Quine talks about conceptual schemes; Wittgenstein refers to language-games. For Quine, a conceptual scheme includes an ontology: ‘One’s ontology is basic to the conceptual scheme by which he interprets all experiences’ (OWTIa 10). Moreover, different conceptual systems invoke different ontologies: ‘disagreement in ontology involves basic disagreement in conceptual schemes’ (ibid., 16). For Wittgenstein, different language-games involve different grammatical rules, and, accordingly, different essences, since essence is expressed by grammar (PI §371). For two or more people to disagree over the essence of something is for them to operate with different grammatical rules. For both Quine and Wittgenstein, meaning is not to be equated with naming: a word doesn’t have to be a name in order to be meaningful. In the minds of both, the confusion of meaning and naming is responsible for numerous confused ontological or metaphysical claims.
For Quine, one cannot prove that one’s own conceptual scheme is correct and that other schemes are false. The only ontological question that transcends linguistic convention is the pragmatic one: ‘how economical an ontology can we achieve and still have a language adequate to all purposes of science?’ (WPEa 68). Another way in which Quine expresses this question is to ask what the simplest conceptual scheme is into which we can fit and arrange the discordant fragments of raw experience (OWTIa 16). Wittgenstein denies that grammatical rules are accountable to reality or experience, and he cautions that we can neither justify as true nor invalidate as false any one or any set of these rules by comparing them with reality. Grammar and language, he insists, are autonomous. To many readers, he also appears to think that pragmatic questions can be raised about language. For example, is a certain language-game workable, given the general facts of nature? Can we offer reasons for a particular rule by reference to its utility relative to our needs?
How similar are the above views of Quine and Wittgenstein? Do we have interesting and important convergence of opinion here, or do we have misleading appearances of agreement that mask profound disagreements? I shall argue that the latter is the case.
The basic difference is this: Quine is an ontologist; Wittgenstein is a critic of the ontological enterprise. Quine struggles throughout his writing to identify what there is and to clarify the way in which we should go about finding out what there is. Wittgenstein maintains throughout his writings that we transcend the bounds of sense when we try to answer ontological questions—at least when we interpret and try to answer them ‘ontologically’. The ontological claims that Quine sanctions, claims such as ‘Universals exist’ or ‘Only particulars exist’ or ‘Physical objects do (or do not) exist’, are seen by Wittgenstein as highly misleading. For him, the only meaning they can have is either as grammatical statements about the meanings of words (posing confusedly as existential claims) or as confused ways of making non-ontological, low-level or trivial empirical statements. There are no ontological commitments embedded in Wittgenstein’s language-games. The only things presupposed by a language-game are grammatical rules. These govern the way we talk since they express the ways in which we define our concepts; they tell us nothing about the nature of reality.
I shall begin to flesh out and defend these claims by looking first at Quine’s notion of ontological commitment as presented in his famous essay ‘On What There Is’. He has several purposes in this essay. In the first place, he wants to show that some famous arguments designed to prove that we must make certain ontological admissions do not work. For example, we do not have to admit ‘unrealized possibles’ into our universe as a consequence of the fact that we can speak meaningfully of people and things that do not exist. Many philosophers have argued that in order for us to speak intelligibly of, say, the mythological horse Pegasus, we have to grant the reality of this non-existent horse; otherwise, they maintain, there would be nothing for the name ‘Pegasus’ to refer to and our claims about Pegasus, even our claim that he does not exist, would be meaningless. Not so, argues Quine. He invokes Russell’s theory of definite descriptions, which shows that sentences with terms such as ‘Pegasus’ can be analysed in such a fashion that these terms do not function as names and thus do not require the reality of a referent in order to be meaningful. ‘Pegasus’, he claims, is not really a name. It is a disguised description, and when we realize this we can see how ‘Pegasus does not exist’ can be true without requiring the postulation of the reality of an unrealized possible, the mythological horse Pegasus.
Quine goes on to show that other arguments designed to force us into accepting an ontology of a certain type likewise do not work. Use of a word like ‘red’ in ‘This is a red house’, ‘This is a red car’ and ‘This is a red tie’ does not force us to admit that there is a universal, redness, which is common to the house, the car and the tie. A different interpretation of this word and these sentences is possible, one which avoids this ontological commitment. To conclude that the universal redness exists because there is a red house, a red car and a red tie is to be guilty of hasty ontologizing.
Do any of our uses of words or sentences force us to accept a certain ontology? Only one, according to Quine. If we speak in such a way that our bound variables (what in ordinary speech would be our pronouns) must refer to certain entities in order for our statements using these variables to be true, then we must admit entities of this sort into our universe and thus into our ontology. Quantification forces an ontology on us. Quine’s second purpose in ‘On What There Is’, then, is to define a criterion of ontological commitment, a test for determining what ontology a person is committed to. ‘To be is to be the value of a bound variable’ is this criterion.
Quine argues that some people—realists—operate with a conceptual system that involves quantification over variables designating universals; others—nominalists—restrict the values of their bound variables to particulars or individuals. He also considers a phenomenalistic language whose bound variables designate sense experiences, and he contrasts it to a physicalist language in which only physical objects are assumed. Adherents of these various conceptual schemes would say, respectively: there are universals (universals exist), there are only particulars (only particulars exist), there are only sense experiences (only sense experiences exist), and there are physical objects (physical objects exist). For purposes of future discussion, let us call statements of this sort ontological statements.
Quine’s third goal in ‘On What There Is’ is to argue that we are not required to use our bound variables in any one or another of the above conflicting ways. We are not required to use a physicalist language involving reference to physical objects; nor are we required to use a phenomenalist language incorporating reference only to events or items of subjective experience. We can, presumably, choose what our bound variables will designate, i.e. their values, and thus we can choose our ontology. But we need not do so arbitrarily. Choosing the physicalist ontology allows us to interpret sets of our experiences as being the experiences of one, enduring object. There are definite advantages of economy in doing so. But the phenomenalist language also has its virtues. For one thing, it grounds our talk in what is most epistemically certain. So our decision to opt for the one conceptual system instead of the other can be made in the light of which advantage plays the more important role in our effort to deal with the data of raw experience. Thus our ontology can receive a pragmatic validation.
One thing that is disturbing about Quine’s reflections on ontological commitment is that although he gives us a criterion for it, he does not give us a clear picture of what an ontological statement is. What, for him, makes a statement an ontological one?1 What is the subject matter of such a statement? We can say with some confidence that for Quine an ontological statement is broadly empirical and hence, if not a part of science, at least continuous with it. Moreover, it appears to be a statement that describes or asserts the existence of abstract kinds. ‘Red exists’ is not a low-level empirical judgment such as ‘this rose is red’, although it may be the result of a inference from such low-level empirical statements (an inference, we might remember, that Quine thinks is hasty). In the end, however, Quine is unclear about what the precise character of ontological statements is.
But let us put this question aside. All Quine’s remarks suggest that whether or not we agree with an ontological statement, or whether such a statement is hasty or not, a statement like ‘There are universals’ is a meaningful existential claim. Ontological statements, whatever their other problems, are quite sensible. Accordingly, they can be debated and criticized. For Quine, the limits of sense fall on the far side of ontology.
I do not believe that this is so for Wittgenstein. Just as, in the Tractatus, his ‘formal concepts’ cannot be meaningfully employed in describing the world, so too, throughout his later philosophy, ontological categories cannot be used to make informative, nontrivial statements about reality. To my knowledge there is no one place in the later philosophy where he deals directly and at length with this issue. So it will be necessary for us to construct a Wittgensteinian position on ontology from a variety of sources in the later writings.
In Zettel §69 (as well as Philosophical Grammar 137) we find Wittgenstein discussing the same problem that concerned Quine, the problem of speaking about something that does not exist:
Socrates to Theaetetus: ‘If you have an idea, must it not be an idea of something?’—Theaetetus: ‘Necessarily.’—Socrates: ‘And if you have an idea of something mustn’t it be of something real?’—Theaetetus: ‘It seems so.’
If we put the word ‘kill’, say, in place of ‘have an idea of in this argument, then there is a rule for the use of this word: it makes no sense to say ‘I am killing something that does not exist.’ I can imagine a stag that is not there, in this meadow, but not kill one that is not there. And ‘to imagine a stag in this meadow’ means to imagine that…. But if someone says ‘In order for me to be able to imagine a stag it must after all exist in some sense’—the answer is: no, it does not have to exist in any sense. And if it should be replied: ‘But the colour brown at any rate must exist, for me to be able to have an idea of it’— then we can say: ‘The colour brown exists’ means nothing at all; except that it exists here or there as the colouring of an object, and that is not necessary in order for me to be able to imagine a brown stag.
This passage is rich with a number of familiar Wittgensteinian themes. It suggests that the argument used by Socrates and many others to prove that something must be (in some sense) in order for one to think about it is based upon a false equation of the grammar of verbs like ‘to kill’ and ‘to have an idea of. This assimilation is the result of misleading surface grammar. Beneath the surface, the grammatical rules for the use of these verbs are very different. There is a grammatical rule that disallows saying that something is killed that does not exist, but there is no such rule for ‘to have an idea of’. On the contrary: the rule governing the latter notion licenses the inference from having an idea of a stag in the meadow to having an idea that the stag is in the meadow, and it may simply be false that the stag is in the meadow. A similar rule does not govern ‘to kill’: ‘he killed the stag’ does not entail ‘he killed that the stag’, because the latter expression makes no sense. The grammar of ‘to have an idea of, like that of ‘to imagine’, equates having an idea of Pegasus with having an idea that Pegasus…which may be false and hence not require the existence of Pegasus.
So far, then, Quine and Wittgenstein seem to agree, albeit for different reasons, that ordinary speech does not force an ontology of unreal objects on us. Quine appeals to the possibility of paraphrase into language with a different logical structure; Wittgenstein appeals to the grammatical differences in our everyday language itself between verbs like ‘kill’ and those like ‘mean’, ‘imagine’, ‘think’, etc. And Wittgenstein explains the error of the Socratic argument in terms of misleading analogies of surface grammar, while Quine seems to blame it on hasty ontologizing.
For our purposes, the most important passage in the above quotation from Zettel is the last sentence, particularly the claim ‘“The colour brown exists” means nothing at all; except that it exists here or there as the colouring of an object.’ Wittgenstein makes a similar point, with an important variation, in PI §58. To the interlocutor in PI §58 who claims that one cannot say ‘Red exists’ because ‘if there were no red it could not be spoken of at all’, Wittgenstein responds:
—Better: If ‘X exists’ is meant simply to say ‘X’ has a meaning, —then it is not a proposition which treats of X, but a proposition about our use of language, that is, about the use of the word ‘X’.
He goes on to suggest that ‘Red exists’ can be taken to say that the word ‘red’ has a meaning and that ‘Red does not exist’ can be taken to say that this word has no meaning. The proposition ‘Red exists’, although it ‘looks as if it were about the colour’, is in fact about the use of a word. He ends the passage by making the same kind of statement we found in Zettel §69: ‘We quite readily say that a particular colour exists; and that is as much as to say that something exists that has that colour.’ ‘Red exists’ is a way (not inaccurate, according to Wittgenstein) of saying that some objects exist that are red.
How are we to understand these somewhat cryptic remarks? The overall message seems to be this: if we mean anything at all in saying ‘the colour brown exists’ (and we may very well fail to mean anything), then we are using this sentence either to say that certain particular objects are coloured brown or to say that the word ‘brown’ has a meaning. Prima facie, the consequent of this claim, especially its second disjunct, sounds implausible: ‘The colour brown exists’ doesn’t seem to be synonymous either with the claim that certain objects are coloured brown or with the claim that the word ‘brown’ has a meaning. The negation of the claim, i.e. ‘brown does not exist’, certainly does not appear (face Wittgenstein) to mean that the word ‘brown’ has no meaning. What is Wittgenstein up to here?
Instead of taking him to make a claim about the synonymy of certain sentences, let us take him, with respect to the first possibility (‘The colour brown exists’ means ‘Some objects are brown’), to be making a point about the criterion for the existence of a colour. The criterion for the existence of a colour, he may be saying, is the existence of an object or thing that has that colour. What if your interior designer asked you: is there any brown in your sitting room? You might answer: ‘Yes, the sofa is brown, and so is the coffee table.’ To be sure, the designer may not be interested in what is brown but only in the present colour scheme in the room—we can take an interest in colour in addition to, and even in opposition to, an interest in coloured things. Nevertheless, the existence of colour is the existence of coloured things; the existence of a particular colour is the existence of something having that colour. The designer might change the colour scheme in the room, but could do so only by changing the things or the colour of the things in it.
So, Wittgenstein may be taken to say that a claim about the existence of a colour either is, or is necessarily connected to, a claim about the existence of coloured things. But the latter are empirical assertions, not ontological ones. The designer’s question, ‘Is there any brown in your sitting room?’, can only be answered by looking to see if there are any brown objects in the room. On this reading, ‘The colour brown exists’ reduces to the unproblematic statement that some objects (somewhere) are brown. This is empirically indisputable, but trivial and wholly uninteresting.
Alternatively, Wittgenstein tells us that ‘The colour brown exists’ might be interpreted to say that the word ‘brown’ has a meaning (and ‘The colour brown does not exist’ to say that the word ‘brown’ has no meaning). These claims are also difficult to understand. But, once again, let us avoid the literal reading, which suggests that in each case the two statements are synonymous. Consider the following scenario: a psychology student studying the psychology of colour perception might ask, ‘Is there any such thing as mauve?’ Or, more ‘ontologically’, ‘Does mauve exist?’ The instructor might answer: ‘Yes, the word “mauve” refers to this colour’ (pointing to a colour on the colour wheel or chart). Or she might just say ‘Yes, this is mauve’ (again pointing). (She is most unlikely to say ‘Yes, mauve exists.’) The instructor’s response to the student amounts to giving him an (ostensive) definition. The student doesn’t know what the word ‘mauve’ means, for if he did, he would have seen the sample of the colour mauve on the chart. If the student had asked instead, ‘Is there a colour anise?’ the instructor could equally have responded ‘No, there is no colour called “anise”‘ or ‘No, “anise” is not a colour word.” Seen in this fashion, the student’s questions about the existence or non-existence of a colour become linguistic ones about the meanings of certain words. No ontological issues need to be decided in order to answer the student’s questions. Definitions are enough (or the lack thereof).
To sum up this discussion of what Wittgenstein says in Z §69 and PI §58, we see him arguing that statements having the look of ontology on them reduce in fact either to low-level or trivial empirical statements or to linguistic ones. ‘The colour red exists’ tells us nothing that we, as moderately experienced and literate observers and speakers, do not already know. What we know, however, are truisms about red objects or the meaning of the word ‘red’.
It should be noted in passing that Wittgenstein’s diagnosis of ‘The colour red exists’ blunts the argument that the Platonic Socrates gives for accepting the reality of transcendent Forms. After noting that various men and societies are just, Socrates pushes his interlocutors to grant that all these entities are just by virtue of possessing one and the same thing, justice. It is but a short distance to the claim that this thing, justice, exists or subsists in a transcendent realm. Wittgenstein in effect takes the claim that ‘Justice exists’ and reconverts it into unproblematic, and non-ontological, claims like ‘Athens is a just state’, ‘Hippias is a just man’, and so on. In this way, ascent to the Forms never gets started.
It might be thought that Wittgenstein’s argument goes too far. Surely it makes sense to say things like ‘Bengal tigers still exist in India’ or even ‘Quarks exist.’ Aren’t these counterexamples to what Wittgenstein has to say? The answer is No. In the Blue Book, Wittgenstein makes an interesting distinction between empirical kinds and grammatical kinds (BB 19)—what he also calls grammatical categories and logical forms. This distinction blunts the above criticism, for Bengal tigers and quarks are empirical kinds. The distinction also, I believe, casts additional light on claims like ‘The colour brown exists.’ Empirical kinds—for example kinds of apples— are defined by sets of properties. A Red Delicious apple has a certain colour, sweetness, hardness, etc.; a MacIntosh apple has an overlapping, but also slightly different, set of characteristics. What kinds of apples exist is an empirical matter, and so is the question of whether we have identified all of the kinds of apples: here, Wittgenstein tells us, nature provides the standard of completeness. Presumably he would allow the meaningfulness of ‘There are MacIntosh apples’ because it amounts to the empirical claim that apples with certain properties have been found in nature. In contrast, grammatical kinds are not defined in terms of properties; whether or not they ‘exist’ is not an empirical matter, and whether or not we have identified all of them is not something that nature decides.
Unfortunately, Wittgenstein doesn’t tell us anything else in this passage about grammatical kinds—he doesn’t give us any indication of their positive features. But what he says in a negative vein allows us to deduce from another passage that colours are grammatical, not empirical kinds. In Remarks on Colour he writes: ‘“The colours” are not things that have definite properties, so that one could straight off look for or imagine colours that we don’t yet know, or imagine someone who knows different ones than we do’ (33). If colours do not have definite properties, and empirical kinds do, then colours are not empirical kinds. No examination of nature, then, will reveal whether the colour brown exists; no examination of nature will reveal whether or not we have identified all of the kinds of colour; and no such examination will show whether we have got the distinctions correct (‘There are no subtle distinctions between logical forms as there are between the tastes of different kinds of apples’ (BB 19)). Questions, then, about the existence and nature of colours do not make sense—unless they are disguised questions about the existence and nature of coloured things, or about the meanings of words.
The claim that colours don’t have definite properties may strike us as a strange one. It appears to suggest that we have no way to distinguish them, which seems absurd. Wittgenstein acknowledges this, but counters:
When asked ‘what is the distinction between blue and red?’ we feel like answering: one is blue and the other red. But of course that means nothing and in reality what we’re thinking of is the distinction between the surfaces or places that have these colours. For otherwise the question makes no sense at all.
(PG 208)
We can’t distinguish blue from red by pointing out that it is blue. ‘Blue is blue’ is an empty tautology which ‘means nothing’. Blueness is not a property of blue, one that distinguishes it from red. To think meaningfully of the difference between blue and red is to think of the things (surfaces, places) that are blue as distinct from those that are red. So, once again, any comments about the grammatical kinds blue and red are, if meaningful, empirical comments about blue and red things. ‘So what I am saying means: red can’t be described’ (PG 209).
This claim may appear counter-intuitive. Don’t we say, for instance, ‘Red is a warm colour’ and ‘The red of the cushion clashes with the violet sofa’? The first of these statements, however, might plausibly be taken as a grammatical statement expressing a rule governing the way we talk about ‘red’. The second seems more like a statement about the red thing, or about the colour of a particular thing, than a statement about red itself.2 ‘Red can’t be described’ comes across as a rhetorical reminder that what we call descriptions of red are often definitional in nature or are descriptions of the specific colours of individual objects. There are no contingent statements that we would call descriptions of red itself or descriptions of its essence.
It might be asked what relevance this discussion of colour has to Quine’s views on ontological commitment. Quine’s argument in ‘On What There Is’ operates at a much higher level of generality than is occupied by the putatively ontological claim ‘The colour red exists.’ He is concerned with statements like ‘Universals exist.’ What can we get from Wittgenstein that will help us see whether this latter kind of statement is meaningful? Red, of course, is said by all the realists I know to be a universal, and hence we might generalize Wittgenstein’s response to the claim that ‘The colour red exists.’ But Wittgenstein also has some interesting things to say about more abstract issues—for instance about the abstract notion of colour. His remarks on this topic will help us get a better handle on ‘Universals exist.’
Surely if the colours—kinds of colour—are grammatical and not empirical kinds, colour is likewise. It shares this status with concepts like noise, shape and number and many others. None of these are distinguished from one another by sets of properties, and none of them must meet a standard set by nature. It follows that there can be no questions about the existence or non-existence of these abstract kinds.
These are large claims. Let us see what can be said in defence of them. Can we distinguish colours from, say, noises (‘Can anyone believe it makes sense to say “That’s not a noise, it’s a colour”‘ (PR 55))? Might we not say: ‘That’s a colour—it has a hue, and noises don’t have hues’? Is hue a property of colour, a property that noise does not have and hence a property that distinguishes colours from noises? First of all, note that ‘Noises don’t have hues’ is a meaningless statement (unless it is a grammatical one about ways in which we are not to talk about noises). We can’t conceive of noises not having hues unless it makes sense to think of them as having hues; our denial is just the denial of the statement that they have hues. But what could ‘Noises have hues’ possibly mean? It is not, to be sure, a contradiction in terms: the term ‘noise’ does not mean ‘something that doesn’t have a hue’. Nevertheless, it lacks any sensible content. But if this is so, ‘Noises don’t have hues’ is equally meaningless, and thus the contrast we have drawn between noises and colours is itself meaningless.
Is hue a property of colour? Not according to Wittgenstein. All descriptions of the properties of things must be, in his view, external descriptions (PG 207). Any description of the properties of things must be capable of falsehood, and ‘Colours have hue’ is not capable of this. Having a hue is constitutive of colour, which means that to speak at all of a colour is to speak of something that can be said to have a hue. This is a grammatical proposition about the way the term ‘colour’ is used. And similar things might be said about other alleged properties of colour, for example intensity. There are no properties of colour.
Couldn’t we distinguish colours from noises by saying that colours are seen, noises heard? Aren’t being seen and being heard external to colour and noise, and hence properties capable of distinguishing them? Once again, however, it appears not. ‘Noises are heard, not seen’, taken as a description of fact, implies that it is intelligible, just factually false, to think of noises being seen. This, however, is not the case: what would it be like to see a noise? There is no application that can be projected for this notion.
Wittgenstein even suggests that we do not put the colours together—put them into a single class and contrast them with the class of shapes or notes (or noises)—because there is a similarity among them, namely the property or properties of colour. And this leads him to raise the provocative question ‘Then might one also take red, green, and circular together?’ (Z §331), to which he gives the only consistent, if astonishing, answer ‘Why not?!’ (ibid.) The conflicting properties of colours and shapes do not prevent such a conjunction, for there are no such properties. We don’t conjoin red, green and circular—but we might (if we spoke a different language). We don’t conjoin colours and noises—but we might (if we operated with a different grammar). Wittgenstein entertains the possibility of some people having a coloured-shape language instead of one that distinguishes colours and shapes. Nature—reality, if you please— does not dictate what grammatical categories exist. Therefore there is no meaningful question of their existence.
If it makes no sense to speak of the existence of the colour red or indeed of colour (in a way other than indirectly speaking of coloured things or the meanings of words) and if similar claims can be made about the other grammatical kinds or categories we find in our language, then what sense can be made of the claim that universals exist? All the possible candidates have been removed from the scene. The mode of speech required to talk about a universal has crossed the bounds of sense.
But the class of universals is only one of the ontological kinds that Quine thinks can be assumed by a language. He maintains that a physicalist language brings with it the assumption that there are physical objects. What would Wittgenstein have to say on this count?
His message is much the same as it was regarding ‘Red exists’ and ‘There are colours’:
‘A is a physical object’ is a piece of instruction which we give only to someone who doesn’t yet understand either what ‘A’ means, or what ‘physical object’ means. Thus it is instruction about the use of words, and ‘physical object’ is a logical concept. (Like colour, quantity,…) And that is why no such proposition as: ‘There are physical objects’ can be formulated.
(OC §36)
Wittgenstein recognizes that this is a conclusion many will hesitate to draw. Surely one might conjecture that there are no physical objects (the Cartesian sceptic certainly appears to do so):
But can’t it be imagined that there should be no physical objects? I don’t know. And yet ‘There are physical objects’ is nonsense. Is it supposed to be an empirical proposition?
(OC §35)
If ‘There are physical objects’ were an empirical proposition, there would be empirically observable properties by means of which we recognized the existence of physical objects, properties the observed absence of which would indicate the non-existence of physical objects. We could, in this case, be said to know that something is a physical object or that physical objects exist. With regard to the question of the meaningfulness of talk about physical objects existing, Wittgenstein focuses, not on the question of whether physical objects have properties by virtue of which we recognize them as such, but on the question of whether it makes sense to talk of our knowing of their existence.
Later in On Certainty he raises the strange question, ‘Are we to say that the knowledge that there are physical objects comes very early or very late?’ (OC §479). I take it that by this question he wants to highlight the oddity of talking about ‘knowing of the existence of physical objects’. If we did know such a thing, we should be able to ascertain when we came to know it or learnt about it. But it makes no sense to talk of such a time. Furthermore, to utilize some of the other things he has to say in On Certainty about knowledge, if we claim to know that there are physical objects, their existence must be something about which doubt is, or was, possible (OC §121). Knowing involves the possibility of doubt that is overcome by evidence. Can we doubt the existence of physical objects? Clearly we can doubt the existence of particular ones, but can we doubt that there are any at all? Isn’t the latter doubt ‘hollow’ (OC §312)? What ‘specific grounds’ could we have for doubting the reality of physical objects in general (OC §458)? But if we cannot conceive of doubting the existence of all physical objects, neither can we speak of over-coming such doubt and thereby achieving knowledge that physical objects exist.
Wittgenstein is careful to note, however, that in instances in which it makes no sense to talk about doubting and knowing, the proper inference is not that we don’t know these things and, at best, only surmise or assume them. In such cases, the very question of epistemic status—knowledge, ignorance, surmise, assumption—is wrong-headed. We can’t say, then, that we assume there are physical objects. ‘Physical objects exist’ is not the unproven assumption behind our talk about individual physical objects, an assumption which we must accept if our more specific statements are to turn out to be true. Nor is it one that can receive any sort of validation, pragmatic or otherwise. Claims about which it makes no sense to speak of knowing can’t be surmised assumed, or validated either, for these are just different moves in the language-game of knowing. As Wittgenstein observes, ‘It is as if “I know” did not tolerate a metaphysical emphasis’ (OC §482). The epistemic language-game does not encompass metaphysics.
Wittgenstein is always insistent that, in order to understand a word or sentence, we must identify its use in a language-game. One of the tasks involved in doing so is to ascertain the purpose or point of its utterance. Claims about the existence or non-existence of certain things usually have as their point the goal of providing information to other people, which presupposes that these people are initially ignorant of the matter at hand and that the speaker communicates the information to them in order to overcome their ignorance. Lacking such a context, the act of making an assertion intended to be informative makes little if any sense. In Remarks on Colours we find Wittgenstein elaborating on such points. Consider:
168. Psychology describes the phenomena of seeing. For whom does it describe them? What ignorance can this description eliminate?
(ROC 40)
328. Could a ‘Psychology’ contain the sentence: ‘There are human beings who see’? Well, would that be false?—But to whom would this communicate anything?
(ROC 61)
These remarks raise some interesting, sceptical questions about the discipline which we call ‘psychology’, but pursuing these questions is not our task at the moment—except for the purpose of explicating the notion of someone’s making a cognitive claim. In effect, §168 asks: who needs to be informed about the phenomenon of seeing? Who is ignorant of it? Certainly not you and I, to whom the description of seeing is presumably addressed. We aren’t ignorant of what it is to see—after all, we see the words on the page of the psychologist’s treatise on seeing. In what sense, then, is this psychologist communicating anything to us? In what sense is he saying anything at all? Unless the term ‘seeing’ has been given a technical sense, and hence doesn’t mean what we ordinarily mean by it, the context for a communicative utterance has been removed. Likewise, if the psychologist makes the claim ‘There are human beings who see’, the fact that prima facie it is impossible to identify anyone ignorant of this and hence in a position to have it communicated to him calls into question the meaningfulness of the utterance. ‘The meaning of the sentence “there are humans who see”, i.e. its possible use at any rate, is not immediately clear’ (ROC 61).
What about those who are blind from birth? Can’t we tell them that there are human beings who see? Wittgenstein gives us the proper response to this question when he follows up ROC III, §328 with the following passage: ‘If we say “There are human beings who see” the question follows “And what is seeing?” And how should we answer it? By teaching the use of the word “see”?’ (ROC 61). The answer to the last question seems to be ‘Yes’, since in ROC III, §337 Wittgenstein writes ‘It is not the psychologist who teaches me the use of the word “seeing”‘ (ROC 62). I already know what seeing is, prior to being informed by the psychologist’s ‘description’, because I already know how to use the word ‘seeing’. My ‘knowledge’ in this case, however, is not a matter of having information, but of having mastery of the use of a word, mastery of a concept. Those who are blind from birth need to learn the concept of seeing. If, in teaching them this concept, we say ‘There are human beings who see’, this statement becomes a grammatical remark, part of the elucidation of a concept.
Let us try to apply these comments to the issue of the intelligibility of ontological statements like ‘There are physical objects.’ To make such a statement presupposes, Wittgenstein’s comments suggest, that there is someone who is ignorant of the matter and to whom one wishes to communicate this information. Who could it be? Not the phenomenalist philosopher. For a number of reasons, he denies that there are physical objects—so it would be disingenuous to attribute ignorance to him. And knowing his position, we could hardly be said to wish to communicate the matter to him. Who else might be in a position to receive this information? Young children? Well, it was never communicated to us when we were children, and we don’t, as a result, seem to suffer from some continuing ignorance about the matter. There is no audience for ‘There are physical objects.’ Therefore it is not a meaningful claim.
Of course, we do not come into this world understanding the notion of a physical object. We have to be taught the use of the term ‘physical object’. But it is not our local ontologist (of a physicalist bent) who teaches us the use of this term. Although it is a fairly uncommon term, it is not totally without a use. But its use primarily seems to be a contrastive one in the context of defining a term that doesn’t refer to a physical object. We might say to a child: ‘an idea is not a physical object; a hammer is a physical object’. In this instance we are defining or elucidating the meaning of ‘idea’ and, perhaps, that of ‘physical object’ as well. Once we learn the use of the term, we hardly need to be informed that there are physical objects.
It very much appears to be the case, then, that ontological statements like ‘There are physical objects’ lack the kind of context which could provide them with a meaningful, informative use. Is there any other use for them? What could it be? Possibly to teach a person that terms like ‘physical object’ have a meaning. But this is a task for a kindergarten or primary school teacher, not a professor of ontology.
The other two examples of ontological statements given by Quine, ‘Only particulars exist’ and ‘Only sense experiences exist’, can be handled without much ado simply by noting that to establish them one must demonstrate the falsity of ‘Universals exist’ and ‘Physical object exist.’ Inasmuch as we have found these latter utterances to be devoid of meaning, it follows that their denials are equally senseless.
In the above and still other ways, Wittgenstein calls into question the meaningfulness of ontological statements. What Quine assumes is the proper subject of debate—the matter of what there is (ontologically)—is for Wittgenstein a set of utterances incapable of being debated. Ontologizing is not a sometimes hasty enterprise; it is once and for all an idle one.3
1 Remarks by John Canfield led me to raise this question.
2 I am grateful to P.M.S.Hacker for suggestions along this line.
3 I wish to thank Hans-Johann Glock, P.M.S.Hacker and John Canfield for their helpful comments on earlier versions of this chapter.