The Limits of Relativism in late Wittgenstein

Bernard Harrison and Patricia Hanna

March 2009

1. Introduction1

Wittgenstein’s later philosophy is commonly taken to be “conventionalist” or “relativist”, not just in detail, but in its entire stance towards the problems of philosophy. Dummett, for example, argues in an early and influential essay that Wittgenstein, in his later philosophy of mathematics, is a “full-blooded conventionalist” who holds that “if we accept [a] proof we confer necessity on the theorem proved, [and do so by] making a new decision, and not merely [by] making explicit a decision we had already made implicitly.”2 And though this and related exegeses have occasionally been challenged3, on the whole they have held the field. This essay is a further exercise in dissent. We shall argue that the later Wittgenstein was far from espousing the global relativism commonly attributed to him. While relativistic on certain issues, the later work is realistic on others. More interestingly still, the kinds of relativism and realism at stake are, dialectically speaking, bound up with one another in such a way that one cannot have one without the other.

That being said, what we have in mind here is not, or not primarily, exegesis; and certainly not an exercise in the sort of philosophical hagiography which Wittgenstein’s work has sometimes attracted. We shall offer some arguments, which we take to be similar in content and general thrust to ones advanced by Wittgenstein, and we shall do something to support the latter claim by anchoring them at a few, widely separated, points in Wittgenstein’s texts. But that is all. The arguments offered here either stand on their own feet, or not at all. And in any case it would be impossible, in an essay of this length to accomplish the massive task of exegesis required to demonstrate them beyond all shadow of doubt to be Wittgenstein’s. As in our recent book4, we have tried to act, in pursuing this enterprise, on the injunction of Wittgenstein’s which perhaps best expresses his legacy both as a philosopher and as a man: “I should not like my writing to spare other people the trouble of thinking.”

2. Anti-Realism and Meaning

To know something to be objectively, or ‘really’ the case, is, we shall assume, to know it to be the case independently of human wishes, beliefs, conventions or cultural practices. Relativism is the denial, for some class of  “claims involving such values as truth, meaningfulness, rightness, reasonableness, appropriateness, aptness, or the like”5, that this condition can be met. 

Realism, by contrast, asserts that, on at least some levels, and for some claims, relativism fails to obtain. Putnam puts it well when he speaks of Realism as

the . . . project of representing ourselves as being “mappers” of something “language-independent6

Earlier philosophical generations would no doubt have preferred to write “mind-independent”. But there is something historically right, as well as faithful to the structure of current debate, about Putnam’s wish to place language at the centre of things. Post-Cartesian philosophy forges an essential connection between epistemology and the theory of meaning. Knowledge, realistically conceived, is taken to consist in a relationship between thought and something external to it.  Hence knowledge of meaning, under the same proviso, must equally consist in a relationship of some sort between linguistic expressions, or the corresponding mental or ‘third world’ entities (concepts, ‘meanings’, statements, propositions) and some class of entities external to both language and thought. Further, since language is the vehicle of thought, and so of any conceptually articulated body of knowledge, the possibility of Realism in epistemology appears to presuppose Realism in the theory of meaning. 

In current discussion, therefore, Realists, in the sense of people who want to find a way of regarding us as mappers of something mind-independent, tend to want a semantic theory which is similarly Realist, in the sense of treating meaning as something determined by “Reality” or “the World”, rather than by mere human convention or stipulation. A familiar locus classicus for Realism about meaning is Socrates’ suggestion, at Phaedrus 265e 1-3, that to be capable of formulating sound definitions of things is to be able to divide up nature into kinds according to joints which are natural, and not to shatter any part, as a bad butcher might. The implicit thought here is that “the world” or “Reality”, or “nature” is already, prior to the institution of any human language or conceptual scheme, conceptually structured; and that our object in constructing systems of descriptive concepts should be, as far as possible, to match the pre-existing conceptual structures displayed by the natural world.  A modern version of that thought is to be found in early Russell, for instance, in the remark that

Every proposition that we can understand must be composed wholly of constituents with which we are acquainted.7

The underlying thought here is that the meaning of a basic name is the actual sensory content which that name picks out. Thus, in becoming acquainted, through encountering a pillar-box or a tomato, with the colour red, we are becoming acquainted not merely with a certain sensory content, but with the very meaning of the term “red” in a sentence such as, e.g., “This coloured patch is red.”

Realism so defined must, it seems, run foul of any view which makes language itself an independent determinant of meaning. For in that case language itself, and the conceptual schemes which issue from it, must permanently frustrate our efforts to describe Reality as it is, or might be, in itself. For William James, pragmatism, the idea that all our thought is rooted in human practical activity, entailed, as an immediate consequence, that “the trail of the human serpent is . . . over everything.” For Putnam, as for many recent philosophers, language is the nest of the human serpent:

What I am saying, then, is that elements of what we call “language” or “mind’ penetrate so deeply into what we call “reality” that the very project of representing ourselves as being “mappers” of something “language-independent” is fatally compromised from the very start. Like Relativism, but in a different way, Realism is an impossible attempt to view the world from Nowhere.8

3. Two Types of Anti-Realism

Putnam’s widely shared sense that any hope of a Realistic account of human knowledge is dashed from the start by the impossibility of thinking other than in terms prescribed by language finds a more specific echo in two very popular current versions of anti-Realism. 

The first is Michael Dummett’s attack on the truth-Realism of Gottlob Frege. Frege held that the primary bearer of meaning is the sentence, or statement, and that the meaning of a statement is the conditions under which it is, respectively, true or false. As Dummett points out, this is a Realistic account only insofar as we can take ‘truth’ to refer to what is actually the case, independently of human knowledge or belief. But as Dummett persuasively argues, we possess no access to truth in the Realistic sense, but only to a weaker notion of warranted assertibility. 

Opposed to [the] realist account of statements in some given class is the anti-realist interpretation. According to this, the meanings of statements of the class in question are given to us, not in terms of the conditions under which these statements are true or false, conceived of as conditions which obtain or do not obtain independently of our knowledge or capacity for knowledge, but in terms of the conditions which we recognise as establishing the truth or falsity of statements of that class.9

The effect of Dummettian anti-Realism is, plainly enough, to render meaningless any claim to the effect that there exists a Reality which altogether transcends whatever means we may happen to possess for finding out about it. It is thus incompatible with the basic picture of Realism: that mind, thought and language are in touch with, and conditioned by, a world which is Real in the sense of being wholly external to, and wholly independent of, all three. 

The second form of anti-Realism we shall be discussing here is the neo- or quasi-Kantianism put forward by John McDowell in Mind and World10. McDowell, in agreement with Geach, Putnam, Goodman and others grants (against what he calls, variously, “rampant” Platonism, or “the myth of the given”) that, in Peter Geach’s trenchant phrase, “the mind makes concepts, and this concept-formation and the subsequent use of the concept never is a mere recognition or finding,”11 But, for McDowell this thought, far from resolving philosophical perplexity about the nature of conceptual thought, merely serves to sharpen it. It does so, McDowell thinks, by forcing us to take seriously Kant’s account of empirical knowledge as “a co-operation of sensibility and understanding”12 To possess a concept is, among other things, to know what warrants, grounds, gives reason for, the judgments in which a grasp of that concept is manifested. The difficulty now, for McDowell – a deeply Kantian one — is to explain how it can be possible for judgments framed in accordance with the demands of a “space of reasons” (to follow McDowell in adapting a phrase of Wilfrid Sellars’) effectively reflecting a humanly-created conceptual scheme, to adequately express the character of a natural world which is in no sense the creation of human beings. If concepts were simply implicit in, and effectively “read off from”, the deliverances of Kantian sensibility, no problem would arise. Nature would, in effect, herself have dictated to us the conceptual vocabulary in terms of which to set about elucidating her mysteries. But if the ‘myth of the given” is just that, then a gap opens between Kantian sensibility and, as Kant would have put it, the “spontaneity” of the understanding.

To close this gap, McDowell thinks, we must conceive of the relationship between sensibility and understanding “in a quite particular way: we must insist that the understanding is already inextricably implicated in the deliverances of sensibility themselves.”13 To do that we must, he thinks, supplement Kant with something like the “notion of second nature . . . all but explicit in Aristotle’s account of how ethical character is formed.”14 This means that “we have to expand nature beyond what is countenanced in a naturalism of the realm of law. But the expansion is limited by the first nature, so to speak, of human animals, and by plain facts about what happens to human animals in their upbringing.”15 The processes of human socialization, in other words, operating on the “first nature” provided by human needs and desires take together with human cognitive and motor capacities, induct the human child into a conceptual realm, a Sellersian ‘space of reasons’ which is neither separate from, nor given by, the space of sensibility, but, in effect, includes the latter space. McDowell’s answer to the question how the demands of a humanly-framed ‘space of reasons’ can hold sway over a wholly inhuman natural world is, thus, in effect, that the natural world, as we possessors of cognitive ‘second nature’ experience it, is not, after all, ‘inhuman’, but is itself, as ‘second nature’, in part a human construction.

McDowell is hence committed to the thesis that the world revealed in sensibility – at least in human sensibility – is a world partly constituted by the induction of the child, considered as “human animal” into the “space of reasons”, a process which McDowell, with a further nod to German Romanticism, terms “Bildung.” What this account of the relationship of thought to experience leaves us with, he argues, is a “naturalised Platonism”. ‘Naturalising’ Platonism in the fashion McDowell proposes permits him to reanimate a version of platonic Realism: the kind according to which concepts, far from being constructions out of some prior experiential substratum, antecedent to, and innocent of, conceptual organization, are “there to be discovered” in experience. This places him at odds with all those interpreters of Wittgenstein who attribute to him “a philosophical stance. . . in which one finds a spookiness in the very idea that requirements of reason are there to have their eyes opened to them, unless the idea can be reconstructed out of independent facts.” This, argues McDowell, “poses a philosophical task, and the thought is that Wittgenstein points to a way of executing it by appealing to social interactions, described in a way that does not presuppose the material to be reconstructed.”16 McDowell’s objection to this way of looking at things is that it simply recasts the fundamental dualisms — between norm and nature, subject and object, thought and world  — which “modern philosophy has taken itself to be called upon to bridge”.17 To have “bridged” –or transcended — them much as German Romanticism, or its English representative Samuel Taylor Coleridge, wished to do – is, for McDowell, the central achievement of the position outlined in Mind and World. “The bare idea of Bildung ensures that the autonomy of meaning is not inhuman, and that should eliminate the tendency to be spooked by the very idea of norms or demands of reason.”18

One might object , however, that, since McDowell’s naturalised Platonism is in effect a Kantianised Platonism, the Realism concerning norm and concept which McDowell regards as its main benefit is, in effect, Transcendental Realism: a Realism, in other words, whose obverse is Transcendental Idealism. Hence if Realism is, as Putnam puts it, the “project of representing ourselves as being ‘mappers’ of something ‘language-independent’, then the Realism of McDowell’s project must be in doubt, since, to the extent that McDowell’s arguments go through, it becomes impossible to make sense of the idea of human cognitive access to a pre-conceptual sensibility. 

The character of McDowell’s view as a version of Transcendental Idealism can be seen at work in what he has to say about what questions it is intelligible to raise concerning animal perception. Thomas Nagel’s question “What is it like to be a bat?” is according to McDowell not so much a question which transcends our powers of sympathetic imagination, as one which it is senseless to ask in the first place. “I am rejecting”, says McDowell, “a picture of a mere animal’s perceptual sensitivity to its environment: a picture in which the senses yield content that is less than conceptual, but already such as to represent the world.”[our italics]19 If we disallow that possibility, then, of course, we cannot raise the question of what the experience of finding its way by sonar feels like to the bat. And the effect of that move is to enthrone the human mind as the ultimate measure of Reality in precisely the way in which philosophical Idealism always has done. “It is true that when the question is what it is like to be a bat or a cat, the answer must try to characterize something on the lines of the creature’s point of view. But the accounts that I envisage do that, to the fullest extent that makes sense, without warranting the thought that in the case of bats at least there are facts that elude our understanding. [our italics]”20  

The transcendental Idealism implicit in McDowell’s conception of the understanding may make us doubt, also, his right to the terms ‘naturalized’, and naturalism; if by the latter we mean, following Quine and Dewey, the thesis that “knowledge, mind and meaning are part of the same world that they have to do with, and that they are to be studied in the same empirical spirit that animates natural science.”21 The Wittgenstein whose arguments we shall attempt, however roughly, to sketch here,  although remote from naturalist or Realist writers like Quine or Nagel in the detail of his thought, is, as we shall see, very much closer to them, in its spirit and implications, than to Idealists like McDowell or Dummett.

4. What function are “language-games” supposed to serve?

  The later Wittgenstein, as we know, cautions against regarding names as referential markers for items or aspects of the world encountered in experience, and advises instead that we understand meaning as deriving from the ‘uses’ or ‘roles’ we assign to expressions in the context of language-games. Many have taken this to constitute sufficient grounds in itself for taking the bottom-line stance of the later work to be some form of relativism or conventionalism. If the very concepts we use to describe the world are framed as much in terms of “games” and social practices of our own devising as in terms of a reality external to human concerns and devisings, must that not mean that, as Putnam puts it, “what we call ‘language’ or ‘mind’ indeed penetrates “so deeply into what we call ‘reality’” as fatally to compromise from the outset the project of realism? For that matter what real contrariety can exist between the Wittgestein for whom our conceptual scheme reflects the roles we assign to words in the context of language-games, and McDowell, for whom it reflects a process of Bildung which involves among other things, introducing the child to a range of essentially practical human concerns and responses.

But is this consensus view as solidly founded as it seems at first sight? It depends on what one is to make of the curious Wittgensteinian notion of a language-game (Sprachspiele). What “use” is, in fact, served by “assigning” an expression a “role” in a language-game. For that matter, what ‘use’ (to us) are ‘language-games’ themselves? Why do we find it necessary to “play” (if “play is the right expression) such ‘games’? Sir Peter Strawson, in one of the earliest reviews of the Philosophical Investigations, complained that a clearer account than Wittgenstein gives us, of what is to count as a ‘difference of use’ (one might equally say ‘a different language-game’) is needed “to save the whole notion from sliding into absurdity”22

Minimal acquaintance with Wittgenstein suggests that, for him, language-games establish the meaning of terms employed in them. They thus determine the content, and the ‘grammar’ of concepts. But what is it to grasp a concept? An ancient but still potent doctrine holds that to possess a concept is to possess what H.H. Price long ago called a  recognitional capacity.23 And that idea has seemed to find confirmation in Wittgenstein’s text. From various passages in which Wittgenstein introduces the term ‘criterion’, it is tempting to conclude, as Chihara and Fodor did early on, that “X is a criterion of Y in situations of type S if the very meaning or definition of ‘Y’ (or, as Wittgenstein might have put it, the ‘grammatical’ rules for the uses of ‘Y’) justify the claim that that one can recognize, see, detect or determine the applicability of ‘Y’ on the basis of X in normal situations of type S”24 On such an interpretation, Wittgenstein’s later views on meaning really are scarcely to be distinguished from Dummett’s, or for that matter from any other version of verificationism or operationalism.

Clearly, then, if we are to present Wittgenstein as any kind of Realist, we need to begin by breaking the grip of the idea that the role of the language-game, if language games are to have any connection with meaning, must be to show how something or other is to be recognised. 

Let’s begin with terms. It is a striking fact, if one infrequently noted by philosophers of language, that we very frequently credit ourselves with knowledge of the meaning of a term ‘F’ when we have not the slightest idea of how to recognise Fs. Take, for example, the term ‘potto’. The OED defines ‘potto’ as ‘A West African lemur (Perodictius potto), adding helpfully that the term is also used to refer to another type of lemur found only on the Calabar coast (Arctocibus calabarensis), and to the kinkajou. Some zoologists will be able to connect these lexicographical jottings with instantly recognisable gestalts. But such people compose a strikingly tiny minority of English speakers. Most of us, I imagine, if we happen to have come across the word ‘potto’ at all, will take it to refer to a small tropical mammal of some sort. It is deeply implausible to suppose, however, that more than a tiny fraction of persons in either group would be able to recognise a potto, even if confronted by one. Most of us, for ‘potto’ (and, actually, for a great many other words in common use) don’t even possess what Putnam calls a “stereotype”25; at least one of any specificity.

Are we to say, then, since only a very few users of the term ‘potto’ are able to recognise pottos, that only that favoured band of users know, or perhaps ‘really know’,  “what the word ‘potto’ means.” That seems implausible. It seems more plausible to say that anyone knows what ‘potto’ means who knows, merely, that ‘potto’ is the name of a kind of animal. But of what use is this exiguous scrap of lexicographical information? What, if anything, does it enable its possessor to do?

Here is a suggestion. A, whose knowledge of English does not extend to ‘potto’ sees someone in a white coat, B. who points to a small animal, and says, (1) ‘That is not a potto’. How is A to understand B’s remark? That depends, surely, on where A takes the word ‘potto’ to figure in the many distinct vocabularies, both ordinary-language and technical, included in the semantic resources of any developed natural language. Has A, in uttering (1) said ‘That is not a meat-eater’? Or perhaps, ‘That is not an animal which figures in the totemic system of the Ashanti’ (with the implication, perhaps: which is why they can eat it)? Or perhaps, ‘That is not a nocturnal hunter’? Or perhaps. . . (the possibilities are endless).

If A happens to know that ‘potto’ is the name of an animal, of course, then these possibilities do not arise for him. He knows that what B is saying in uttering (1) is that the specimen indicated is a specimen of some other kind of animal. So we can say, for a start, that what lexicographical knowledge of the form “‘F’ is a φ-name”, even when it is as exiguous as the knowledge that ‘potto’ is the name of a kind of animal, does for its possessor is to enable him to disambiguate, to rule out all but a single interpretation of the assertoric force, or content, of statements of the form ‘~F’.

That is not all it does, however. If A does not know that, in English, ‘potto”’is a name for a kind of animal, then he will be at a loss to understand, not only what B means when he says ‘That is not a potto’, but what he means when he says ‘That is a potto’. Does A mean that the animal indicated as instancing pottohood is totemic to theAshanti? That it is a carnivore? That it is a nocturnal hunter? Or what? Each of this forest of possibilities corresponds, in effect, to a certain account of what would constitute non-pottohood. What is gained by possessing the knowledge that ‘potto’ is, in English, the name of a kind of animal, is that this forest of possibilities is instantly reduced to one: by uttering the words ‘That is a potto’, B states that the creature in view is that kind of animal, with the further implication that it would be a non-potto only if it were some other kind of animal.

Knowledge of meaning – for instance, knowledge of what ‘potto’ means – has on this view nothing whatsoever to do with recognising pottos, and nothing whatsoever to do, either (here we are in agreement with Putnam), with fixing the extension of the term “potto”. The business of meaning is neither with recognition nor with extension-fixing, but with the determination of assertoric content. What A gains, as a result of knowing the meaning of ‘potto’ (knowing, i.e., that ‘potto’, in English, is the name of a kind of animal) is neither the ability to recognize pottos nor the ability to fix the extension of ‘potto’; but, rather, the ability to determine the assertoric content of statements employing the term ‘potto’. A gains that ability in virtue of having gained access to a simple rule for deriving the purely semantic implications of the denial of such a statement from those of its affirmation, and vice-versa.

A familiar philosophical tradition, which begins with Plato’s suggestion in the Phaedrus to the effect that the good framer of names is one who knows, like the good butcher, how to cleave Nature at the joints, suggests a different view. It suggests that to enlighten A concerning the meaning of ‘potto’ it is enough to pronounce the word in the presence of such an animal while performing some suitably ostensive gesture. The actual animal, on this view, is the meaning of the term. What the ostensive gesture says, in effect is: ‘This is what “potto” means’. This is the account of meaning which Wittgenstein encapsulates in an aphorism at Tractatus 3.203: ‘A name means an object. The object is its meaning”’26  It is the account which, in due course, he repudiates at Philosophical Investigations I.28-3027. In this section we have closely followed the argument of I.28-30. Wittgenstein there points out that an ostensive definition of “the number two, ‘That is called “two”’ — pointing to two nuts” can be variously interpreted. “the person one gives the definition to doesn’t know what one wants to call ‘two’; he will suppose that ‘two’ is the name given to this group of nuts.” Wittgenstein then proposes (start of section 29) that the ambiguity problem can be resolved by informing the learner that “two” is supposed to be the name of a number. And he goes on to suggest that to take that step is to tell the learner, not something about mathematics, or about anything else that might be considered an aspect of “Reality”, but rather about the workings of language. 

Perhaps you  say: two can only be ostensively defined in this way: “This number is called ‘two’ “. For the word “number” here shews what place in language, in grammar, we assign to the word. But this means that the word “number” must be explained before the ostensive definition can be understood.—The word “number” in the definition does indeed shew this place; does shew the post at which we station the word.”

We have added to this argument the suggestion, culled from other parts of Wittgenstein’s corpus which need not concern us here28, that the object of this move from world to language – for instance, from the large-as-life potto before us to “the post’ in language at which ‘we station’ the word ‘potto’– is to make it possible for the learner – for A in our example – to determine the assertoric content of statements containing the term – ‘potto’ or ‘number’ — to be defined. In effect we have suggested that the problem addressed by Wittgenstein’s familiar remarks concerning “ostensive definition” is not that of determining the meanings, in the sense of the referents, of  terms, but rather that of determining the assertoric content (in the sense of determining what kinds of consideration are relevant to the truth or falsity of) of the sort of propositions Quine liked to call “occasion sentences”: sentences like ‘Gavagai!’, or ‘Look, there’s a potto!’, or ‘Hey! Some red!’.

If that is what Wittgenstein was talking about, though, isn’t it odd, someone might object, that he should have written (Investigations I.29) the words ‘But this means that the word “number” must be understood before the ostensive definition can be understood’. Here, as elsewhere with Wittgenstein, it is important not to be put off by natural irritation with the admittedly somewhat high-handed, teasingly gnomic style of his writing, but to read a little further. The passage continues:29

—The word “number” in the definition does indeed shew this place; does shew the post at which we station the word. And we can prevent misunderstandings by saying “This colour is called so-and-so, “This length is called so-and-so, and so on. That is to say; misunderstandings are sometimes averted in this way. But is there only one way of taking the word “colour” or “length” —Well, they just need defining. —Defining, then, by means of other words! And what about the last definition in this chain? (Do not say: “There isn’t a last definition”. That is just as if you chose to say” “there isn’t a last house in this road; one can always build an additional one”.)30

Whether the word “number is necessary in the ostensive definition depends on whether without it the other person takes the definition otherwise than I wish. And that will depend on the circumstances in which it is given, and on the person I give it to.  And how he ‘takes’ the definition is seen in the use that he makes of the word defined.

  At first sight this brief passage of discussion appears blatantly circular. It is tempting to read Wittgenstein here as denying the very possibility of explaining ‘the post at which we station’ any word in a language without presupposing some prior understanding of language, and hence some prior understanding of the post at which, in language, some word or other is stationed. Were that the only possible way of reading this passage, Wittgenstein would no doubt be the semantic nihilist which Kripke31 and others have made of him. But let us take the argument step by step. Wittgenstein is considering the suggestion that we can clarify the purport of the attempt to define ‘two’ ostensively by reference to a group of two nuts. His reaction is to reject it, on the grounds that while in many cases it may do the trick, it is not guaranteed that it will do so. It is not guaranteed because there is not “only one way of taking the word ‘colour’ or ‘length’. The problem with ‘two’, after all was that there was more than one way of taking it in the context of the attempted ostensive definition. Wittgenstein is now generalising that point: it is not a problem for ‘two’ alone, but one which arises for any word, at any level of generality. Hence, clarifying the application of one word by appealing to an antecedent understanding of the application of another is a fruitless exercise. Just as there must be a last house in a row, so defining one word in terms of another must come to an end. 

There is a way out of this apparent impasse, and it is one of which Wittgenstein, here and elsewhere, shows himself to be perfectly aware. It involves, roughly speaking, the splitting apart of two notions which, thanks to a very long philosophical tradition, have come to appear inseparable both to the vast majority of philosophers, and to a wide variety of other theorists of language: the notion of meaning and that of application. Normally, when we think of explaining the meaning of a term, we equate that with explaining its application. And we think of such explanations as proceeding either by means of the demonstrative indication of actual instances of things to which the term applies, or by means of some verbal specification of a rule – often understood as a set of sufficient and necessary conditions – for picking out such instances. Similarly, we take it for granted that A’s understanding of the meaning of a statement S must result either from A’s  being made acquainted with actual instances of sets of circumstances under which S comes out true, or from A’s coming to know some in principle verbally specifiable rule for picking out such sets of circumstances. 

We have sketched a contrary proposal. A knows the meaning of ‘potto’, according to us, just in case he/she understands the defining project of the “game”, or practice, of sorting animals into labelled sets of morphologically similar individuals, and knows that ‘potto’ is one of the verbal labels used in that game. It is possible that A may know that, yet be unable to apply the term ‘potto’ correctly, and hence unable to discriminate correctly between circumstances under which the statement ‘That is a potto’ comes out true and circumstances under which it comes out false. Nevertheless, if A understands the place of the expression ‘potto’ in the morphological-type game, he understands enough to resolve the problem raised at Investigations I.28. That is, he is in a position to correctly assess the assertoric content of B’s remark. ‘That is a potto’. And, that being so, there seems every reason to say that A understands the meaning of ‘potto’.

But, if that is the case, coming to understand the meaning of a term N is a matter of coming to understand how to engage intelligently, with others, in a practice. P, and thus to understand what sort of role linguistic expressions such as N play in that practice. Learning to engage, with others, in social practices, is, evidently, a central aspect of socialization for human beings. It is an aspect of socialization already well-advanced in pre-linguistic children. Hence we don’t need to ask, as philosophers of language, in what terms (linguistically speaking) A had the conduct of the morphological-type game explained to him. Maybe he never did have the  workings of the morphological-type game explained to him, at least not by means of verbally-formulated explications. Maybe he just picked it up. Children do ‘just pick up’ a lot of the practices current in the communities into which they are born, and we can safely leave it to the psychologists to work out precisely how they do that. 

It follows that, at Investigations I.29, Wittgenstein may not, after all, have painted himself into the corner that virtually all his interpreters have supposed him to have painted himself into. He is right, clearly, to say that, while a verbal clue – ‘this length is called so-and so’ may help to resolve misunderstandings in some case, it may not do so in others, and is in any case not necessary to understanding. It will work, indeed, if A already knows the meaning of ‘length’, but what A needs to possess, in order to know that, is not access to the meaning of some further verbal explication, of the form “‘length’ is what we call N’s” – there is, as Wittgenstein justly observes, no last house at the end of that road – but rather a grasp of how to engage in a practice. Whether he or she possesses that grasp will show itself, as Wittgenstein suggests, in ‘the use that he makes of the word defined’. If he uses ‘potto’  in further discourse to refer to individuals of the same morphological type, then he has understood the definition correctly. If, on the other hand, he uses it as a proper name for the individual designated (cf. ‘Fido’), then he has not. Which of these things, or others, he does with the word will not depend, however, on his possessing any set of verbal instructions for using it. It will depend, rather, on the nature of the practice in which he has chosen to site the word; the ‘post’ he has assigned to it, and whether that post – as evidenced by his use of the word in one or another context of discourse — is the one other speakers assign to it. 

It might be objected that what we are offering here is not so much an interpretation of Wittgenstein’s remarks at Investigations I. 28-29 as a substantial gloss upon them. That has some force, but only up to a  point. Firstly, the suggested interpretation provides a straightforward answer to Strawson’s shrewd query. Secondly, it does not add all that much to Wittgenstein’s remarks at Investigations I.29. Effectively, it fills out a few blank cheques: namely, those represented by the words ‘how’, ‘takes’, and ‘use’ in the sentence “And how he ‘takes’ the definition is seen in the use that he makes of the word defined”. Thirdly, the way in which we have filled out those blank cheques is consistent with much else in Wittgenstein’s texts: for instance with the well-known remark at On Certainty 204:

Giving grounds, however, justifying the evidence, comes to an end;—but the end is not certain propositions striking us as true, i.e, it is not a kind of seeing on our part; it is our acting, which lies at the bottom of the language-game.32

Effectively, we have suggested a way in which language-games, and with them the notion of meaning, as that is encountered in connection with natural languages, might indeed come down to, be ultimately a matter of, ‘our acting’. Current philosophical wisdom offers us two competing accounts of what it is to know the meaning of a general term — of ‘potto’, for instance – the ‘description theory’, and the theory of ‘direct reference’. Both of them, in effect, make knowing the meaning of ‘potto’ a matter of knowing – or holding true – ‘certain propositions’ about pottos. According to the description theory, to know the meaning of ‘potto’ is either to know enough about pottos to be able to recognise one, or at the very least, if we are to follow Putnam on stereotypes, to attach some descriptive content to other speakers’ talk of pottos. According to the theory of direct reference (adapting Putnam’s well-known views on  ‘water’ as an indexical expression33), to know the meaning of ‘potto’ is to possess access to a sample of the kind so designated  — to know, i.e., that Sample P is a potto-sample, and to know how to establish the existence of some sameness-relation Rs linking P to other potto-samples. Both theories make propositionally-formulable knowledge, rather than our acting, ‘what lies at the bottom of’ our grasp of meanings. And the role of the propositional knowledge in question is, in either case, to make clear the relationship of a linguistic expression, ‘potto’, to the extra-linguistic: to the zoological realm amongst whose ‘constituents’ as Russell liked to put it, pottos figure. Both theories, in short, help to forge, by assuming it, precisely the linkage between meaning and the epistemic that Wittgenstein is at pains to rupture.

Te account proposed here suggests a third option, according to which knowing  the meaning of ‘potto’ is a matter of knowing (i) how to conduct a certain social practice P (the practice of sorting individual animals into groups sharing a distinct bodily morphology, and bestowing a unique verbal label upon each such group), and (ii) that ‘potto’ is one of the verbal labels employed in the conduct of P. The second of these is a bit of propositional knowledge, something that ‘strikes us as’, or is held to be, true. But it is nor a piece of knowledge about pottos. It is a piece of knowledge about the expression ‘potto’: about the ‘post at which we station’ that expression in language. Furthermore it is a piece of information intelligible only to someone who already knows how to conduct P. So the knowledge of how to conduct P remains the primary component of semantic knowledge: remains what ‘lies at the bottom of’ what it is to know the meaning of, ‘potto’. To know that is simply to know how the word is used in the conduct of a practice. And to possess knowledge of a practice (and here we get finally to the bottom of things; reach a point, as Wittgenstein would say, at which ‘the spade turns’) is merely to have grasped the nature and point of a project, and with that to have acquired a habit of acting as other speakers act in pursuit of that project.

5. Realism and (Dummett’s ) Anti-Realism

What are the implications for current debates concerning Realism of the apparently — but, as we shall see only apparently — starkly conventionalist account of meaning at which we have arrived by glossing – but only very modestly glossing – some of what Wittgenstein had to say about language-games?

That  account is clearly incompatible with Metaphysical, or Phaedrus-style, Realism. The framer of names cannot, like Plato’s butcher, carve along the conceptual joints of the world, because the world has no conceptual joints. Concepts are linguistic constructs, and as such internal to language. 

Realism about truth is another matter. Anti-Realism about truth rests on the assumption that the notion of meaning is conceptually linked to that of recognition, and more generally, to that of the epistemic. If we have no means of recognising when p obtains and when it does not – if there is nothing, or nothing to which we enjoy, even in principle, epistemic access, that would warrant us in asserting ’p’ – then, since the “meanings of statements”, as Dummett puts it, are “given to us” only in terms of “the conditions which we recognise as establishing the truth or falsity of the statements in question”, then ‘p’, though the English words which compose it may be familiar to us and the syntactic principles upon which they are assembled in the sentence valid ones, offers us only a fictive and illusory appearance of meaning.

The account we have just developed short-circuits this argument34 As it presents things, it is not that meaning has nothing to do with truth. It has a great deal to do with the ability to assess the content of truth-claims, but virtually nothing to do with the ability to determine whether or not such claims are warranted. Knowledge of the meaning of ‘potto’, for instance, confers upon a speaker the ability to determine what is being asserted by the claim ‘That is a potto’, namely, that the subject of the claim is an animal of one of the morphological types, or species, known to English speakers as pottos. It does not, and cannot, however, confer upon him the ability to determine, by inspection of the animal indicated, whether the claim is being truly or falsely advanced. Nor is this a particularly outré, or surprising (even mildly surprising) suggestion. It is surely deeply implausible to suggest that the competences, in that regard, which distinguish a trained zoologist or naturalist, one, moreover, enjoying some familiarity with the fauna of West Africa, require no more in the way of foundation than the ability to speak and understand English. In general, the ability to recognise conditions establishing the truth or falsity of a statement S is not an ability acquired automatically in learning the meaning of “S”. It follows that the meanings of statements are not, pace Dummett, “given” in terms of the conditions that warrant the attribution of truth or falsity. The establishment of the meaning of a sentence is prior to, and independent of, the establishment of the truth or falsity of any statement made by means of it. It follows that the conventional arguments in favour of anti-Realism concerning truth are insufficient to disturb the common-sense conviction that we enjoy access to a robustly Realistic notion of truth; that is, one for which bivalence holds. Doubtless we have no means of establishing whether or not the Royal Menagerie housed in the Lion Tower of the Tower of London, and occasionally opened to the public in the reign of Queen Elizabeth I, contained a potto. Nevertheless, we know perfectly well what it means to assert that it did; and hence there seems no reason not to be Realistic about the question whether it did or not: to hold, that is to say, that there is a truth of the matter, whether or not we could ever, even in principle, come to know it.

6. Resisting Transcendentalism 

 If we reject Dummett’s style of anti-Realism on these grounds, however, don’t we immediately encounter the fundamental question which fuels McDowell’s neo-Kantianism: How can we know that judgments framed in terms of practices we invent, practices which determine the assertoric content of those judgments, are in principle capable of representing the content of a world altogether independent of our inventions and devices?

Putting the question this way seems to demand that we show, by way of an answer, some way in which our judgments, or rather the concepts in terms of which they are framed, might be said to match or to be in accord with, the nature of Reality. This is, in effect, a demand that McDowell accepts as reasonable, and to which he proposes that we should respond by conceiving of the understanding as “already inextricably implicated in the deliverances of sensibility themselves.” 

By contrast, Wittgenstein seems to regard the whole issue of accordance between our concepts and nature as beside the point. In the 80’s of the Philosophical Investigations, for instance, he embarks on what appears to be a campaign to debunk the very idea that the notion of exactness in discourse, in the sense of the fidelity to a presumed real natural order of the concepts in terms of in which we frame our judgements, is of the slightest philosophical interest. Closer inspection reveals that his target is not the ordinary notion of precision in the framing of sentences: the sort of precision that a watchful judge in a courtroom might demand of lawyers, or a watchful journal editor of contributors. Rather, it is the expressly philosophical notion of an ideal or final exactness, an exactness which would determine the meaning of a sign so precisely as to leave no space for dubiety or uncertainty at any point over its application: no possibility of a gap of any sort between the sign and what it signifies. Here are some samples, or snatches, of his discussion:

85. A rule stands there like a sign-post.—Does the sign-post leave no doubt open about the way I have to go? Does it shew which direction I am to take once I have passed it; whether along the road or cross-country? And where is it said which way I am to follow it; whether in the direction of its finger or (e.g.) in the opposite one? . . .

[87.]. . . The sign-post is in order—if, under normal circumstances, it fulfils its purpose.

88. If I tell someone “Stand roughly here”—may not this explanation work perfectly? And cannot every other one fail too?

But isn’t it an inexact explanation?—Yes; why shouldn’t we call it “inexact”? Only let us understand what “inexact” means. For it does not mean “unusable”.35

The drift of the discussion here seems to be towards the idea that it does not matter if a sign is inexact, so long as it serves some purpose, is usable. The sense of “use” operative in these passages seems also to have shifted. We seem no longer to be talking about the “use” given to a sign in the context of a language-game or practice, but rather about the “use”, or “usefulness of practices, such as setting up signposts or giving instructions concerning location, in mediating one speaker’s relationship to other speakers and to the non-human environment.36

We can perhaps shed some light on Wittgenstein’s intentions here by way of a further example; one which combines the suggestion offered at the start of Investigations 88 with Wittgenstein’s interest in measurement as one paradigm of what a language-game might be: the interest displayed, for instance, at Philosophical Grammar 88:

The role of a sentence in a calculus is its sense.

      A method of measurement – of length, for example – has exactly the same relation to the correctness of a statement of length as the sense of a statement has to its truth or falsehood.37

Suppose one of the authors, Hanna, wants to convert a storage room into an office. She is never able to find a builder’s tape when she needs one, so she uses her feet to measure the room. After successively placing her feet heel-to-toe, and counting each iteration. she writes down that the dimensions of the room are 12 Hanna-feet by 14 Hanna-feet. Going to the store to buy oak flooring she measures that out by pacing in the same way.

A friend, call her Mary, who has been following these deliberations from afar, asks her how big the new room is. Knowing that her friend’s feet are much the same size as hers, Hanna gives her the dimensions — 12X14 – in Hanna-feet. But her friend isn’t entirely satisfied. She wants to know how big the room is really. What are its exact measurements?

Notice, first of all, that Mary has no problem attaching a meaning, in the sense of an assertoric content, to the statement, “It’s ten by twelve Hanna-feet”. She knows that “Hanna-foot” is functioning here as the name of a modulus of measurement. Her problem is, rather, that she regards the Hanna-foot as an intrinsically inexact modulus of measurement, and wants to know what “12X14 Hanna-feet” comes to, in more exact terms. One obvious thought is that it is going to be rather difficult to meet this demand. Suppose Hanna measures her feet, in the shoes she wore when pacing the room and buying the flooring, with a footrule, and finds that they are 11.25 inches long. Can she simply convert measurements in Hanna-feet into measurements in ordinary American feet and inches using this formula. Well yes, but the results, given the essential uncertainties built into the method of measurement in which Hanna-feet figure as a modulus (pacing a room by placing one’s feet alternately heel-to-toe) , will hardly be any more ‘exact’, in the sense intended by Mary, than the original measurement expressed in Hanna-feet.

Suppose, then, Hanna finally obtains a builder’s tape and remeasures the room using it. Mary now objects that there are inexactitudes built into the method of measuring a room using a builder’s tape, just as there are in the method of measuring it by pacing it heel-to-toe. Wouldn’t one get a more exact result, she asks, by using her new hi-tech measuring gadget which fires a laser across the room and calculates dimensions from the length of time taken for the light to return from the opposite wall. This, she claims, will produce measurements accurate to a ten-millionth of a millimetre.

At this point. Mary has come out in her true colours as a certain type of Metaphysical Realist. She means it when she says that she wants to know the real dimensions of the room: not just how big we find the room to be, but how big it really is. But her strategy for pursuing this goal is surely at odds with the goal itself. The strategy is to replace a relatively inexact method, or technique, of measurement with a relatively more precise, or exact, technique: to replace toe-to-heel measuring with a builder’s tape, the builder’s tape with a laser device, and so on. She is, in a way, right to adopt this strategy. “Exact” and “inexact” are, in this context, predicates which attach to techniques of measurement, rather than, say, to choice of modulus, or to the degree of care with which a given act of measuring is carried out. What makes measuring with a builder’s tape more exact than measuring by Hanna-feet are the possibilities of standardization and the reduction in the number of physical applications of the measuring implement which characterise the latter case. What makes measuring with the laser device more exact is, among other things, that it altogether excludes manual manipulation. But to admit the connection between degree of exactness and choice of measuring technique is in the same breath to deny the possibility of a final, or ultimate degree of exactness which would (as Mary-style Realists think of it) capture the dimensions things really have, rather than merely the dimensions our measuring practices credit them with having. For one thing there seems no reason in principle why it should cease to be possible to invent new and more precise techniques of measurement for special scientific purposes, as physicists and others from time to time do. More fundamentally still, in admitting a conceptual connection between exactness and technique, one admits, in effect, the impossibility of giving sense to the notion of measurement except in the context of some concrete technique of measurement or other, which will necessarily display some specific type of exactness, and thus, equally necessarily, relative to other concerns and methods, of inexactness.

Moreover, it is unclear either what is gained – for Hanna’s purposes — by moving away from Hanna’s inelegant but effective technique of measuring-by-foot, or what is lost, even from the point of view of Realism, by sticking to it. Measuring by Hanna-feet serves Hanna perfectly well for her purposes. Moving to the builder’s tape or the laser device, while if might well be a justifiable move relative to other practices, remains a pointless one relative to hers. It is as Wittgenstein says at Philosophical Investigations I.88:

“Inexact” is really a reproach, and “exact” is praise. And that is to say that what is inexact attains its goal less perfectly than what is more exact. Thus the point here is what we call “the goal”.  Am I inexact when I do not give our distance from the sun to the nearest foot, or tell a joiner the width of a table to the nearest thousandth of an inch?

No single ideal of exactness has been laid down; we do not know what we should be supposed to imagine under this head—unless you yourself lay down what is to be so called. But you will find it difficult to hit upon such a convention; at least any that satisfies you.38

The dream of an ideal of exactness transcending relativity to human techniques of measurement and correlative human purposes is on this account an empty one. No ideal of exactness is to be had, other than ones that we ourselves lay down. 

This seems at first sight an anti-Realist conclusion. In certain ways it is; in others it is not. Most importantly, from the latter point of view, it suggests a way of resolving MacDowell’s transzendentale Hauptfrage without appeal to McDowell’s or any other version of Transcendental Idealism. MacDowell’s question was, in effect: How is it possible for our thinking to be “constrained from outside the conceptual sphere”39? How is it possible, in other words, for a natural phenomenon to acquire the status, the function, of a reason? What we need to see is that this is precisely the question implicitly addressed by Wittgenstein’s remark, “A method of measurement – of length, for example – has exactly the same relation to the correctness of a statement of length as the sense of a statement has to its truth or falsehood.”

Let us see how this might work. Linear measurement is simply the device of comparing the relative size two of bjects by means of repeated end-over-end applications of some third, arbitrarily-chosen object, which is designated as a permanent standard, or modulus, for such comparisons. That object may be the Standard Metre, in Paris, or some measuring tape or rod derived from it, or it may simply be Hanna’s foot. Either way, the device has many uses, in the sense of purposes: it enables us, for example, to determine whether some heavy object will pass through a doorway without going to the trouble of moving it, to compare lengths of cloth, or areas of land, and many more. 

Knowing this much about what linear measurement involves and what it is for, one knows how to assess the assertoric content of a statement of the form “X is n M’s long”, where n is a number and M the name of a specific modulus of measurement. That is, one grasps what considerations are truth-relevant to statements of that form. Knowing what considerations are relevant to the truth or falsity of a sentence often, though not always, amounts to knowing how to set about establishing the actual truth of falsity of a specific statement of that form, and so it is in this case. One does so by carrying out a measuring procedure of some sort: measuring with a yardstick, or whatever. The result obtained by such a procedure is not, however, determined by the nature of the procedure: it is determined by the nature of the object measured. In short, it is the practice of measurement itself that provides the bridge between the natural universe and McDowell’s or Sellers’ “space of reasons”. That is, its operation effectively serves to transform mere natural circumstances (in this case the brute bulk of the object measured) into truth-relevant considerations: in this case the fact that, the bulk of the object being what it is, its edge coincides with a certain gradation on the yardstick.

It is, in short, neither just some kind of happy accident, nor a deep philosophical puzzle requiring for its elucidation elaborate exercises in neo-Kantian transcendentalism, that our conceptual inventions connect with the deliverances of an obdurately extra-conceptual sensibility in such a way as to secure, for the judgements framed in terms of them, the possibility of being grounded in Reality. They could not be otherwise, given the manner in which, according to Wittgenstein, judgment is rooted in the operation of a vast range of discrete practices or “language-games”, of which linear measurement provides merely one example. To spell it out, a practice which did not, by its very nature, engage with Reality in such a way as (a) to make clear the nature of the considerations relevant to the truth and falsity of judgments framed by means of concepts defined in terms of it, and (b) afford concrete means of attaching, at least in principle, a truth-value to at least some specific judgments so framed, could not give rise to any concepts. No terms could be introduced in connection with such a practice, because none would have been assigned a use. (That, on one level, of course, is what Wittgenstein is getting at when he argues that it is not intelligible to envisage giving a meaning to a name by “associating” it with an inner sensation, when there is no external, public criterion by which not only others, but the speaker himself, could identify a recurrence of the sensation supposedly in question.)

It follows, further, that if the gap between understanding and sensibility can be bridged on this way, we have no reason to think of understanding, with its freight of conceptual distinctions, as “inextricably implicated in the deliverances of sensibility themselves”. We remain at liberty, pace McDowell, to think of ourselves as sharing with animals a sensibility which serves as the foundation for the erection of conceptual distinctions, but which is itself pre- or a-conceptual. And that in turn allows us to opt for a version of naturalism far closer to Quine’s or Dewey’s understanding of the term than to McDowell’s at times Platonizing, at times Kantianizing one. Brian Morrison has objected to McDowell’s  account of Bildung,”, on the grounds that it far too general and lacking in substantive explanatory content. How, exactly, does “language acquisition turn the ‘mere animal’ as which McDowell claims human infants are born into reasoners to whom it is second nature to apprehend a structured world”40? McDowell, of course, is prevented by the logic of his position from offering the sort of account Morrison suggests — rightly, I think – is needed. For it to be in principle possible to give such an account it would need to be possible for us, as fully conceptually equipped human observers, to possess access at the same time to the deliverances of a pre-conceptual sensibility, and that is precisely what is ruled out by McDowell’s transcendentalism.

We, however, not being burdened with McDowell’s transcendental baggage, are in a position to respond to Morrison’s demand.  Peter Geach defines  concepts as “capacities exercised in acts of judgment”41, a definition which also effectively informs  McDowell’s discussion. We have suggested, following Wittgenstein, that to possess a concept, or rather a collection of intrinsically connected concepts, understood in that sense, is to grasp the interconnected roles which the corresponding terms serve in connection with the operation of a socially devised and maintained practice serving certain functions in the life of a human community. Man the Thinker, in effect, turns out to be merely Man the Toolmaker in another guise. And we have argued that, since such a device itself serves to bridge the gap identified by McDowell, following Kant, between sensibility and the “space of reasons” (in effect, between mere natural circumstance and truth-relevant circumstance), there is no obstacle to offering a fully naturalistic account of the construction of conceptual schemes upon the foundation, and against the background, of a pre-conceptual sensibility. There is no more difficulty in giving an account of the genesis of measurement concepts, say, than there is of giving an account of the genesis of the capacity to ride a bicycle or to use a chisel. The one can be just as naturalistic, in the fully down-to-earth sense adumbrated by Quine or Dewey, as the other. It might be objected, of course, that the very idea of such an explanation conceals a hidden circularity, since on the one hand it presumes the possibility of describing the content of a pre-conceptual sensibility while at the same time ascribing to it a conceptual content as the inevitable price of describing it. But this is merely a version of Berkeley’s sophistical argument that to form an idea of an unperceived object – i.e., one of which no idea exists in any mind — it is necessary to form an idea of precisely that object. The answer to this is that while the notion of perceiving an unperceived tree involves a contradiction, the concept of an unperceived tree involves none. Similarly, while it would be contradictory to explain the formation of the concept of length in ways which presumed a sensibility already structured in terms of that concept (call it concept L), no contradiction is involved when we explain the formation of a concept in ways which require us to describe in terms of other concepts the content of the mode of sensibility on which the processes of concept L-formation operate.

7. Wittgensteinian Realism

In conclusion, let us briefly try to draw up a summary of the senses in which the philosophy of the later Wittgenstein, if his intentions were anything like those we have been ascribing to him, can be regarded as variously relativist, or constructivist, or realist in its implications. On our reading the later Wittgenstein was not a Platonic or any other kind of Realist about concepts. The content of a conceptual scheme is always, for him, relative to the character and purposes of the system of practices in relation to which the corresponding terms have been assigned specific roles. On the other hand, the later work is certainly Realistic in the negative sense of subverting the most persuasive arguments currently available for the each of the two most historically important forms of Idealism, Subjective Idealism (in the exemplary form of Michael Dummett’s anti-Realism about truth) and Transcendental Idealism (in McDowell’s version). More fundamentally still, perhaps, since it offers us a way of construing the relationship between understanding and sensibility which allows us to retain a conception of the former as constituted, in ways susceptible of fully naturalistic (i.e., in principle natural-scientific) elucidation and explanation, it allows us rational grounds for dissenting also from Putnam’s pessimism concerning the prospect of representing ourselves as “’mappers’ of something ‘language-independent’”. The sensibility which we share with animals is, on the present account, entirely language-independent. And the business of language, as of empirical investigation, is indeed to “map” it. But the mapping proceeds, not by forcing sensibility to fit the procrustean bed of present human concerns and capacities, whatever those may be, but rather by the far more difficult process of devising word-strewn practices through whose conduct sensibility becomes, as it were, enabled to reveal to us, on its own terms, potentialities implicit in it but hitherto hidden. Language functions in short, as a means through which inhuman reality can penetrate and determine our thinking concerning it; not as Idealists of all stamps imagine, vice-versa. Pace James, the trail of the human serpent is not, after all, over quite everything.

A brief postscript: the late Professor Gilbert Ryle once remarked to one of the authors that in his opinion Absolute Idealism could be viewed as a form of solace for Nineteenth Century post-Christian intellectuals, who having lost their faith, had lost with it the assurance of  a central place in the universe (or in “Reality”) for the human mind and its concerns. This is no doubt correct as far as it goes. However it does not seem to us that every kind of religious sensibility requires that assurance. What finally satisfies Job’s desire to reason with God, for instance, is God’s insistence on the radical limitations of Job’s understanding: “canst thou draw up Leviathan with an hook”, and so on. Our feeling is that we have much to learn from the Job poet. Our impression, further, is that Wittgenstein, both early and late, in effect thought so too.

NOTES

  1. The order of our names on the title page is alphabetical, and implies nothing about our relative contributions; the paper is a joint effort. The paper has benefited from correspondence with Dr. Thomas S. Huddle, of the University of Alabama School of Medicine and the Birmingham VA Medical Center, and with Dr. Alan Tapper, of the Centre for Applied Ethics and Philosophy of the Curtin University of Technology, Perth, Western Australia. Some of its central ideas were presented at the XXII World Congress of Philosophy, 2008.Thanks are due to all who have helped improve it. ↩︎
  2. Dummett, Michael, “Wittgenstein’s Philosophy of Mathematics”, in Truth and Other Enigmas, London: Duckworth (1978), 171 ↩︎
  3. For instance, by Guy Robinson, in “Following and Formalization”. MIND (1964), revised and updated in Robinson, Philosophy and Mystification, London and New York: Routledge (1998), 39-57 ↩︎
  4. Patricia Hanna and Bernard Harrison, Word and World: Practice and the Foundations of Language, Cambridge: Cambridge University Press (2004) ↩︎
  5. Michael Krausz, Introduction to Krausz, ed, Relativism: Interpretation and Confrontation, Notre Dame, Indiana: University of Notre Dame Press (1989), 1. ↩︎
  6. Hilary Putnam, Realism with a Human Face, Cambridge, Mass.: Harvard University Press (1990), 28 ↩︎
  7. Bertrand Russell, The Problems of Philosophy, London, NY, Toronto: Oxford University Press, 1912, p.58 ↩︎
  8. Hilary Putnam, Realism with a Human Face, Cambridge, Mass.: Harvard University Press (1990), 28 ↩︎
  9. Michael Dummett, “The Reality of the Past”, in Dummett, Truth and Other Enigmas, London:Duckworth (1978), 358 ↩︎
  10. John McDowell, Mind and World, Cambridge, Mass, London: Harvard University Press (1994) ↩︎
  11. Peter Geach, Mental Acts, New York, Humanities press, inc. (1957), p.40 ↩︎
  12. McDowell, op. cit., p.46 ↩︎
  13. McDowell, op,cit., p.46 ↩︎
  14. McDowell, op,cit., p.84 ↩︎
  15. McDowell, op,cit., p.109 ↩︎
  16. McDowell, op,cit., p.92 ↩︎
  17. McDowell, op,cit., p.93 ↩︎
  18. McDowell, op,cit., p.95 ↩︎
  19. McDowell, op,cit., p.121 ↩︎
  20. McDowell, op,cit., p.122 ↩︎
  21. W.V. Quine, Ontological Relativity and Other Essays, New York and London: Columbia University Press (1969)p.26 ↩︎
  22. P.F. Strawson, Review of Wittgenstein’s Philosophical  Investigations, MIND, v.LXIII (1954), 70-99 ↩︎
  23. H.H. Price, Thinking and Experience, Cambridge, Mass: Harvard University Press (1953), 35 and passim. ↩︎
  24. Charles S. Chihara and J.A. Fodor, “Operationalism and Ordinary language”, American Philosophical Quarterly, v.II (1965), 281-295 ↩︎
  25. Hilary Putnam. Mind, Language and Reality: Philosophical Papers, Volume 2, Cambridge: Cambridge University Press (1975), 249-252 ↩︎
  26. Ludwig Wittgenstein, Tractatus Logico-Philosophicaus, tr. D.F. Pears and B.F. McGuinness, London: Routledge (1961), 23 ↩︎
  27. Ludwig Wittgenstein, Philosophical Investigations, tr. G.E.M. Anscombe, Oxford: Basil Blackwell (1958), 13-15 ↩︎
  28. For an account of the sources of this move in Wittgenstein’s middle-period writings, see Bernard Harrison, “Truth, Yardsticks and Language-Games”, Philosophical Investigations, v.19, no.2, April 1996, 105-130; Bernard Harrison, “Criteria and Truth”, Midwest Studies in Philosophy, ed Peter A.French and Howard K Wettstein, v.XXIII, New Directions in Philosophy, Boston MA and Oxford UK: Blackwell Publishers (1999), 207-235. ↩︎
  29. ↩︎
  30. Wittgenstein, Philosophical Investigations, 14 ↩︎
  31. Saul A. Kripke, Wittgenstein on Rules and Private Language, Oxford: Blackwell (1982), passim. ↩︎
  32. Ludwig Wittgenstein, On Certainty, ed. G.E.M. Anscombe and G.H. von Wright, Oxford: Blackwell (1968), 28 ↩︎
  33. Hilary Putnam. Mind, Language and Reality: Philosophical Papers, Volume 2, Cambridge: Cambridge University Press (1975), 229-235 ↩︎
  34. For a related way of short-circuiting a parallel argument of Dummett’s, this time concerning proper names, see Hanna and Harrison, Word and World: Practice and the Foundations of Language, 102-104 ↩︎
  35. Wittgenstein, Philosophical Investigations, 39-41 ↩︎
  36. For a more extended discussion of these two senses of ‘use’, See Hanna and Harrison, Word and World: Practice and the Foundations of Language, 48-53 ↩︎
  37. Ludwig Wittgenstein, Philosophical Grammar, ed Rush Rhees, Oxford: Blackwell (1974), 130 ↩︎
  38. Wittgenstein, Philosophical Investigations, 42 ↩︎
  39. McDowell, op,cit., p.5 ↩︎
  40. Brian Morrison, “Mind, World and Language:McDowell and Kovesi”, Ratio (new series) XV, 3 September 2002, p.297. I am obliged to Alan Tapper for drawing this excellent essay to my attention ↩︎
  41. Geach, op.cit., 7 ↩︎