Wittgenstein’s On Certainty contains a commitment to the existence of hinges: items - of some sort - which are ‘held’ fast and about which inquiry in some sense turns. My aim is to see whether this basic idea can be combined with a plausible account of knowledge and then to see what happens to scepticism. Along the way I will discuss my differences from Norman Malcolm, Avrum Stroll, Danièle Moyal-Sharrock, Annalis Coliva, Marie McGinn, Duncan Pritchard, Crispin Wright and Michael Williams.
Now hinges might be extra-conceptual and purely animal behaviours: the Malcolm and Stroll view (Stroll thinks that Wittgenstein slowly moves to this from a more propositional view). Perhaps the animal certainty with which my cat uses her limbs is some form of hinge for her. But Stroll and Malcolm link this view to the rational standing of linguistic agents too: people with inherited ‘realistic pictures’ of the world. I follow neo-Fregean thinking here. If hinges contribute to the epistemic standing of rational agents then they must be conceptual. They must be something for the agent and that implies they are entertained as something like a Fregean sense. That is, they lie in the realm of the conceptual, the space of reasons. Thus one key claim of this book is that Wittgensteinian hinges are conceptual items standing in the space of reasons.
Against this, one might object that we not stand in any specifically epistemic relation to hinges. That denial may be claimed to be part of the point of Wittgenstein’s discussion of hinges. But if so, what is it that they do for the agent? Why are they not blankly external to him/her? Two general strategies exist in the literature to respond to this question.
First, one might claim that hinges are non-epistemic but contribute in some other way to a rational agent’s mental economy via some other sort of propositional attitude. Both Wright and Pritchard suggest this. I do not think that either Wright’s or Pritchard’s proposals seem plausible even by their own terms. Wright makes the attitude a piece of prudential reasoning, severing the connection from knowledge. Pritchard conjures up a bespoke propositional attitude (akin to Andy Egan’s conjuring up of ‘bimaginings’ in the philosophy of delusion) but a) does not specify its functional properties and b) disguises this lack behind his further construction of an ‘uber-hinge’.
Second, one might deny the conceptual or propositional form of hinges while locating them outside the merely animal. Danièle Moyal-Sharrock, Annalis Coliva (and in a sense Stroll too) do this. For Moyal-Sharrock, this depends on a semi-technical notion of propositional form. Since hinges lack ‘bipolarity’ they do not count as propositions. But their status with respect to sense is not akin to nonsense as resolute readers of Wittgenstein (e.g. Conant) would hold. Technically, they lack propositional status and hence cannot be objects of propositional attitudes or knowledge. Conveniently, however, it is possible to take a hinge and convert it into a meaning-related doppelganger so as to explain how it can seem to have an empirical sense. I do not think that there is conceptual space for this, however. Such hinges are not mere syntactic patterns. They must be patterns with some sort of use. But what use? Not an empirical use. Nor are they akin to tautologies: emptied of all specific empirical content. So what could they be?
I take it that hinges lie within the space of reasons, are conceptual and contribute to an agent’s epistemic standing even if they do not form premises for arguments to knowledge that the agent makes. Their holding might contribute to the doxastic responsibility of the agent. Contentiously, I take it that they are known even if they are not known as the result of an inference nor a direct perception. (Michael Williams is a sometimes ally of this point.) They are part of a conceptually structured world picture held in place by holistic considerations.
Wittgenstein suggests, however, that they differ in some respects from other knowledge claims. Now it may be tempting to suggest that a difference lies in the fact that knowledge claims are advanced for good enough reasons but are not certain. Coliva implies this with her description of mere criteriological ascription of knowledge on a roughly Baker and Hacker line on criteria (contrast: McDowell's rather more plausible view). This, however, seems to me to be wrong for reasons that McDowell - speaking perhaps as a late representative of Oxford Realism - has outlined especially in response to Brandom. Knowledge, too, is certain.
But the difference is suggested by something right in Marie McGinn’s and in Coliva’s rather different accounts. Certainties comprise the framework of representational techniques. Normatively, ‘This is a hand’ can serve as an instruction for the correct use of ‘hand’. The utterance is conceptual and prescriptive/normative. So far this accords with McGinn and Coliva. But, going beyond their descriptions, accord with such a prescription involves something. What? I propose that it is knowing that this is a hand. In other words, I take the idea that hinges are normative – a view suggested by both McGinn and Coliva – implies that they are known, a view contrary to McGinn and Coliva.
Now knowing that this is a hand might seem to involve two distinct achievements: grasp of the meaning of ‘hand’ and a practical recognitional grasp of hands. As Wittgenstein stresses, agreement in meaning involves agreement in judgements. This is shown in the application of ‘hand’ to the world. So knowledge of hinges is a form of tacit knowledge - knowing how to deploy a set of representational techniques - but which also carries with it ‘knowledge-that’. It enables a speaker to recognise a hand in paradigm conditions.
This rapprochement is made more plausible by my previously articulated account of tacit knowledge as conceptually structured, context specific practical knowledge. Linking the tacit to the conceptual via McDowell’s relaxed account of the conceptual helps to locate certainties within the space of reasons. It helps to show how hinges can seem akin to the animal while still being both conceptually structured and also knowledge.
Both Wittgenstein and commentators seems to make heavy weather of Moore’s ‘I know’ (see especially Coliva). One worry seems to be that Moore misrecognises knowledge as a mental state that could serve as a source of certainty. But that doesn’t seem to me to be the problem. A sincere ‘I know’ claim expresses objective certainty though it cannot guarantee it. ‘I know’ is fallible. So it can be a mental state - as Oxford Realism takes to be uncontentious - even if not a source of certainty. Nor is it a problem that Moore attempts to enumerate what he knows (even the usually reliable Williams gets caught up on this because of his own signature dish: epistemological realism). That too seems perfectly possible on a fallibilistic mental state picture. But the fallibility of knowledge avowal does not apply to knowledge itself, contra what Coliva assumes when she suggests merely a defensible criteriological relation between reasons and knowledge. Knowledge is factive and we can be both fallible about the contents of a knowledge claim and the fact that we think we know it. Too much ink is spilled worrying about the ‘I know’ while a retreat to ‘he knows’ would have been clearer. Moore's failure to say something in his strangely contextualised assertion does not carry over as to whether he actually knows there is a hand. (On this point: imagine if his defence in court had been that he did not know that he had a hand. Imagine if Wittgenstein’s pair of tree labelling philosophers claimed in court that they had not known there was a tree in front of them, in daylight, with non-defective vision. As Avner Baz stresses, sometimes knowledge is a burden that is not easily denied. That we do not know the point of a hypothetical knowledge claim does not imply that no knowledge is had.)
If hinges are known then neither Pritchard’s non-belief nor Moyal Sharrock’s non-propositional view will help defeat scepticism. Neither of their arguments could be used, for example, to block closure. So the full weight of an anti-sceptical argument will fall on whether supposedly known hinges can evade sceptical hypotheses in something like the way that rejecting a merely highest common factor picture of experience in response to the argument from illusion can undercut scepticism about perceptual knowledge. What’s needed is the idea that our epistemic standing can take in how the world is even if there is no additional assurance from outside our epistemic practices that it does. There is, however, a partial analogy between hinges as construed here and disjunctivism about perceptual knowledge. Of course, the picture offered by disjunctivism of a direct embrace of the world in object-dependent experiences or singular thoughts will not apply to all hinges. But there is an analogy in the use of samples and the role of showing in the later Wittgenstein. The representational system cannot float free of the world because its elements are made up of the world.