Wittgenstein on Scepticism and Certainty - page 2

Written by AC Grayling.

Strawson brackets Wittgenstein with Hume as a naturalist because of 'resemblances, even echoes' in OC, but says that Wittgenstein does not make 'explicit appeal to Nature'. As we have just seen, this is not so; the appeal is explicit enough. Strawson goes on to cite passages constituting the foundationalist component of OC1 as 'echoes' of naturalism. I think one should keep these two –isms clearly apart; they are not the same thing, and do not entail each other. The naturalistic streak in OC is not as strong as Strawson claims; it is a mere echo indeed, much muffled, as things stand, by OC2. But it suggests a genuine alternative, suitably worked up, as a way of protecting OC1 from OC2.


What explains Wittgenstein's inability to shake off OC2-type views is his muddling together contingent or empirical propositions with those he calls 'grammatical propositions'– see e.g. 57, 58, 136: Wittgenstein somewhat vaguely describes these latter as propositions which have the "peculiar logical role" of fixing the framework–giving the meaning, setting the conditions of intelligibility–for ordinary discourse; they cannot be called into doubt without thereby impugning the whole discourse for which they stand as foundational. This is the fatal flaw that generates the OC1-OC2 conflict. It is simply demonstrated: inspect 93-4, 106-111, 128-9, 143, 159, 167, 234, 273-4, 449, 505, and 614. Here are examples:

93. Everything that I have seen or heard gives me the conviction that no man has ever been far from the earth. Nothing in my picture of the world speaks in favour of the opposite.

106. If now the child insists, saying perhaps there is a way of getting [to the moon] which I don't know, etc. what reply could I make to him? ... But a child will not ordinarily stick to such a belief and will soon be convinced by what we tell him seriously.

234. I believe that I have forebears, and that every human being has them. I believe that there are various cities, and, quite generally, in the main facts of geography and history. I believe that the earth is a body on whose surface we move and that it no more suddenly disappears or the like than any other solid body ... If I wanted to doubt the existence of the earth long before my birth, I should have to doubt all sorts of things that stand fast for me (WR255).

These are offered as examples of beliefs 'standing fast', but one notices that in 93 and 106 the beliefs mentioned are contingent (true when Wittgenstein wrote them, but false if uttered now), while in 234 grammatical beliefs (everyone has forebears) and contingent ones (there are cities) are mixed together indiscriminately. There are examples of what might uncontroversially be called foundational beliefs–('there are physical objects', 51)–and when Wittgenstein addresses the problem at 318-323 ('But there is no sharp boundary between methodological propositions and propositions within a method' 318, and see 319) he does not resolve it, but turns directly to a claim about rationality that forms part of his positive account of knowledge, as if, whether or not a proposition is grammatical or contingent, its sense-giving foundational role is conferred on it by its being what 'the reasonable man believes' (323).

The rationality view is, indeed, unexceptionable, in having it that one of the marks of systematic propositions is the epistemically normative authority they exercise. Both grammatical and contingent propositions can be systematic in this way, for among the latter there can be propositions of a high degree of generality which key given areas of discourse, the sense of which presupposes the truth of the proposition: and the proposition is contingent. One can pluck from history examples of such propositions which have since been shown false, with the consequent withering of the discourse, as if its artery had been pinched closed.

But such propositions are not transcendental or grammatical. They are scepticism-rebutting only with respect to challenges to the less general propositions which assume them, and themselves lie open to sceptical challenge of that same internal variety. Their defence against it is supposed to rest on appeal to the system they belong to, that is, to genuinely grammatical propositions. But Wittgenstein at times accords them a status indistinguishable from genuinely grammatical propositions; at 136, for example, he speaks of 'a lot of empirical propositions which we affirm without special testing; propositions, that is, which have a peculiar logical role in the system of our empirical propositions.' The difficulty is clearly apparent here: for these 'special empirical propositions' turn out not to be empirical in the ordinary sense: 'We don't, for example, arrive at any of them as a result of investigation' (138). So they are a priori; and therefore to explain the sense in which they are also 'empirical' we must suppose them akin to Kant's synthetic a priori propositions. But these latter are transcendental in the way WittgensteinÕs grammatical propositions are, when he describes them with more care; and as 318-9 shows Wittgenstein is alive to the difference. So the problem remains.

I have no brief here to reconcile Wittgenstein's views in this connection; I simply point out the conflation to explain the presence in OC of OC2. The explanation is that if one includes among the foundations of the system propositions which are in fact contingent even if they have some kind of special status in their language-games, one is bound to accept that their status might change. Hence OC2; and hence the inconsistency in OC as it stands.


The well-known, and persuasive, central tenet of OC is its view that claims to knowledge only make sense where the possibility of doubt exists. Knowledge and doubt are correlative notions, and both knowledge claims and expressions of doubt get their content from their inherence in a framework of assumptions stable both for claims and challenges to them. We take from this the idea the thought that were matters otherwise we would be disabled from grasping that such-and-such a doubt relates to such-and-such a claim to know–that they compete, so to say, over the same epistemic territory. Knowing and doubting are internal to a framework (a language game, a practice), and the framework is its own court of appeal. All this depends on OC1 (and is threatened by OC2).

Many passages in OC urge this view. Among the key paragraphs are 121-3 (WR249), 317, 341-2, 354, 450, 519, and 625. Here are exemplary passages:

354. Doubting and non-doubting behaviour. There is the first only if there is the second.

450. A doubt that doubted everything would not be a doubt.

519. Doubt itself rests only on what is beyond doubt.

I take as key passages those that focus on doubt because what Wittgenstein's theory of knowledge responds to, taking its cue from Moore and via him the tradition of debate, is scepticism. He offers the other barrel of the shotgun too, in the long debate he has with himself from 483 until the end on 'I know' ('I know that my name is Ludwig Wittgenstein'). Getting his central tenet from those paragraphs requires the complete disentanglement of the contingent and grammatical levels of knowledge, which Wittgenstein here thoroughly mixes; yet the underlying work is already done by the respective components of OC1 described in I above.

As in Moore and the tradition of debate that sees scepticism as sharpening the point of epistemological concerns, the resolution of the crux about doubt yields the required account of knowledge. The thesis of OC, resting on its principal OC1 theme, is clear (and cogent). Of course it only sketches a kind of view; it amounts to recognising that theories of knowledge like, say, Kant's–framework-invoking theories–are on the right lines. Now one would like to see the hard detail of such a theory.

The role of certainty in Wittgenstein's view invites comment. A response sometimes offered to the familiar traditional Cartesian quest in epistemology is to point out that certainty is the wrong goal, because it is a psychological state one can entertain with respect to falsehoods: you can be certain that Red Rum will win next week's Derby, yet lose your shirt. One might accordingly argue that the goal should instead be knowledge, so understood that it is definitionally something more than the psychological states (believings) an epistemic subject has to be in as a necessary condition for entering the richer, truth-constrained, relation in which 'knowing' consists. However: Moore followed his predecessors in the Cartesian tradition by seeking to forge a connection between enjoying, as an epistemic subject, a particular kind of certainty, with the unsustainability of scepticism about what that attitude addresses. One can make 'being certain' the criterion of knowledge when the proposition one is certain of is entertained as such without option, that is, at risk of incoherence or loss of meaning. Wittgenstein, in his turn, follows Moore in adopting this strategy, but offers a deeper explanation of why there is no option: he in effect plays Kant to Moore's Hume.

Consider 8, 30, 42, 193, 194, 308. Wittgenstein acknowledges the difference between knowing and being certain, and offers an account of why the latter is sufficient for the former in the optionless cases: namely, that the certainty is (not identical to, but) a function of indubitability, which in turn is a function of the framework. Certainty is not identical with indubitability because it is a psychological state whereas indubitability is a property of a sense-constituting propositions of a definable class, viz. the grammatical propositions.

Note that Wittgenstein's apparent inability to hold apart genuinely grammatical and contingent propositions destabilises this thesis too, for relative indubitability will not do for certainty, as the remarks in the cited paragraphs clearly show. So this is indeed an aspect of OC in need of housekeeping.

Is there a lost opportunity in OC? Its argument is rooted in the same intuitions as the private language argument and its related rule-following considerations, in rejecting the 'I'-perspective of the Cartesian tradition, accepted without question or even awareness by Moore, in which the quest is for radical agent-certainty, without a backdrop of publicity constraints on the articulation of thoughts, and arguing in its place for a perspective which admits its debts to a 'we' perspective, in which, that is, the speaking and knowing agent is indebted for his capacities in these respects to the resources of an epistemic-linguistic community (see 440). But this makes it all the more striking that Wittgenstein does not use the private language argument against scepticism, for this argument at very least suggests that the existence of a public realm of referents is a condition of the existence of language, and hence of sceptical doubt itself being articulated. And this goes precisely in the direction he sought.


Apart from the vitiation of Wittgenstein's thesis threatened by OC2, there are other difficulties in OC, of which I here mention one: his conflation of scepticism with idealism. It is not fatal to the OC1-based account of knowledge, but that account needs to be shrived of it.

Wittgenstein identifies scepticism with idealism in 19, 24 and 37 (WR247). In 37, moreover, he shows that he takes realism to be the thesis opposed to idealism. This is an error which many besides Wittgenstein make. Realism and idealism are not opposed theses; they are not competitors for the same territory, for realism is an epistemological thesis and idealism a metaphysical one. There is no entailment from the truth of either to the negation of the other. Moreover, the chief varieties of idealism are intended to show that, in an associated epistemological sphere, scepticism gets no grip. Consider Berkeley; it was the avowed aim of his construction of an anti-realist epistemology and (what is a different and further matter) idealist metaphysics, to refute scepticism.

Idealisms form a various family of theses about the nature of reality, but it is safe to say that their characteristic common thesis is the metaphysical one that the universe is mental. Their chief historical opponent is materialism, the metaphysical thesis that the universe is material, that is, ultimately consists of 'material substance', a view that should not be confused with physicalism, which claims that the universe consists of what can be described by physics. (What can be described by physics is not only not coterminous with matter, but might well entail that there is no such thing as matter.)

Idealism is not the same thing as anti-realism. This latter is an epistemological thesis which denies that the relations between thought and its objects, perception and its targets, experience and the realms over which it ranges (these are different, though related, relations) are external or contingent relations. There are realistic forms of idealism (see, for example, Sprigge ), and there is no reason in principle why there should not be anti-realistic forms of materialism or–even more plausibly–physicalism. I take it that quantum theory under the Copenhagen interpretation is an anti-realist physicalism.

The claim that the relations between thought and its objects (etc.) is internal is far from the claim that all objects of thought are causally dependent upon thought (or, more generally, experience, or sentience) for their existence. Certain forms of idealism (for example, Berkeley's) put matters this way, and doubtless this is why some confuse idealism with anti-realism. Rather, anti-realism is at most the claim–until more is said; as to which, there can be much variety–that no complete description of either relatum can leave out mention of the other.

It is important to be clear about what this means. Realism is the view that the relation between thought and its objects is contingent or external, in the sense that description of neither relatum essentially involves reference to the other. This is the force of saying that realism asserts the independence of things from any mental or perceptual acts that might intend them. Call this the 'independence thesis'. Anti-realists argue that this thesis is incoherent. A simple way of showing why is afforded by the idiom of relations already adopted. A little reflection shows that the independence thesis, understood as the claim that the relations between thought and its objects are external, is a mistake at least for the direction object-to-thought, for any account of the content of thoughts about things, and in particular the individuation of thoughts about things, essentially involves reference to the things thought about–this is given by the least that can be said in favour of notions of 'broad content'. So realism offers us a peculiarly hybrid relation: external in the direction thought-to-things, internal in the direction things-to-thought. It is an easy step for the anti-realist to show that thought about (perception of, theories of) things is always and inescapably present in, and therefore conditions, any full account of the things thought about. The poorly-worded 'Tree Argument' in Berkeley, aimed at showing that one cannot conceive of an unconceived thing, is aimed at making just that elementary point. The best statement of such a view is afforded by the Copenhagen interpretation of quantum theory alluded to, in which descriptions of quantum phenomena are taken essentially to involve reference to observers and conditions of observation. Such a view does not constitute a claim that the phenomena are caused by observations of them. No more does anti-realism claim this. However, a moment's thought shows that if this claim–that the relations between thought and things is internal–is correct, then one needs to think again about truth, objectivity, the modalities and knowledge.

One often sees an opposition posed between realism and idealism, as if the labels marked competitors for the same terrain. As the foregoing shows, this is a surprisingly common, simple, but serious mistake.

Wittgenstein makes this mistake. But he also makes the mistake, or seems to, of confusing idealism and scepticism. This mistake stems from the crude view that idealism consists in the denial of the existence of the external world, and that this is what scepticism denies too. But as we see, idealism is the metaphysical claim that the world is ultimately mental in some sense, and scepticism is an epistemological challenge to us to justify our beliefs and our methods of acquiring them. In view of this one is sometimes puzzled as to what exactly Wittgenstein takes scepticism to be. It has already been noted that he confuses the 'grammatical' and the contingent as targets of sceptical attack; here he seems to imply that a sceptic claims something (viz. that the world is ideal). But it is obvious that scepticism had better not be an agniology. The scepticism that consists in challenges to justify our beliefs and epistemic practices, rather than claims that (weakly) we lack or (more strongly) cannot have knowledge in some domain, is the scepticism most worth addressing.

Strip away all but OC1 as characterised in section I above, and it can be seen as addressing scepticism thus conceived. So protection of the central insight of OC is possible: it requires no more than selective pressure on the 'delete' key.


OC is uncharacteristic of Wittgenstein in at least one striking way: that it is straightforward workaday philosophy of just the kind he earlier thought his views demonstrated to be fly-in-the-bottle. Perhaps this is evidence of a third turn; had Wittgenstein lived we might have seen him engaging even more with the problems of the philosophical tradition, thus tracing a journey from, first, thinking he had solved all its problems, to, secondly, articulating a different vision of how we misunderstand the workings of our language and thereby generate spurious problems, to, thirdly and finally, seeing that philosophical problems are real ones after all, amenable to investigation–and solution.

Wittgenstein makes a contribution to solving the central problem in epistemology in OC. His contribution is to insist on the internal connection between the concepts of knowing and doubting. This is useful to the work of showing that epistemic justification is provided by the conceptual scheme within which it alone gets content. The provisional character of OC leaves much hanging: OC2, the grammar-contingency matter, and the unworked conception of scepticism are examples. One of the most serious of the matters left hanging was recognised by Wittgenstein himself as such: the vague and generalised appeal to practice and a 'form of life' as the basis of the scheme, carried over from PI: 'Now I would like to regard this certainty, not as something akin to hastiness or superficiality, but as a form of life. (That is very badly expressed and probably badly thought as well)' (358). (Read this remark with e.g. 94, 105 and 162 [WR252] open before one.) But even as a provisional and sketchy view OC offers convincing support for a set of possibilities–admittedly, familiar ones–debated elsewhere in the epistemological tradition, namely, the framework-invoking or 'conceptual scheme'-invoking refutation of scepticism .