intuitionist mathematicians’ deductive practice!


268. The second remarkable fact is the manifest circularity in the standard explanation of the meanings of the logical notions (26/258). Of course, it is absurd to suppose that there could be such a thing as genuinely explaining their meaning, that a person might ‘not know’ their meanings and need to have them explained to him. But what is remarkable is that, lacking such an ‘explanation’, there remains nothing else to back up our faith in the correctness of our notions of what counts as a ‘logically valid’ inference. At this point one teeters on the edge of the ‘sceptical’ dialectic. But that is not now the issue. Rather, the possibility has been raised that Classical logic is not the right logic; that we should change our notions of what counts as a ‘logically valid’ inference. The possibility is that, far from being only a description of the inferences which should be sanctioned in mathematical practice, Intuitionist logic is intended as a revision of our notions of what counts as a logically valid inference; that the forms of inference rejected as mathematically ‘invalid’ are ‘invalid’ tout court. In that case, Classical and Intuitionist logicians will ‘understand’ the logical notions, and consequently the ‘circular account of the logical notions’, differently. This ‘difference in understanding’ is what ‘explains’ the difference between what the two parties regard as a ‘logically valid’ inference. But the ‘difference in understanding’ remains unexplained.


269. In mathematics, proofs essentially relying on Classical logic are called ‘non-constructive’. Intuitionist mathematicians reject the very coherence of such ‘proof’; its use embodies the belief in a false metaphysic. For example, a proof of Goldbach’s conjecture that ‘every even number is the sum of two primes’ would consist in a ‘mathematical induction’; a demonstration that the conjecture applies to the number 0 and that if it applies to any number n then it applies to n+1. By contrast, a proof that its negation entailed a contradiction would not constitute a proof of the conjecture, but only of its double-negation. Only the application of the Classical principle of double-negation elimination, or equivalently, the law of excluded middle, would permit one to conclude that Goldbach’s conjecture was true. Our first task is to determine how the claims of Intuitionist mathematicians should be viewed both from the point of view of the dialectic of realism and from the point of view of a conception of metaphysics as the dialectic of illusion.


270. Intuitionist mathematicians reject a certain conception of the reality which makes mathematical propositions true. And they take that as sufficient to justify their rejection of non-constructive proof, their refusal to assert the law of excluded middle. However, we have seen how the rejection of the illusion of realism does not entail the refusal to assert the law of excluded middle (26/261, 4). Rather, we refuse to allow that asserting that law means what the realist intends it to ‘mean’. Now if mathematics were confined to bounded generalization (‘for all x less than n...’, ‘for some x less than n...’) there would nevertheless exist propositions whose proof was beyond our physical capacities. What the realist ‘means’ by asserting that their associated questions ‘have’ answers would be no less problematic than in the case where the generalizations are unbounded. But the Intuitionist objection to non-constructive proof would no longer apply. The crucial difference between bounded and unbounded generalization is that only the latter concerns infinite domains; the dispute between Intuitionist and Classical mathematicians is not over the validity of realism but over the concept of infinity.


271. The series of natural numbers, or the series of ‘constructions’ yielded by an inductive proof, can be potentially carried on ad infinitum. However, we find ourselves tempted to raise questions which concern ‘actual’ and not merely ‘potential’ infinity I ‘Never mind the question whether there exists an inductive proof of Goldbach’s conjecture, does mathematical reality make it true or not?’ A non-constructive proof would then be taken to establish a truth about the actual infinite; that means we should not regard it as necessarily a ground for belief that the conjecture is inductively provable. Where does this conception of actual infinity come from? If the idea of a finite but unsurveyable totality were unthinkable then we should not even be able to comprehend the meaning of those bounded generalizations whose proof is beyond our physical capacities. To say that such a totality would be surveyable by a being with greater sensory and/or intellectual powers than our own would not supply an adequate ‘explanation’ were comprehension found lacking. Rather, it serves as a reminder that even to credit ourselves with the conception of an ‘ability to survey’ entails thinking both sides of the ability; the ability which we have and the ability which, owing to certain physical limitations, we lack. By contrast, the ‘actual infinite’ is a new concept, not implied by an adequate description of the limits of our cognitive powers. (Calling our powers ‘finite’ imports the concept only of potential infinity, ) To say that an infinite totality is ‘surveyable only by a being with infinite powers’ can only be taken as a joke; that would be to ‘explain’ the possibility of ‘actual’ infinity in terms of the possibility of an ‘actually infinite being’. If that is the picture which lies behind talk of ‘actual’ infinity then the idea that non-constructive proof might be the only route to a supposedly ‘given’ mathematical truth is indeed incoherent.


272. If one rejects the notion of actual infinity then the truth conditions of mathematical propositions imply the existence of constructive proofs; mathematical propositions are aimed at potential infinity. When the law of excluded middle is used in a mathematical proof either it is being assumed that a constructive proof ‘must exist’ or the non-constructive proof is being taken to establish a truth about the actual infinite. If the former is the case then the proof rests upon an unproved assumption; if the latter then to ‘understand’ what it is taken to ‘prove’ involves belief in a false metaphysic. And there rests the Intuitionist’s case: ‘Either prove that this mathematical question can be decided by constructive proof or explain what you mean by ‘actual’ infinity.’ If the Intuitionists are right, some revision of mathematical practice is inevitable; for it turns out that certain propositions of Classical mathematics can be disproved in a mathematics based upon Intuitionist principles. So one could not justify the practice of Classical mathematics as merely an ‘anticipation’ of the ‘possibility’ of finding Intuitionistically valid proofs.


273. The Intuitionist account of the truth conditions of mathematical propositions should not be understood as a reduction of truth to a mathematically defined notion of ‘constructive provability’; just because our concept of an Intuitionistically acceptable proof remains open-ended, the truth conditions for each proposition must remain truistic. The Intuitionist revision of mathematical practice is rather the revision of our conception of mathematical reality itself; the rejection of a certain metaphysical illusion. However, the use of the notion of the ‘existence’ of constructive proofs suggests an objection: If the Classical mathematician is wrong to regard the non-constructive proof of a proposition as entailing the ‘existence’ of a constructive proof, then in what sense can it be all right for our understanding of the truth conditions of an unproved mathematical proposition for which we do not possess a decision procedure to ‘entail the existence of a constructive proof’; which it does simply by virtue of the fact that the truth conditions of mathematical propositions are understood to relate to a reality where all truths are ‘constructively provable truths’? — The Intuitionist is committed only to the following pair of conditionals: If the proposition is true then it is constructively provable; and if, instead, the proposition is false then it is constructively disprovable. But that is consistent with the refusal to assert that the proposition is ‘either true or false’; either constructively provable or constructively disprovable. Still, there remains a problem with the Intuitionist’s charge against the Classical mathematician that, foregoing the concept of ‘actual’ infinity, to take a non-constructive proof as establishing the existence of a constructive proof is an unjustified assumption. For it is not as if we are waiting either for a proof that all meaningful mathematical propositions are either constructively provable or constructively disprovable, or for a proof that some propositions are neither. The latter is of course ruled out since it would entail, via the Intuitionist’s ‘pair of conditionals’, that some propositions are neither true nor false; which contradicts the Intuitionist principle of ‘not-not- (P or not-P), But the absence of the former proof does not itself suffice to justify the placement of onus implied by talk of an ‘assumption’. The Classical mathematician may reply that the very idea that non-constructive proof might not, for all we know, capture constructive provability is itself an ‘unjustified assumption’.


274. The foregoing account of the grounds for Intuitionism in mathematics was based on the premise, already established by the dialectic of realism, that the rejection of realism does not entail the refusal to assert the law of excluded middle (270). The grounds for the rejection of Classical mathematics are not grounds for the rejection of Classical logic, But then nothing much turns on this question; for we have already seen how non-mathematical practice lacks the features necessary to determine that Classical, rather than Intuitionist logic is being used (26/262). The ‘practice’ which the meaning theorist may take to ‘realize’ the false pre-meaning-theoretical metaphysic rejected by a correct account of the form of a theory of meaning (265) Can only be mathematical practice: A global ‘anti-realist’ theory of meaning rejects the use of Classical logic in mathematics; and it does this by giving a revisionary ‘explanation of the meanings of the logical notions’ according to which ‘P or not-P’ is no longer a logical law. Briefly, knowledge of the meaning of a proposition is knowledge of ‘canonical methods’ for its verification (24/241). To verify that ‘the supposition that ‘P’ has been verified is absurd’ is itself absurd is not to verify that P. That is a truism; only when the meanings of the logical notions have been ‘revised’ by an anti-realist theory of meaning does it have the substantial consequence of rejecting the validity of double-negation elimination, and equivalently, the law of excluded middle. Now in the dispute over the philosophical justification for Intuitionist mathematics this account is a non-starter. The demand for an ‘anti-realist theory of meaning’ has already been rejected (24/242-4). We have seen, moreover, that its ‘overlooked possibility’, the dialectic of the truth illusion, does not resolve the question of the justification for Intuitionism (270). Our own reconstruction of the grounds for Intuitionism proved less than conclusive (273): Have we ‘overlooked a possibility’?


275. In ‘The Philosophical Basis of Intuitionist Logic’ Dummett envisages the ‘hard-headed’ denial of the reality of mathematical objects as the only alternative ‘basis’ to the global application of an ‘anti-realist theory of meaning’. The ‘hard-headedness’ of this option consists in a construal of unbounded quantification which justifies Intuitionism only at the cost of maintaining that a proposition cannot be said to be possibly ‘true now’ while it has not actually been proved (P.246-7). This restriction is necessary because any less severe ‘denial of the reality of mathematical objects’ appears compatible with the belief in the ‘actual infinite’; truths concerning ‘actual’ infinity might nevertheless be regarded as truths about mere ‘creations of the mind’, as Dedekind believed (P.246). The ‘hard-headed’ option may be put in this way: We have no difficulty in forming a conception of truths concerning an actual infinity of physical objects; i.e. Unbounded empirical generalizations over an infinite domain which do not possess the status of physical laws. But whereas physical objects ‘already exist’, mathematical objects are only ‘creations of the mind’. An actual infinity of mathematical objects could be created only by a mind of infinite capacity. Therefore, mathematical propositions which we are able to assert can only concern potential infinity; either because we do not know whether such an infinite mind exists or because even if it did its thoughts would not constitute a valid subject matter for our mathematical assertions. Needless to say, the Intuitionist who rejects the concept of actual infinity has no sympathy for this argument.


276. If Intuitionism is not based on the hard-headed denial of the reality of mathematical objects then its only possible justification consists in a possibility which Dummett has overlooked: That the existence of inductive or otherwise ‘constructive’ proofs in mathematics, as indeed the existence of physical laws in the case of empirical generalizations, exhausts our conception of the ‘reality’ which satisfies or fails to satisfy the truth conditions of propositions about the infinite. But we have also seen that the Intuitionist’s case is less than conclusive; that there remains the possibility of rejecting the metaphysical illusion of an actual infinity of mathematical objects while not taking that as sufficient to justify the rejection of Classical mathematics. It is no part of the concept of a ‘dialectic of illusion’ to assert a dogmatic ‘non-revisionism’; the question cannot be decided without further debate.


28. The reality of time; and of other times


277. The sense of shock caused by the thought of what it is that the dialectic of illusion rejects gradually dissipates as one loses oneself in ‘working out the details’. As the new metaphysic appears increasingly familiar, finding replies to objections, deducing ‘consequences’ reduce to making the right moves in a newly-mastered language game; becoming more and more easy as one gains in skill and confidence. Meanwhile the reality of the dialectic, its underlying meaning, disappears from view. The illusion is lost. To remedy that I shall now describe a simple metaphysical ‘experiment’. In its naive directness it approaches as close as possible to expressing the inexpressible ‘what-it-is’, not by finding new words but rather by inducing the feeling of disorientation at the thought of its rejection in its most acute form.


278. I put a coin into an opaque container, close the lid and shake the container. After spinning for a short while the coin comes to rest on the bottom. All I know is that the coin has landed either on heads or on tails. Now I have a choice: shall I remove the lid and see how the coin landed? or shall I shake the container again? The possibility of removing the lid is like an open door; if I choose to pass through then I shall know; the question will be answered. Instead, I shake the container I The door shuts permanently. I shall ‘never know how the coin landed’. But what does that mean? The realist in me wants to say: Behind that permanently shut door a portion of reality has closed itself off from all future inquiry. The answer to my question is there; only it is no longer an answer for anyone to know. The dialectic refuses to understand what I ‘mean’: There is nothing ‘there’; the question whether the coin landed on heads or on tails does not ‘have’ an answer. It did have an answer, in a different sense (22/228, 25/251). For a while the door remained open; I might have chosen to pass through. But now the door has not simply ‘shut’! it has ceased to exist. There is no ‘closed door’, no ‘other side’, nothing ‘behind’ or ‘closed off’. — The experiment has provoked me into saying that once an event becomes inaccessible to investigation it is as if its ‘reality’ were extinguished, as if it never happened. — The thought causes nothing, short of sheer panic.


279. Panic is panic; not a contribution to an argument, not an objection. Of course saying that makes no difference to what I feel. Nor does the reminder that the ‘destruction’ of past events is not the physical destruction of objects; or that events do not simply go ‘missing’ but rather that the chain of causes and effects traces back into different ‘possible worlds’, so constituted that the question which world is the actual world does not ‘have’ an answer; and that both these statements are meant, not as a description of ‘metaphysical facts’ but only the rejection of an illusion (23/237). It remains the case that, he ‘destruction of past events’ is how one feels compelled to describe the result of the dialectic; that is how reality appears to ‘appear’ once the illusion is stripped away. To dismiss that reaction as merely the illusion’s description of its own rejection may be the right thing to say but remains deeply unsatisfactory. However, that is not surprising; while the illusion is still in us, we ought to remain ‘unsatisfied’.


280. Let us attempt to describe the ‘result’ of our experiment without falling into the trap of describing the realist’s picture of an ‘anti-realist reality’. The past is past, gone forever. I cannot ever return to the state of intimacy with an event which I enjoyed while the event remained within my present perception. But I possess a memory. Whatever I learned through perception I retain; only now fixed and immobile; perception loses its ability to enlarge upon itself. There is also inference from present evidential traces. But what becomes of memory and inference if past events are ‘real’ only for possible knowledge? That is already an illusion-infected description. We should ask instead. In what sense is there a ‘past’ at which we aim our thoughts? — understanding a proposition about the past entails grasp of its truth conditions; moreover, these conditions concern how things were, and not facts about ‘memory impressions’ or about present evidential traces. In order to grasp the point of this seeming concession, however, we must ask ‘how it might be if things were otherwise’; what could lead someone to ‘deny the reality of the past’. Had we accepted the argument for a global, meaning- theoretical anti-realism (24/241) we should have been obliged to view a speaker’s knowledge of the ‘rules’ which constrain the procedures for investigating the ‘truth’ of propositions about the past as exhausting his conception of ‘the past’. In the completed meaning-theoretical ‘explanation’ of these ‘rules’ all mention of ‘the past’ would be expunged; the whole sense of this language game consists in the manifestations of an implicit grasp of ‘canonical modes of truth-determination’ whose explicit formulation would completely avoid use of the past tense (and only a God, observing the language game from the ‘outside’, could complete this ‘explicit formulation’; cf. 25/249). To assert the reality of the past within the context of the ‘rejection of realism’ means simply to reject the erroneous conception of the ‘illusion to be rejected’ embodied in the false project of describing the ‘form’ of an ‘anti-realist, theory of meaning’.


281. What now accounts for the validity of what we take to be ‘procedures for investigating questions about the past’ (24/242-4): since the question no longer concerns the present ‘content’ of the speaker’s ‘implicit knowledge’ of ‘canonical modes of truth-determination’ we need have no reservations about using the past tense: The ‘validity’ of memory as a means of access to the past is explained by its causal relation to what happened in the past (transparently refusing to address the sceptic who questions the possibility of having knowledge of the past; 282). But we also conduct inferences from things remembered to things not remembered; and moreover, infer from present evidence how things stood in a past not remembered. The validity of the general forms of these inferences is accounted for by their application to cases where what is inferred can also be remembered: Inferences from the present to an unremembered past are validated by induction from parallel cases where the past is remembered. Inferences from one past state of affairs to another cohere with inferences which are performed when similar states of affairs occur in the present. Ability to perform the inferences entails knowing what it would have been to have witnessed what the passage of time has placed beyond the reach of perception, to be an ‘ideal observer’ who does not need to infer what took place because he remembers it (24/240). Our reliance upon the notions of ‘parallel case’ and ‘coherence’ is itself justified by the success of our practice (the world might have been different, with physical laws changing with the passage of time; because our world is not like that this ‘reliance’ never once comes into question). — These observations are not meant to perform the work which would have been performed by an account of the ‘form’ of an ‘anti-realist theory of meaning’. They present a valid alternative only when seen in the light of an ‘overlooked possibility’ (24/242). For the terms of reference of the original problem required us only to ‘explain why truth is not an arbitrary convention’ (24/240); in the present case, how the point of past-tense propositions is to describe how things were in the past; how a past reality is able to determine the present truth or falsity of our judgements of ‘truth’ and ‘falsity’ of propositions about the past.


282. Note that an ‘anti-realist theory of meaning’ would make it impossible to express the ‘hypotheses’ which the sceptic employs in order to cast doubt upon our purported knowledge of the past; for example, that it is ‘possible’ that the Earth came into existence five minutes ago. If the meaning of past- tense propositions simply consists in the ‘rules’ which constrain the way we investigate their ‘truth’ then the ground is removed from under the sceptic’s feet. There is nothing to be sceptical about, for his unfalsifiable ‘hypotheses’ cannot by definition be associated with ‘rules of investigation’. By contrast, our use of the past tense in ‘accounting’ for the validity of those methods of investigation seems open to a sceptic’s charge of begging the question. But we have already seen how the ‘corrigibility of verification’ does not itself lead to scepticism unless seen under the aspect of the ego or truth illusions (9/83-9, ZI/Z07, Z3/Z38). That point may be extended to the case where there is no question of overturning our judgements, no matter what ‘evidence’ turned up (as one might do if, say the Earth’s true age was discovered to be five years). The coherence of a ‘description of a possibility’ does not entail that it describes something which, for all we know, is possible.


283. There is a further dialectical point in asserting the ‘reality’ of the past (Z80), Phenomenalism entails the unreality of time and thus destroys the pastness of the past; as indeed the presentness of the present, the futurehood of the future. For the passive observer, all that ‘is the case’ consists in facts about the object of his possible experience (5/45-6). But the temporality of the subject’s experience is not itself an ‘object’ of his experience; rather, it constitutes the ‘form’ in which experience presents itself to him (cf. Kant on the ‘transcendental ideality of time’; Critique of Pure Reason B46-58). Phenomenal reality consists in a series of events ordered by the asymmetrical relation of ‘before-and-after’. What is ‘past’, ‘present’ or ‘future’ merely reflects the changing point of view of the ‘window’ of consciousness as it ‘moves’ along the series. The ‘representation’ which the passive observer constructs of a world of ‘temporally ordered events’ remains outside time; the ‘representation’ of a timeless reality in which there is no ‘past’, ‘present’ or ‘future’; for there is no ‘representing’ the fact that it is now. — But ‘representation’ is not a ‘private content’; representing reality is an action with consequences for a subject whose activity necessarily extends beyond acts of representation (18/178). The fact that a representation represents what is past, or present, or future, and not merely something that ‘happens’ at a particular point in the asymmetrical ‘series of events’ crucially affects its significance for the users of representations. The fact that it is now determines what changes an agent has the capacity to effect in his surroundings; for at every moment events become past and consequently immune from his causal influence. Time is ‘real’ because the subject of experience is ‘necessarily embodied as an agent’.


284. The rejection of phenomenalism secures the futurehood of the future, for just the same reason that it secures the pastness of the past. However, the question remains whether one may speak of the future as possessing the same reality which we attribute to the past. Consider first now the future appears in the meaning theorist’s metaphysical vision: Whereas the ‘truth’ of past-tense propositions reduces to the possibility of their present verification, no such reduction is necessitated by the application of an ‘anti-realist theory of meaning’ to future-tense propositions. The ‘canonical’ method for determining their truth is to wait and see whether the possibilities which they describe become present; the present, provisional ‘verification’ of future-tense propositions waits upon the determination of their truth. The future is thus more ‘real’ than the past. Now this is just the ‘realist’s picture of an anti-realist reality’ (280): The pastness of past events makes them inaccessible to ‘direct’ inspection; they are consequently ‘unreal’. By contrast, future events have not yet passed us by; we can still catch them before they disappear into the past, before their ‘reality’ becomes an ‘unreality’.


285. For us, the question to ask is whether the ‘reality’ of the future can be secured by an argument parallel to that which we used to secure the reality of the past (280-1): In the case of the past, memory may be said to constitute a ‘direct’ means of verifying propositions in the past tense. (For the meaning theorist, ‘memory’ reduces to a ‘method of verification’ whose employment the completed ‘anti-realist’ theory would describe without using the past tense.) The conception of what would have been remembered then ‘regulates’ our employment of ‘indirect’ methods of investigation involving inductive inference from present evidential traces or from what we already know of the past. Now intention cannot be viewed as a ‘direct’ means of verifying propositions about the future, parallel to verification by memory; for there will always be a more ‘direct’ means, viz., to wait and see if the intention is realized (notwithstanding the observation that it is not merely a happy accident that intentions are generally realized). Moreover, whereas memory can be of events which were not my own actions, I can intend only what is or will be under my control. The form of ‘indirect’ verification by inductive evidence is such as only would have ‘cohered’ with a ‘direct’ knowledge of future states of affairs, parallel to memory; but just because that ‘conception’ is necessarily a fantasy, the inductive inferences which it ‘regulates’ cannot be regarded as ‘determining the truth about the future’. The very fact that we must wait for what is supposedly ‘in the future’ to really happen and not disappoint our expectations means that our thoughts cannot strictly be said to aim at the future. Instead, one can only try either to anticipate what the future will bring, or else to bring about states of affairs. For an agent, ‘describing the future’ does not have the same point as ‘describing’ either the present or the past.


29. The antinomy of phenomenalism and realism


286. To someone thrown into a state of shock or panic by the thought of what the rejection of realism with respect to propositions about the past ‘really means’ (28/278-9) it is no comfort to learn that, according to certain rejected views, the past might have been even less ‘real’ than it in fact is. The ‘dialectical point’ of asserting its ‘reality’ (28/280, 3) reduces to the rejection of one mode of ‘unreality’ in favour of another; and borrowing the term ‘real’ appears only a cheap and clumsy cover-up. Nor is it any comfort to be told that such a reaction is only a symptom of that very illusion which is being rejected (28/279). That is like being told: ‘If you disagree with me then I shall refuse to take what you say as an argument; it is impossible to disagree with my theory because any semblance of ‘disagreement’ is in reality only a symptom.’ Now a metaphysic cannot coherently admit the ‘possibility’ of its being wrong; it is forced to explain away the ‘arguments’ of all who oppose it. On the other hand, to formulate the arguments for one’s own ‘position’ in such a way that nothing could ever be perceived to count as ‘disagreement’ is to renounce all claim to objective status; it becomes only the self-refuting attempt to render one’s metaphysical vision incorrigible. But the difference between the two can sometimes seem vanishingly small. One must somehow contrive to be sensitive to opposing arguments while remaining steadfastly insensitive to all ‘symptoms of shock and panic’. Whatever reactions may be prompted by its seeming ‘results’, one can only follow the dialectic wherever it may lead.


287. But how long can one’s commitment to a metaphysical inquiry last out against the conviction that ‘there must be a way round the argument’? We are now faced with two equally powerful illusions, and so must ask that question twice: If phenomenalism and realism are after all only ‘truisms’ which have been ‘falsely described’ and whose ‘refutation’ refutes only that false description, or if, on the contrary, ‘ego’ and ‘truth’ are metaphysical absolutes which cannot be ‘refuted’, that has yet to be shown. The onus remains with anyone who refuses to ‘accept’ the refutations to find a gap in the dialectic if he can; we are ‘sensitive to opposing arguments’. But if he cannot, then he must keep silent. The refutations of phenomenalism and realism are like walls, blocking the path of one’s thoughts. Now a real ‘wall’ is a more or less temporary obstruction; walls can be climbed, or knocked down. One is tempted to think of metaphysical arguments in the same way: that one dislikes the conclusion is sufficient reason for attempting to overcome the wall. But let us now assume that it is possible for a metaphysical argument to be valid; that the ‘challenge’ which it issues cannot be met. Such a ‘metaphysical wall’ is not like a real wall: There is no ‘blocked path’; only the illusion of a ‘path’. I reach the wall only to find that I am once again facing in the other direction. The crux of the simile is this: If you are not able to ‘overcome the wall’ don’t stand there silently ‘hoping’ that it will fall down or even pretending that it ‘isn’t really a wall’. Find a new path.


288. There is no longer any question for us whether the ego and truth illusions ‘really are illusions’; whether the arguments which refute them are ‘really valid’. The only question is what the arguments really mean: What is the underlying nature of the illusions? where do they come from? and whence do they derive their power to deceive? It now seems as if the investigations up to this point were only a preliminary ground-clearing in order that we might face this problem directly. For our goal must surely be to throw off the illusions; to destroy and not merely ‘refute’ them. The reason that their refutation does not succeed in destroying them cannot be that they have not, after all, been ‘refuted’; from that ‘possibility’ we must turn away. Suppose instead that their refutation is only insufficiently constructive (4/35); that the illusions have not yet been ‘adequately expressed’ (1/9, 10). The dialectic of the ego and truth illusions would then be in that sense incomplete. But that supposition is ungrounded; and we should not know in any case where to search for a ‘completion’. Perhaps then there remains other ‘constructive’ work to do, outside the dialectic of illusion. But our account of the very possibility of metaphysics (1/4-6) commits us to a ‘full-blooded dialectic of illusion’ (8/77). ‘Outside’ the dialectic of illusion there is nothing that can call itself ‘metaphysics’. There remains one last alternative: that the completion of the dialectic of the ego and truth illusions does not complete the ‘dialectic of illusion’; that there remains the most fundamental component of that dialectic to be worked through. Now the ego and truth illusions entail the illusion of transcendent ‘metaphysical facts’. That suggests that a dialectic of the illusion of transcendent metaphysics will supply the missing insight; a knowledge which destroys the illusions by identifying their common source in the illusion of transcendent metaphysics.


289. How different is our present situation from that normally encountered in a philosophical ‘dispute’! Having refuted one’s opponent’s position, explaining his ‘mistake’ serves as an inessential heuristic. The mistake may indeed have multiple ‘causes’; or perhaps there is no determinate diagnosis to be found, the ‘position’ just crumbles away into nothing. But for us now, recognition of the urgent need to destroy the illusion means that ‘diagnosis’ becomes the whole problem, the question which has been waiting to confront us from the beginning.


290. Our first task is to determine the dialectical relation which holds between the truth and ego illusions: When I first pause to reflect I happily embrace both illusions. I am a Cartesian ‘ego’ whose knowledge of an ‘external world’ can never match up to the certainty with which it ‘knows’ its own ‘mental contents’ (3/28). When I then begin to doubt whether my common sense I is right in attributing to me any kind of ‘knowledge of an external world’ I do not, at first, also doubt whether questions concerning its existence or constitution may be said to have answers. That doubt only occurs to me on deeper reflection, when I find myself incapable even of thinking about ‘an object which is not part of my own mind’. But that ‘naive’ idealism gives way in turn to its sophisticated reformulation as a ‘metaphysic of experience’ (3/31, 4/42)., Now the crucial question arises whether, with the rejection of naive idealism, I have after all found a way to hang on to the thought that ‘undecidable questions nevertheless have answers’; whether it is possible to be, simultaneously, a phenomenalist and a realist.


291. Once again I am contemplating the ancient bronze vase, giving full rein to the truth illusion (20/198-9). It occurs to me that I can perform the same operation upon my conception of the vase as I performed with the red lampshade (3/28). The vase-out-there dissolves away, to be replaced by a vase-in-my-mind. Accepting the phenomenalist refinement of that naive statement of idealism I am left with this thought: Any proposition which I entertain concerning the history of the vase concerns ‘my possible experience’ (5/46). But what exactly does that mean? My entire system of beliefs consists in an ‘interpretation’ of my subjective experience I Its correctness or incorrectness is determined by the way it ‘matches’ or fails to ‘match’ the world of my experience; its meaning consists in its ‘anticipation’ of future perceptions which it is possible for me to have. Recognition of undecidability means that a proposition simply has no work to perform in the ‘interpretation of experience’; I can only imagine perceptions which it might have served to ‘anticipate’. For I cannot, as Kant seems to imply at one point (Critique of Pure Reason B52l-5) meaningfully aim my thoughts at the ‘object’ of ‘possible experience’ which only a suitably placed observer might have had, the one true causal explanation out of the many epistemically possible causal explanations of experience which it is possible for me to have. The ‘possible experience which an observer might have had’ is experience I can only imagine; there is nothing for my thoughts to aim at except the world determined by the experience which it is possible for me to have. If I believed that the world is a rigid, deterministic system and if I did not rule out the possibility of an indefinite expansion of my knowledge then I might indeed aim any proposition concerning the history of the vase at what future knowledge of that system would determine concerning all its past states (this is a possible interpretation of the ‘Analogies of Experience’ B2l8-65). But then no proposition concerning the past would be undecidable.


292. How do things stand with the illusions themselves? It seems to me as if the very identity of the vase, its underlying ‘metaphysical nature’, undergoes a transformation: First, the vase appears to ‘dissolve’, becoming just a character in the story of ‘my world’; existing wholly for me and not in itself. Only with the rejection of ‘existence wholly for me’, of my idealist ‘dream world’, could the ‘in itself’ assert itself. But then it does so in a manner which destroys the existence of the vase for me. If there is any ‘subject’ for whom the vase may be said to exist that subject can only be God (23/236, 25/251-2). It is now tempting to argue, At least one of the ‘conceptions’ of the vase must be an illusion. But there are no grounds for preferring one to the other. Therefore both might as well be illusions. As long as I keep my mind fixed upon this point their power indeed diminishes. But the argument is ‘non-constructive’; it serves only as a psychological defence against the illusions.


293. Phenomenalism and realism form a dialectical antinomy. An antinomy is not a logical contradiction; the semblance of a ‘contradiction’ derives rather from the falsehood of a common presupposition (Critique of Pure Reason B448-50, B530-5). If we could only attain a clearer view of that presupposition we might be able to work back to the source of the two illusions. But the fact of the antinomy tells us nothing except that there is something to be found, an element common to both illusions. Moreover, reworking the dialectic of the ego and truth illusions can only tell us what we already know. The source of the ego and truth illusions is something whose recognition would entail the destruction of the illusions (288); but nothing we have learnt so far promises to do that.


294. The first clue is this: Both phenomenalism and realism entail global scepticism (9/83-4, 21/207). On closer examination, there appears a formal correspondence between the two arguments. In both cases there is a question which must have an answer, but to which ‘nothing could count as an answer’ The phenomenalist must ask, but cannot answer, the question: ‘Do my beliefs really correspond with what they purport to represent?’ (9/87). The realist must ask, but cannot answer, the question: ‘Does this purported ‘verification’ mean that my belief represents how things really stand? (23/238). In both cases, because what purports to be ‘knowledge’ is only the property of a representation it cannot be knowledge. Something is missing from the picture.


295. The second clue is the formal correspondence between the constructive refutations of idealism and realism; and, equivalently, the correspondence between the models of the relation between mind and reality determined by a phenomenalist and a realist metaphysic I The phenomenalist’s model of the thinking subject is a ‘passive constructor of representations’ 5/45); the realist’s model is of ‘representations with a life of their own’ (23/236). But the models are incoherent. Only its potential consequences for the interpretation of an agent’s behaviour can determine a distinction between ‘correct’ and ‘incorrect’ applications of a concept (6/58); only its potential consequences for the interpretation of an agent’s behaviour can differentiate the act of ‘conjecture’ from the act of ‘mere supposition’ (23/232-3). We may present the two refutations in the form of a single, indiscriminate argument. The act of ‘representing reality’ is only one among the many kinds of action performed by the thinking subject. The behaviour of an agent is the only context in which we can so much as conceive of ‘representations’ existing; whatever a representation accomplishes is necessarily accomplished by the agent whose act of representation it is. The power of the two illusions is just that it doesn’t seem like that. The act of representation appears utterly self-sufficient and complete. To ‘prove’ that all I have to do is just think; and there, before my own mind’s eye, is the very example of what I ‘mean’.


296. The third and final clue most strongly illuminates the element of ‘transcendent metaphysics’: The reduction of thought to bare ‘representation’ forces the recognition that ultimate reality cannot be made an object of thought. I can make the vase an object of my thought but not the ‘fact’ that the vase is ‘wholly for me’; the ‘fact’ that all the ‘objects’ of thought and indeed the ‘self’ that thinks are only modes of an underlying ‘this...’ (3/31, 4/42, 5/46, 6/56). Alternatively, I cannot represent to myself the ‘fact’ of the brute reality of the vase as it is ‘in itself’; its role in ‘determining’ the truth or falsity of thoughts concerning it. For in that role it is no longer a ‘vase’ but only what all the ‘in themselves’ have ‘in common’; to be ‘that...’ (19/95, 21/209, 23/229). The rejection of the ego and truth illusions is the rejection of these two opposed ‘ultimate realities’; which one alternately sees but cannot describe.


297. The three ‘clues’ strongly suggest that the source of the ego illusion and the source of the truth illusion are one and the same. But they do no more than that. For we have not yet identified the source, but merely given an illuminating redecoration of the two illusions. The crucial question remains: Why are we disposed to regard the notion of a representation’ in this way? It is not simply a philosophical error which can be put right by our being forced to admit’ the truth. Accepting the validity of the ‘refutation’ of the illusions does not destroy them; but nor does the discovery of their ‘dialectical relation’. We now see the illusions for what they are; we still cannot see through them.


30. The metaphysical attitude


298. What is the meaning of ‘metaphysics’? It is said that the word was originally intended to refer simply to ‘what comes after the Physics’ in the corpus of Aristotle’s works. While I do not wish to challenge scholarship on this point, it is hard to believe that, had Aristotle’s Metaphysics ‘come after’ the Poetics, metaphysics would have been called ‘metapoetics’. On the contrary, what the scholars regard as a ‘misconstrual’ of the word ‘metaphysics’ as a description of the object of metaphysical knowledge proves to be the closest that it is possible to approach to a definition: Metaphysics deals with a reality which stands beyond the objects of all other forms of knowledge; and which therefore has always been and will always be the same, however much those other objects might alter. Because metaphysics is ‘beyond’, it is possible to know everything else there is to know and still completely lack metaphysical knowledge; it is not the ‘foundation’ which ‘underpins’ the meaning or the certainty of other forms of knowledge, for its lack is not in any sense a ‘deficiency’ in any of the other things we take ourselves to know. Because metaphysics is always ‘the same’, it cannot satisfy the changing forms of human needs and desires as they progress through time; if there is such a thing as ‘knowledge of the meaning of life’, its content must necessarily vary according to the content of those needs and desires and is therefore not metaphysics. Metaphysics is thus complete in itself and without consequences (save for the consequences of rejecting the supposed ‘consequences’ of a false metaphysic; 27/276). Indeed, only such a ‘self-contained’ metaphysics can set itself the ideal of knowing its Object ‘completely and in such a way as to integrate all partial perceptions into a meaningful and consistent whole’ (1/3).


299. Metaphysics is neither super-physics, nor para-physics: It cannot be reduced to the ‘most general theory of the nature of the world’, for that implies a continuity with scientific theory and consequently an interdependence between metaphysics and science. This remark is not asserted dogmatically, but only as an observation of the fact that no such ‘connection’ has been revealed by the dialectic. Nor, on the contrary, does metaphysics stand in competition with physics, a ‘theory’ of a ‘non-physical reality’ which does not exist for any form of inquiry which respects the aims and results of physics (12/110). Attempts at a positive definition indeed prove futile, Metaphysics cannot, for example, be defined as knowledge of what is necessary, remaining true in possible worlds where other, contingent forms of knowledge would become false; for that does not distinguish it from mathematics. Nor can metaphysics be defined as knowledge of an ultimate reality; for that does not distinguish it either from fundamental physics nor from psychoanalysis. All one can say is that its own ‘necessity’ and ‘ultimate reality’ must be understood in a special sense. This ‘special sense’ is the ‘beyondness’ of metaphysical knowledge. But that only tells us what metaphysics is not; and not what it is.


300. A ‘metaphysics of meaning’, the dialectic of the relation between mind and reality, and consequently of the metaphysical illusions of ‘representation’ (29/294-6), entails a metaphysics of metaphysics. Not only must it establish the meaning and possibility of a dialectic of illusion in the face of its rejection of ‘transcendent’ metaphysics (1/4-10); the illusion which it rejects is an illusion of transcendent metaphysics (1/6, 29/296, 31/312) and will indeed prove to be the illusion of transcendent metaphysics (Ch.32). But what is ‘transcendent metaphysics’? and supposing the illusion of its possibility to inform and sustain the very meaning of ‘metaphysics’, with what right can a project of inquiry which rejects that illusion call itself ‘metaphysics’? I wish to argue that no person who has not to some degree been gripped by the metaphysical attitude can be brought to understand what ‘metaphysics’ means. In the eyes of such a person, the dialectic of illusion fails to ‘adequately express’ anything ‘metaphysical’, for he lacks the power to see its illusions. A definition of ‘metaphysics’ is only for someone who already knows its meaning, just because the dialectic expresses the illusion only for someone who already has it; and all there is to learn about the rejected project of transcendent metaphysics and the essentially ‘metaphysical’ nature of the dialectic which rejects it can only be learned by working through the dialectic. The ‘outsider’ must remain forever on the outside. Now this will appear to the uncomprehending critic just another ‘self-refuting’ attempt to render ourselves ‘insensitive’ to all opposing arguments (29/286). We may reply that within the ‘community of metaphysicians’ there will indeed be argument and counter-argument. For them, metaphysics is an objective fact. However, the question of the possibility of a dialectic of illusion then transforms into the question of how one can know that one’s inquiry is really ‘objective’ in the face of the conceivability of the critic’s suggested alternative. that the appearance of ‘rational activity’ is only a collective hallucination.


301. The metaphysical attitude is not a posture which one takes up but a predicament ore finds oneself in. It begins with the discovery of questions; or rather, what appear to possess the character of ‘questions’; only so perversely unfamiliar that nothing one has ever learned and no technique of art or science could have equipped one to answer them. For they lay claim to a staggering generality and an endless depth of meaning that threaten to rule out the very possibility of intelligible answers. Indeed, they appear to throw the whole of reality into question, subverting and undermining the point of every thought and every action, including all ‘questioning’ and all ‘knowing’. Yet at the same time no identifiable consequences follow from this radical scepticism. Everything is equally affected and so everything continues on as before. It is as if a giant chasm appeared from nowhere and swallowed up the whole world, only leaving every object in the world intact and unchanged; or, as if before everything was ‘up’ and now everything is ‘down’. The upshot might be described as the ‘discovery’ of one single, bare question mark; were that not utterly nonsensical. What the metaphysical attitude ‘discovers’ seems beyond ‘description’: Something happened, one felt the question that was there; while at the same time ‘nothing actually happened’ (3/28). — Now this ‘feeling’ is where metaphysics begins and ends, its sole subject matter and its sole rationale.


302. The metaphysical attitude ‘discovers’ nothing in the world. The bare question mark does not question any particular thing because it questions everything: the world in its totality. In so doing, it places the self which discovers it outside the world. Previously, one took the here-and-now to be the ‘ultimate reality’; or, rather, until the question appeared, the ‘here- and-now’ did not even have a name; it was simply the reality of life itself, of projects and commitments, risk and struggle, success and failure; ‘ultimate reality’ did not have a ‘metaphysical’ sense. Then came the discovery that one can detach oneself in thought from the ‘here-and-now’: While from the point of view of practice the thinking subject remains ‘attached’ — one does not cease to live — that attachment takes on an unfamiliar aspect; it now appears to the viewpoint of detached ‘contemplation’ as something that merely happens in the world, alongside all the other things that ‘merely happen’. The detached self ceases to identify itself with the ‘actions’ of its attached counterpart.


303. While it remained ‘attached’ not only in practice but also in thought the self was lost in the ‘here-and-now’; that is how the detached self views its previous attachments. The act of mental detachment thus presents itself as the rediscovery of a previously lost self through a more intense self- consciousness: This description of the discovery of the metaphysical ‘point of view’ should not appear unfamiliar even to someone who has never felt the necessity of its questions. For an analogous process of ‘detachment’ and ‘self-discovery’ occurs when one stands back from one’s projects and commitments in order to question their value. But this ‘standing back’ is only a retreat from a previously unquestioned position to another within the world of projects and commitments; to question one’s values is itself a ‘project’. From the point of view of metaphysical detachment that project is just another thing that ‘merely happens in the world’.


304. Up till now, we have described ‘detachment’ in terms whose meanings are themselves undermined by the metaphysical attitude. For even to talk of things ‘merely happening’ misrepresents their true ‘aspect’. The denial of the ultimate reality of life contained in the view of the world-as-totality is a denial of the ultimate reality of time. Because action is ultimately unreal, nothing ‘happens’; there are no ‘events’. The world taken in its totality is, literally and not merely figuratively, a world viewed sub specie aeternitatis. Now when one speaks of the ‘activity’ of metaphysics, one means the attempt to comprehend the world thus viewed, and the ‘intellectual struggle’ of the detached, contemplative self for whom metaphysics sets that self-negating task. But in a world where time is unreal, there is no such ‘activity’: no ‘task’ and no ‘self’ to carry it out. The metaphysical attitude itself timelessly ‘comprehends’ its object. Meanwhile, the metaphysician continues to conceive of himself as ‘striving’ for comprehension; while simultaneously admitting the ultimate meaninglessness of that very description.


305. We shall see in the next chapter (31/313-5) how the aims of the original ‘metaphysical project’ are transformed in a dialectic of illusion. The question now is: What is the ‘project’ of metaphysical detachment? how does the detached self set about ‘comprehending’ the world-as-totality? One’s first reaction is to reject the very possibility of a ‘project’; the only consistent expression of the metaphysical attitude is complete mental paralysis. Indeed, we have seen how little there is in the way of projects for the pure idealist or realist. But then the history of metaphysics is very far from being ‘dialectical’; one finds all sorts of ‘projects’ whose motivation is strictly extraneous to the task of comprehending ultimate reality: The analytical taxonomy of ‘ultimate kinds’, proofs of the existence of God, deductions of; the ‘first principles’ of science or ethics. It seems the last thing a metaphysician would admit is that there is nothing to do; or perhaps it is because all we know of past attempts at metaphysics comes from books, and silence means not saying or writing anything. The original, pure ‘metaphysical attitude’ is identified, not through any projects, but by a single question. Previously, we took this ‘bare question mark’ to be without content (301); its all-inclusive generality rendered it vacuous. But now there appears an object sufficiently ‘inclusive’ to fit the question, the world in its totality. What the question puts into question is why there is something rather than nothing.


306. I shall not dwell on the manifold self-deceptions which have persuaded metaphysicians that the question of the existence of existence can be answered through rational argument: All such attempts are transcendent metaphysics; and transcendent metaphysics is an illusion. The only alternative is to reject the question. But that does not go without saying. Whether it accepts or rejects the possibility of transcendent metaphysics, the metaphysical enterprise appears Promethean in its ambitions; inspiring a reaction of hopelessness and defeatism: ‘Of course we cannot grasp the mystery of existence. We are so puny; while reality is so sublime. Either metaphysics pretends that we have powers which we could not possess; or it self-defeatingly cuts reality down to size in order to fit the powers which we do possess.’ -We must maintain that the final destruction of all metaphysical illusion would nullify that reaction. But what if that goal proves to be beyond the resources of metaphysics? There might be other ways to destroy the illusion; but we could not count on it. The only way to maintain faith in the enterprise is simply to ignore the defeatist’s protest. When we put the question: ‘Is metaphysics possible?’, that is the one ‘possibility’ to which our eyes remain shut.


31. The rejection of transcendent metaphysics


307. Something exists: But so what? The statement is so obvious that no-one would assert it even as a ‘statement of the obvious’; say, in the way one says: ‘It’s raining’ as the rain begins to fall. But the metaphysician means it differently: Lost in the ‘here-and-now’ of our mundane concerns, we have failed to notice a profoundly significant fact: that ‘something exists rather than nothing’ (30/305). Now, if it is ‘significant’, there must be such a thing as taking notice of it; giving that fact its rightful share of attention amongst all the other facts of which we ‘take notice’. One might think, therefore, that this is a fact which calls, as other facts call, for explanation. But in that case ‘explanation’ cannot be taken in its normal sense. For any physical object, one may ask what processes of construction, development or generation led up to its coming into existence and thus ‘explain’ the fact of its existence. An ‘explanation of existence’ necessarily refers to something beyond, in relation to which the existence of the given object is rendered intelligible. But the world itself is not an object in the world; it has no ‘beyond’. On the contrary, ‘the world’ is what stands ‘beyond’ every given object. Thus, to ask for an ‘explanation of the existence of the world’ in that sense is just to babble. The metaphysician will readily agree: ‘Taking notice’ of the fact that something exists entails a completely different form of cognitive activity from that involved in taking notice of the fact that some particular object exists. For it entails taking up the metaphysical attitude.


308. What does the metaphysical attitude require us to do? Suppose first that the fact that something exists still waits to be ‘explained’; but that this explanation is ‘metaphysical rather than physical’. Metaphysical, as opposed to physical explanation does not appeal to contingent facts; its truths are necessarily true, true in all possible worlds, no matter what the facts might be. Then the fact that something exists rather than nothing must be, contrary to what we first thought, a necessary truth; and the project of metaphysics is to establish its necessity. Now it would not suffice to prove that something had to exist, if that proof does not extend to a proof that whatever exists could only have existed in the way it actually exists; that reality could not have been constituted in any other way from the way it is constituted. For otherwise the original question would simply be replaced by the question. ‘Why do things exist in just this way rather than in some other way?’ As before, the answer to this question could not be a merely conditional, physical explanation. In philosophical terminology: The ‘ontological’ proof that the world had to exist must further establish that this world of all ‘possible’ worlds ‘had to exist’, if its point is not to be undermined. For the metaphysician to whom this project, of all possible projects, appears impossibly ambitious, there remains the alternative of renouncing ‘explanation’: ‘something exists’ reduces to the bare ‘invocation’ of the metaphysical attitude; a prayer, or a mantra.


309. I shall define transcendent metaphysics as the attempt to express recognition of the ‘profound significance’ of the fact that ‘something exists’ (30/306). The definition must be understood ‘dialectically’. It is indeed the least likely of all ‘definitions’ to communicate the meaning of the term to someone who has not studied metaphysics; the least useful of all dictionary entries. For transcendent metaphysics is a metaphysical illusion; and adequately expressing the illusion entails its location within a dialectic. The definition merely sets up the right ‘connection’. However, its very obliqueness requires some preliminary comment. The rejection of a Kantian I ‘metaphysic of experience’ entails giving up the project of ‘drawing the bounds of sense’. Whereas Kant defines non-transcendent questions as those which concern either the matter of possible experience or its a priori form, so that a definition of the ‘questions’ of transcendent metaphysics, and indeed a diagnosis of the illusion of transcendent metaphysics, follow as mere corollaries, the metaphysic of action makes no analogous claim for the ‘limits’ to the possibilities of ‘meaningful action’, and consequently of possible knowledge or experience. On the contrary, establishing the very possibility of a metaphysic of action as a ‘full-blooded dialectic of illusion’ requires the ‘adequate expression’ of the illusion of transcendent metaphysics. Now the ego and truth illusions have already revealed themselves as illusions of transcendent metaphysics (29/296); in that case the connection with the ‘transcendent metaphysics’ of the above definition waits to be proved, for it is certainly not immediately apparent. An even more difficult fact to assimilate is that pure ‘transcendent metaphysics’, by contrast with its ‘undialectical’ manifestations (30/305), is not even a project; for it chooses to remain silent rather than attempt to ‘express’ a ‘vision’ which cannot be expressed (308, 10): Its ‘attempt at expression’ is simply the intended ‘meaning’ of that meaningful silence.


310. Transcendent metaphysics is reduced to silence by the critique of its epistemology: If, per impossibile, there existed transcendent metaphysical facts, they would remain unknowable; for it would be impossible to reach objective agreement upon what those ‘facts’ were (2/14, 12/112-3): Pure thought sets itself the task of determining the nature of that ‘ultimate reality’ first revealed by the ‘discovery’ of the metaphysical attitude (30/301-4). It aims to begin with premisses which in no way depend upon the ‘results’ established by other methods of inquiry, metaphysical ‘axioms’ whose truth consists solely in the fact that it is impossible to deny them, and end with the completed system of their inferred consequences. The ‘system’ would then speak for every rational being. But intuitions about undeniability and completeness are incorrigible. One ‘system of metaphysics’ turns out to be as ‘valid’ as any other; the author of the system speaks only for himself. The intention of the metaphysical project is to embody, in its finished system, the ‘unaided movements of pure thought’; the rational mind then follows those movements as inexorably as it conforms to the laws of logic. But practice refutes the ideal; each system invents its own ‘logic’! — Now this argument stands on its own; and is perfectly fair. Nevertheless, it remains non-constructive; it brings us no nearer to grasping the meaning or the motivation for the rejected project; does not yet adequately express the illusion of transcendent metaphysics.


311. The idea that one can even address, let alone refute the ‘meaningful silence’ of pure transcendent metaphysics seems at first absurdly optimistic. How can one deny a person the right to wonder at the mystery of existence, or even so much as raise a question mark against his attitude? Indeed, the one absolutely necessary qualification for understanding metaphysical argument is to have experienced the grip of the ‘metaphysical attitude’ (30/300). The reality of the experience could only be denied by someone who rejected the very possibility of metaphysics. Of course, this question could have been raised earlier, against the attempted ‘refutations’ of the ‘meaningful silences’ of realism and idealism. We learned then that it is possible for a refutation to be constructive; that the rejected ‘silence’ is not simple silence but rather a silence of paralysis, an opposition of ‘forces’ — the desire to speak, the realization that ‘nothing can be said’ — whose structure can be ‘adequately expressed’ by means of a dialectic. The rejection of that ‘silence’, far from refusing to admit the reality of its ‘experience’, presupposes that the experience is ‘objective’; that the dialectic which expresses it speaks to a genuine audience. It is only in the present case, where the inquiry appears to hit rock bottom, and the ‘Promethean’ ambitions of a dialectic of illusion finally reveal themselves (30/306), that one finds oneself threatened with loss of nerve.


312. The silence of pure transcendent metaphysics has a ‘structure’; it is the end point of a process of thought which finds itself ‘reduced to silence’. First, there appear ‘questions’ which ‘put reality itself into question’ (30/301). These questions seem to reveal the possibility of ‘detaching’ oneself in thought from the ‘here-and-now’ in order to view the world as a ‘totality’ (30/302-3). From that detached viewpoint time reveals itself as something ultimately ‘unreal’, together with all that that implies (30/304). Then one discovers the ultimate question ‘why something exists rather than nothing’ (30/305-6). Meanwhile, the posture of detachment, the ‘taking seriously’ of metaphysical questions, including that ‘ultimate’ question, appears to demand that one attempt to answer them; that the ‘metaphysical attitude’ express itself in projects of transcendent metaphysics (308-10). Finally, with the rejection of those seeming ‘projects’, only silence remains. For the clue to the ‘constructive’ refutation of transcendent metaphysics we must return to the beginning: ‘Detachment’ must be shown to be an illusion; the world cannot be viewed as a ‘totality’; time is real. Now this is just the ‘metaphysical vision’ embodied in the dialectic of the metaphysic of action: The self is necessarily an agent and cannot detach itself from its worldly associations, either by becoming a disembodied ‘transcendental ego’ or by losing its subjectivity and finitude in the mind of God. What remains to be shown is that the antinomy of phenomenalism and realism is none other than the illusion of transcendent metaphysics; that the illusion of representation is the illusion of detachment. In that case we should not require in addition a constructive refutation of transcendendent metaphysics, for that will have been taken care of by the dialectic of realism and idealism. We shall return to that task in the next chapter.


313. The ultimate goal of the dialectic of the ego and truth illusions is the diagnosis and critique of the illusion of transcendent metaphysics. But a question raised earlier against the self-understanding of such an enterprise has not yet been adequately answered: Supposing the illusion of the possibility of transcendent metaphysics to ‘inform and sustain’ the very meaning of ‘metaphysics’, ‘with what right can a project of inquiry which rejects that illusion call itself ‘metaphysics’?’ (30/300). It does not suffice simply to recognize the existence of the ‘metaphysical attitude’: The self-understandings of positivists, pragmatists, existentialists and marxists, and their conceptions of their mutual differences, essentially involve a conception of how their different projects serve to combat the metaphysical attitude. Each takes the metaphysical attitude ‘seriously’; yet none deserves the title ‘metaphysics’, nor welcomes it. What each fails to perceive is the inner necessity of metaphysical questioning, its demand for satisfaction through a form of inquiry which promises ‘complete and whole’ knowledge of its special ‘object’ (30/298): The dialectic of illusion inherits from transcendent metaphysics the desire to know the ultimate reality.


314. A dialectic of illusion, in its rejection of transcendent metaphysics, regards the primary ‘disclosure’ of ultimate reality in the metaphysical attitude at one remove: from the viewpoint of a second-order attitude which treats the ‘metaphysical’ attitude as an experience against which a question has been raised, whose real meaning is not what it ‘appears’ to be. Just as transcendent metaphysics has its ‘appearance and reality’, the ‘here-and-now’ of perception and action and the ‘world-as-totality’ into which the ‘here- and-now’ disappears, so a dialectic of illusion discovers an ‘ultimate reality’ different from the ‘world of appearance’: The world of appearance is the ‘here-and-now’ as transcendent metaphysics sees it; the thought that thinking that the things I touch and see, the world of my projects and commitments are the ‘ultimate reality’ is an illusion. Ultimate reality is what one perceives when, having worked through the dialectic of illusion, one comes to realize that what transcendent metaphysics takes to be an ‘illusion’ concerning the ‘here-and-now’ is its own illusion. The ‘here-and-now’ is, after all, the ‘ultimate reality’; only not the ‘ultimate reality’ of transcendent metaphysics. But nor, on the contrary, is it the ‘here-and-now’ which existed prior to metaphysical reflection. Only by working through the dialectic can one gain an insight into the ‘ultimate reality’ of the ‘here-and-now’; only after one has lived through the attempted destruction of the ‘here-and-now’ by the illusion of detachment can one see the world under the aspect of ‘ultimate reality’.


315. From transcendent metaphysics, the dialectic of illusion inherits not only the concept of an ‘ultimate reality’, but also the fundamental problem of the possibility of metaphysical inquiry; the possibility of knowledge of ultimate reality. Transcendent metaphysics is refuted by its ‘problem of beginnings’ (310): The very attempt to make one’s ‘system’ speak for ‘every rational being’ means that one is speaking only for oneself. No person has the right to question another person’s ‘metaphysical vision’ of ultimate reality. But the dialectic of illusion at first appears to be in an even worse predicament; for it admits that its starting point is not beyond question, but is only the brute given of ‘metaphysical illusion’ (1/5-10). It is possible to be wrong about what one takes to be a ‘metaphysical illusion’. However, what first seems a fatal defect proves to be the necessary condition for objectivity: I may claim to be speaking not only for myself only if I admit another person’s right to question my ‘vision’. For I wish to allow the ‘metaphysical experience’ to speak for itself; to ensure that the voice of the dialectic overrides private fantasies and obsessions. My judgements on these matters may turn out false. But only by indeed recognizing that possibility of error can I hope to attain ‘knowledge of ultimate reality’.


32. The transcendence of the ego and of truth


316. The ego and truth illusions are illusions of ‘transcendent metaphysics’ (29/296); yet neither the phenomenalist nor the realist need have regarded his metaphysic as an ‘attempt to express recognition of the profound significance of the fact that something exists’ (31/309). However, that proves nothing I Just as it is possible to be a partisan of the transcendence of either the ego or of truth, and consequently to refuse to admit both the antinomy of phenomenalism and realism (29/291-3) and the identity of the two illusions in the illusion of representation (29/294-7), so neither party need have followed through all the consequences of his own metaphysic. To fully understand either position entails seeing how its metaphysical questions lead inexorably to the metaphysical attitude of detachment (30/301-4). For both the phenomenalist and the realist refuse to accept that the self is ‘necessarily an agent’ that ‘cannot detach itself from its worldly associations’ (31/312). But if the self can detach itself than it must: To view the ‘here-and-now’ as something from which it is possible to detach oneself is to detach oneself from it. All that remains to do is to recognize the fact and to follow through its consequences. — Our task now is to establish the reverse implication: that the attitude of detachment, supposing its ‘discovery’ to have occurred in the absence of the ‘primary manifestation’ of either the ego or truth illusions (3/28, 20/198), leads inexorably to the antinomy of phenomenalism and realism.


317. The attitude of detachment ‘discovers’ the possibility of placing the mind in a ‘new relation’ with reality; the self now stands ‘outside the world’ (30/302). The ‘here-and-now’ of its projects and commitments reduces to the object of pure contemplation: Whereas the world appears as a world of both thought and action, in reality all ‘action’, all involvement in the ‘here-and-now’, is merely an object for thought; the ‘here-and-now’ disappears in a world-as-totality (31/314), and with its disappearance the world becomes a world of pure thought. Now thought thinks propositions; and what it thinks is true or false of reality. How is that possible? A proposition can be true or false of an object in reality only if its terms are secured a reference. Let us first ask now reference is ‘secured’: In some cases, the propositional sign contains indexical terms, ‘this’, ‘here’, ‘now’ etc.; the proposition is true or false of the particular object or objects to which the indexical terms refer. Where there are no indexical terms, the proposition is said to be ‘true or false’ simpliciter; it goes without saying that to be true or false, the proposition must be true or false of the real world rather than merely being a piece of fiction. But ‘reference to the real world’ ultimately presupposes indexical reference; however much one packs into descriptions of objects, those descriptions are guaranteed a unique reference only through the location of those objects in a spatio-temporal ‘framework’ whose point of origin is the ‘this’, the ‘here’ and the ‘now’. (The reader may be referred to Ch.l of Strawson’s Individuals; with the proviso that the argument there not be understood as a ‘phenomenology’ of reference; 8/72-3, 11/101.) However, that account of reference presupposes the impossibility of detachment: From the detached viewpoint, thought is in no position to refer to a ‘here-and-now’; the uniqueness of descriptive reference must be secured instead through the metaphysical principle of the identity of indiscernibles. With the disappearance of the ‘here-and-now’, the only thing which distinguishes this ‘thought world’ from a world of fiction (for which the identity of indiscernibles is indeed trivially true) is the brute, inexplicable fact that it is.


318. It might be mistakenly supposed that the identity of indiscernibles could be ‘refuted’ while allowing the possibility of the detached viewpoint I Can we not imagine the ‘brute, inexplicable fact’ of two identical worlds, with a detached subject contemplating their ‘qualitative identity’ and ‘numerical diversity’? — What one imagines is only a non-detached spectator who views the two worlds from a vantage point between them; the one world is this, the other that. But let us ask instead how one should conceive of the nature of thought and of propositions if the possibility of reference presupposed the identity of indiscernibles: A proposition is a meaningful propositional sign. It is not the inscription or sequence of sounds qua object, but rather their being used to mean something, together with all that that entails. If one wished to ‘gather together’ everything which contributed to a proposition’s meaning, one would have to include events of demonstrative reference and the objects which they involved, upon which the reference of propositions finally depends. That is tantamount to saying that the proposition cannot coherently be thought of as something self-contained; essentially separate from the world of which it is true or false. But that is just how one would have to think of propositions were one to deny the reality of indexical reference. The vehicle of thought must remain untainted by connections with the ‘here-and-now’, as much ‘detached’ from the world as the thinking subject. To regard propositions in that manner is to treat them as ‘representations of reality, in a vicious sense; and that illusion is the illusion of representation.


319. The detached self discovers that its thoughts are ‘only representations’: The argument may be read as beginning a ‘genetic analysis’ of the illusion; a timeless ‘history’. It is not intended as an empirical ‘hypothesis’ concerning actual thought processes; nor is the inference ‘logical’, for in logic anything whatever may be taken to ‘follow’ from a necessarily false premise. It is rather an attempt to place the various ‘moments’ or ‘aspects’ of metaphysical illusion in the right dialectical order. In practice, we simply encounter one of those aspects as a ‘primary manifestation’ (316); the dialectic fills out the missing ratiocination, and that is its way of comprehending what the illusion ‘really means’. Thus, what the dialectic presents as a stage in the ‘history’ of the illusion need never actually manifest itself; and that is indeed the case with the abstracted notion of ‘representation’ which we now discover. The illusion of representation manifests itself either in the form of phenomenalism or realism — we have yet to determine why those exhaust the possibilities — and prior to that it is nothing but an ‘abstraction’. Indeed, that provides the clue to the next stage of the analysis: The ‘world of thought’ which the abstracted ‘representation’ represents is a pure form without a matter. To say that its actuality is constituted simply by the ‘brute fact that it is’ (317) does not yet give the matter. But the ‘form’ is unthinkable in abstraction from a specific ‘matter’; the concept of thought as ‘representation’ is unthinkable outside a conception of how thought relates to reality, its role in a metaphysic.


320. The relation between the detached self and the world-as-totality is the one thing we have so far omitted from the description of the metaphysical attitude. The self places itself ‘outside the world’ (30/302), ‘in a new relation with reality’ (317). Then is the self itself ‘real’? The detached self contemplates the actions of its attached ‘counterpart’ as something as ‘real’ as all the other constituents of the world-as-totality. But that does not answer the question. The denial of the reality of the ‘here-and-now’, of agency as the mode in which the self relates to its world, is the affirmation of the reality of a different mode of relating to reality; the reality of ‘pure contemplation’ and of its ‘representations’, the thoughts which it directs at the world-as-totality. Once again we press the question: What is the ‘reality’ of the contemplating subject, the detached self? There are just two possibilities: Either that self exists or does not exist; either the self belongs to ultimate reality, exists for the metaphysical viewpoint, or it does not. Now the detached self can exist neither outside the world-as-totality nor within; for ‘outside’ the world there is nothing ‘real’, while ‘within’ the world one finds only the attached self. The only remaining possibility is that the self is none other than its world; ‘I am my world’ (Tractatus 5.63), the world whose ‘matter’ is ‘my possible experience’ The identity of indiscernibles is discarded; only to be replaced by the transcendental unity of apperception (5/45-6). To reject that possibility means rejecting the ultimate reality of the ‘subject’ of detached thought: The reality of ‘pure contemplation’ is merely that of thoughts thinking themselves, of thoughts ‘with a life of their own’ (23/236), whose ‘world’ is simply the ‘determination’ of their truth or falsity.


321. The illusion of detachment leads either to the ego illusion or to the truth illusion. But we have given no ‘dialectical’ reason why it should lead to the one rather than to the other. So one might think: ‘If there is no reason why the metaphysical attitude should. choose to embody itself in one metaphysic while rejecting the other, then it must remain suspended in the middle, ‘accepting’ or ‘rejecting’ neither. The interpretation therefore falls short of comprehending what it is to be a phenomenalist or a realist.’ — We freely admit: There is no dialectical reason for ‘choosing’ a particular metaphysic. The real meaning of the illusion of detachment/representation is the antinomy of phenomenalism and realism (29/291-3). However, it does not follow from the fact that the dialectic does not explain why one of the elements in the antinomy manifests itself an any particular occasion, that metaphysical explanation ‘falls short’ in any way. Anyone of a number of irrelevant, psychological reasons might push one to one side rather than the other; just as one’s personal psychology might block one’s ability to see either illusion. A person remains in the grip of a particular illusion only for so long as he fails to grasp its ‘real meaning’, fails to see what the subject who works through the dialectic is brought to see. For that reflective subject, there exists only the antinomy.


322. Transcendent metaphysics, the illusion of detachment is the illusion of representation, is the antinomy of the ego and truth illusions. However, one may still feel that, by comparison with other forms of knowledge, this ‘metaphysical proposition’ falls short of certainty. To assert that there is ‘no other alternative’ besides just those two illusions means only that they are the only possibilities which fit our dialectical reconstruction. But perhaps there exists a more complete dialectic, allowing ‘possibilities’ which we have overlooked. Now the right response to that doubt is to assert that the single-minded pursuit of the ideal of ‘certainty’ is shallow and barren; that we should resist the temptation to play the ‘comparisons’ game. But that is no excuse for refusing to submit the argument to the closest possible scrutiny; to attempt to show up any shortcomings or inconsistencies by means of test cases. Metaphysical ‘insight’ cannot claim incorrigibility without relinquishing its claim to knowledge (29/286). The argument of this chapter indeed provides the opportunity for such a test case: If detachment is equivalent to the disjunction of the ego and truth illusions, then all that detachment entails must be entailed by each disjunct. Detachment entails the unreality of time (30/304); so indeed does the ego illusion (28/283). How do things stand in the metaphysic of realism?


323. Thoughts detached not only from the world but also from the thinking subject, thoughts which ‘think themselves’ (320), are, the realist would like to say, ‘essentially God’s own thoughts’ (23/236): The finite subject disappears, only to be replaced by an infinite subject. Not surprisingly, when pressed to explain what it means for an undecidable question to ‘have’ an answer, the realist asserts that ‘God knows the answer’ (25/247-8, 251-2). Let us first suppose that this explanation suffices. One possible conception of God is that of a being who lives in the ‘timeless present’ of eternity, viewing our temporal existence from ‘outside’ the series of temporally ordered events. And according to that conception, time would be ultimately ‘unreal’ (Dummett ‘The Reality of the Past’ P.369-70). But in the present dialectical, context, such a conception would not have been required: The omniscient being whose existence guarantees that the answer to every question concerning the past is known by someone need only be an ‘ideal observer’ who witnesses every event and whose knowledge of the past derives from memory (28/281). However, the supposition that to posit an omniscient being, whether temporal or timeless serves to ‘explain’ what the realist ‘means’ proves false (25/253-4). For the realist’s view of the nature of time we must return to the simple assertion: ‘The question has an answer’. The realist now realizes that ‘has’ cannot refer to the knowledge of an omniscient being. Now only if, per impossibile, it could have so referred, and provided that the omniscient being was a temporal being, the realist’s ‘has’ would have had a temporal meaning. That is indeed the only meaning of ‘having an answer’ which we understand. For the only point in supplementing the assertion that a proposition concerning the past is ‘either true or false’ by the assertion that its associated question ‘has an answer’ is to convey information about our knowledge or our ability to acquire it (25/251). But the realist cannot, after all, mean ‘has an answer’ in a temporal sense: We are forced to return to the conception of a ‘timeless omniscient knowledge’; but now a knowledge not belonging to any subject. The detached thoughts which ‘think themselves’, and indeed ‘know’ their own truth or falsity, stand outside of time; and the reality which they think about is correspondingly a timeless reality.


33. The source of the illusion


324. The illusion of detachment displays the ultimate meaning of the antinomy of the truth and ego illusions; with that, the dialectic comes to an end. Nothing can be more fundamental to the illusion than the metaphysical attitude itself; if we attempt to press the inquiry further, it will have ceased to address the metaphysical attitude, cease to be a ‘dialectic of metaphysical illusion’. That does not mean that the work of metaphysics is finished; one continues to strive for greater clarity; moreover, each person must begin the task anew, for metaphysics belongs only to those who are willing to undertake the struggle. But there remains a sense in which one may speak of the end of metaphysics; with the completion of the dialectic, the motivation for the metaphysical project is shown to be nothing but illusion, the illusion which ‘metaphysics’ itself is. And with that realization comes despair. What metaphysics promised, it cannot supply: ‘ultimate reality’ is nothing but the knowledge that there is no ultimate reality, that even the ‘desire’ to know ultimate reality is not a coherent desire. For consolation, there is only the dialectic; to be endlessly ‘worked through’ until every grain of meaning has been extracted from it; a record played over and over again, while each time one strains to hear something beyond the words and the melody; and each time, with the last fading chord, one admits the same disappointment.


325. The world in its totality cannot be made an object of thought. It is not an ‘object’, not something concerning which it makes sense to ask: ‘Why does it exist?’ Yet that is not the ‘reason’ why the world exists, why there is ‘something rather than nothing’ — as if one thought one could prove that the world ‘necessarily exists’ — but only a rejection of the question. The world is open-ended, not a ‘totality’: One cannot stand back from the world; while within the world nothing could count as defining its ‘limits’. Moreover, that ‘open-endedness’ is, the reality of time. For the world does not stand still; it is the flow of action, a lived world. The thinking subject who ‘comprehends’ that world is an agent, a living person whose every action further strengthens the bonds of attachment. — For those who take comfort in metaphysical visions, let that be their ‘metaphysical vision’.


326. The description of our metaphysical vision is, after all, nothing more than a heuristic device for recalling what has been learned from the dialectic; and now that dialectic is ‘finished’. Yet there remains something we still wish to know: The ‘refutation’ of metaphysical illusion serves to identify the illusion in our own selves, to know that it is an illusion and how its illusory nature motivates the things we say and the projects we undertake. Let that be seen with complete clarity, let the vision be articulated in the finest detail; still, the illusion itself remains. Knowing the illusion does not make the illusion go away. That is just the same as to say: There is no satisfaction in knowing that metaphysics cannot supply what the metaphysical attitude desires, for the attitude will not allow itself to be abandoned; it continues to desire what it cannot have. — One is moved to ask whether that is indeed our fate; or whether, on the contrary, some task remains undone: a final destruction of metaphysical illusion.


327. The idea of a ‘dialectic of transcendent metaphysics’ promised to I destroy the ego and truth illusions by exposing their common source (29/288). We now know that the ‘ultimate, dialectical meaning’ of those illusions is the illusion of detachment; the metaphysical attitude itself inexorably leads to the antinomy of phenomenalism and realism. But just as the identification of the illusion of representation merely served as an ‘illuminating redescription’ of the two illusions (29/297), so too the illusion of detachment is not their source; knowing that the illusions are illusions of detachment does not make them ‘go away’. Now the illusions must have a source; there is no effect without a cause. And that source is in us; our failure to identify the source is a failure of self-knowledge. But if we do not even know what kind of thing it is how do we know what to look for?


328. We at least know what would not count as an ‘identification of the source’. A purported ‘identification’, of the form. ‘The illusions arise because we do F’, is only an ‘illuminating redescription’ if, first, the question arises: ‘Why do we do F?’; or, second, our knowledge that we do F does not provide the means for the destruction of the illusions but merely contributes to our knowledge of their illusory nature. These two negative criteria do not provide the means for locating the source of the illusion. If nothing more could be gleaned from them, we could only wait passively for new ‘identifications’ to turn up. However, the very fact that there are two criteria suggests a previously untried question: What has the one got to do with the other? Intuition suggests that an ‘identification’ would satisfy the first criterion if and only if it satisfied the second. The only raw material we have to work on is that very intuition.


329. In the case of other kinds of ‘illusion’ the two criteria pull apart: understanding why we are deceived by visual illusions makes no impression upon the phenomenological facts. The illusions retain their character; but then, unlike metaphysical illusions, they cease to affect us, do not need to be ‘eliminated’. Conversely, it is possible to eradicate another person’s superstitions without knowing their source, simply by indoctrination; all we need to ‘identify’ is what his superstitions lead him to falsely believe. That in the case of metaphysical illusion the two criteria coincide tells us something about the concept of ‘metaphysical illusion’ of which we had previously not taken notice: Originally, the term ‘illusion’ was chosen to underline the contrast with ‘mere philosophical error’; the illusion ‘forces itself upon our consciousness’; indeed, the objectivity of the dialectic presupposes that the illusion is ‘universal’ in that sense (1/6). It amounts to the same thing to say that only illusion can be expressed ‘dialectically’; the explanation of ‘mere error’ is ‘psychology, not metaphysics’ (18/174). We did not then anticipate that there would remain anything to be ‘explained’, once the dialectic had finished its work. And now the coincidence of the two criteria provides the only clue to what the missing explanation might be.


330. Metaphysics aims at complete clarity. That means discovering answers to one’s questions which do not raise further, unanswerable questions; it also means whole-heartedly believing those ‘answers’ and not resigning oneself to a perpetual conflict between the results of the investigation and one’s own ‘intuitions’. This demand is not merely an ideal which we choose to set ourselves; for ‘a ‘metaphysics’ which stops short of attempting to solve all the problems which present themselves in the course of its investigations simply risks reduplicating those very problems for which it claims a ‘solution’ in the form of an ineliminable residue of unanswered questions or unsatisfied intuitions’ (1/3). Each of our two ‘negative criteria’ follows from the demand for complete clarity. First, the source of the illusion must not be allowed to remain a mystery; its identification must satisfy the demand for a complete explanation of its ability to deceive. Second, that same identification must lead to a resolution of the conflict between the psychological influence of the illusion and the ‘results’ of the dialectic.


331. We must now face the fact that the terms of reference of that task rule out the possibility of its being carried out. For the propositions of a ‘system of metaphysics’ can serve only to refute metaphysical illusion; once one departs from that negative function there is nothing upon which to base the development of the system except the appeal to an ‘incorrigible metaphysical intuition’ (31/310). But that is just what the task of ‘identifying the source of the illusion’ would require us to do. So long as the dialectic is confined to its ‘negative function’ it can yield only ‘illuminating redescriptions’ of the illusion; we may cast those descriptions in ever more revealing forms, but the source of the illusion itself remains untouched. From the point of view of the dialectic, the fact that we remain unsatisfied with ‘mere description’, the fact that the illusion continues as a source of temptation are brute facts, beyond the scope of metaphysical inquiry.


332. The conclusion appears inescapable: Our metaphysic’ admits, after all, an ‘ineliminable residue’; completeness is unattainable. But if its ‘ideal’ is an impossible ideal then, on its own admission, the inquiry has failed. Indeed, it should never have been begun. — But now I wish to suggest an ‘overlooked possibility’ Identifying the source of the illusion is indeed a necessary task; but it is not a task for metaphysical inquiry. For its necessity belongs, not to metaphysics but to psychology. It is that necessity which differentiates the explanation of the source of ‘metaphysical illusion’ from the explanation of a ‘mere error’ (330), rather than the discipline for which the explanation is set as a necessary task.


333. Within the context of psychology, the two criteria for identifying the source of the illusion would ‘necessarily coincide’ (329) if hypotheses concerning the nature of that source resulted from psychoanalytic investigation. The criteria for their truth would then be ‘acceptance’ by the subject under analysis and ‘cure’ The question ‘why the source is of such-and-such a nature’ would no longer arise; and at the same time the illusion would be eliminated from the subject’s mental life. And now we are forced to confront the following question: Might there have existed a race of intelligent beings who simply could not be brought to see the ‘metaphysical illusions’, to understand the meaning of ‘metaphysics’, not because their preoccupations, or their culture, or even their psychological make-up blocked their ability to see the illusions but because there was nothing for them to see? While in the grip of illusion, compelled by the ‘necessity’ of the metaphysical attitude, we find this impossible to conceive. Yet the conclusion is inescapable. The questions of metaphysics are not the discovery of ‘reason itself’ but reflect only the contingencies of a specifically human psychology.


Prefatory note (1982)


‘The Metaphysics of Meaning’ is a six-fold expansion of a twelve and a half thousand-word thesis submitted for the B.Phil at Oxford in 1978. My supervisor then and now has been John McDowell, to whom I am indebted for persistent, constructive criticism. While our philosophical disagreements are really not substantial, the deceptive appearance of fundamental disagreement highlights the gulf that separates the aspirations of language philosophy from those of metaphysics; or so it seems to me. McDowell would not put it in that way.




To my mother and father.


Abstract (1982)


The ego and truth illusions belong to every person to whom the dialectic addresses itself, distinguished from mere error only by that ‘universality’; illusions of an ultimate reality of ‘metaphysical facts’ beyond the reach of language, whose exposure as illusion simultaneously rejects the project of transcendent metaphysics. The motivation for this project, the attempt to take up a ‘metaphysical attitude’ to a world viewed sub specie aeternitatis, ultimately translates into the antinomy of idealism and realism; an irresolvable conflict between two opposing conceptions of the nature of thought’s representations; a transcendental ego versus a transcendent truth. However, the illusions themselves cannot be described or communicated in any way save through the dialectic: for example, by tracing their influence in a distorted understanding of the meaning of ‘meaning’, in the ill-motivated rejection of the possibility of knowledge, and other false models of mind’s relation to reality; finally, in the inner structure of a paralysed silence striving to express the inexpressible. But for the brute ‘given’ of illusion, metaphysics would not be possible; yet accepting that given charges metaphysics with the additional, and most difficult task of self-justification: Unable to rest content merely with affirming its purely critical and negative function it must discover its own possibility through reflection on the meaning and purpose of ‘dialectic’. Meanwhile, false conceptions of the nature of metaphysics must be rejected: for example, the attempt to construct a deductive system with ‘axioms’ and ‘theorems’, or, alternatively, the search for a ‘theory’ of meaning which explains the workings of language, based upon the ‘rules’ for the ‘game’ of representing reality. But dialectic can claim no results, no established propositions; only at most a change in the inner state of the reader who has worked it through. Metaphysics indeed sets forever the same task, demanding completion yet never finished.


Writings referred to in the text


Bradley, F.H. Appearance and Reality 2nd. Edn. Oxford 1897


Davidson, D. ‘Mental Events’ (Experience and Theory Foster and Swanson Eds. Duckworth London 1970)


Dummett, M.A.E. Frege: Philosophy of Language 2nd. Edn. Duckworth 1981


_________ ‘The Philosophical Basis of Intuitionist Logic’ (Truth and Other Enigmas Duckworth 1978)


_________ ‘The Reality of the Past’ (Truth and Other Enigmas Duckworth 1978)


_________ ‘What is a Theory of Meaning?’ (Mind and Language Guttenplan Ed. Oxford 1975)


_________ ‘What is a Theory of Meaning? (II)’ (Truth and Meaning Evans and McDowell Eds. Oxford 1976)


Frege, G. ‘On Concept and Object’ (Philosophical Writings Geach and Black Eds. Blackwell Oxford 1960)


_________ ‘On sense and Reference’ (Philosophical Writings Geach and Black Eds. Blackwell Oxford 1960)


Kant, I. Critique of Pure Reason smith Tr. Macmillan London 1929


Kolakowski, L. ‘Karl Marx and the Classical Definition of Truth’ (Toward a Marxist Humanism Peel Tr. Grove Press New York 1969)


Quine, W.V. ‘Ontological Relativity’ (Ontological Relativity and Other Essays Columbia university Press New York 1969)


Strawson, P.F. Individuals Methuen London 1959


_________ The Bounds of Sense Methuen 1966


Wittgenstein, L. Philosophical Investigations 2nd. Edn. Anscombe Tr. Blackwell 1958


_________ Remarks on the Foundations of Mathematics 3rd. Edn. Anscombe Tr. Blackwell 1978


_________ Tractatus Logico-Philosophicus Pears and McGuinness Trs. Routledge London 1961