With Wittgenstein originates the idea that philosophical problems are a sort of ‘muddle’ that we fall into when we reflect upon ourselves in certain ways.1 In the course of such a reflection we, for example, make the ostensible discovery that it is impossible in principle for us to have knowledge of the external world or of the inner states of other subjects. Our mistake here, according to Wittgenstein, is not that we have come upon a false theory of the subject matter of our enquiry. Rather, we are caught in a confusion which springs from the form of our reflection whose character is not transparent to us. Wittgenstein therefore responds to the problems raised by scepticism not with a philosophical theory which would solve these problems. He rather offers a ‘therapy’ which is meant to free us from being exercised by them.2
Philosophy is therapy, not theory. This does not imply that either Wittgenstein himself or the present-day philosophers who in his wake advocate a therapeutic method in philosophy, abandon positive conceptions of knowledge. Yet it does suggest that we will understand the point of such conceptions only by attending to the therapeutic function they are intended to fulfil. We must not conceive of them as theories of knowledge which do or do not give a convincing answer to sceptical problems. Rather, their content consists in the way in which they attempt to liberate us from the persistent inclination to seek such an answer in the first place. It is unclear, however, what form a Wittgensteinian therapy of scepticism ought to assume. I think one can distinguish at least two conceptions of philosophical therapy. The first one is exemplified in the writings of, for example, John McDowell, Cora Diamond and Marie McGinn. According to this conception, a philosophical therapy seeks to dissolve scepticism: it attempts to show that the sceptic's doubt is intelligible only in the light of philosophical preconceptions which are by no means unavoidable.3 In challenging these preconceptions, it recommends to the sceptic a point of view from which her doubt no longer makes sense. The second conception is exemplified most notably in the work of Stanley Cavell. According to it, a philosophical therapy performs an operation that is properly paradoxical: it seeks to overcome doubt, not by dissolving, but by affirming it. However, as it affirms the sceptic's doubt, it reinterprets its meaning. Its goal, then, is not to show the sceptic that her doubt is meaningless, but that it has a meaning different from the one she herself ascribes to it. The sceptic's doubt contains an incontestable truth, one the sceptic herself, however, cannot comprehend.
In this chapter, I shall argue that a philosophical therapy must take the form of a paradoxical reinterpretation of the sceptic's problem. As long as the philosophical therapy is driven by the aim of dissolving the problem raised by scepticism, it cannot prevent itself from getting re-entangled in the very problem it wanted to eliminate.
The reading of Wittgenstein which sees him as offering not a refutation, but a therapy of scepticism, is not uncontested. According to an influential interpretation, Wittgenstein's notion of a criterion provides the means for rebutting sceptical doubt. Hence, in section 1, I shall first expound this allegedly Wittgensteinian anti-sceptic theory. In section 2 I then present the dissolving therapy as a critique of this anti-sceptical reading of Wittgenstein. I go on, in section 3, to show that the dissolving therapy fails. The sceptic's doubt must be affirmed, but understood differently. On the basis of this revised understanding, I offer in section 4 an account of how the sceptic is led to misinterpret her own doubt. I end with suggesting that scepticism is a possibility internal to the enterprise of philosophy (section 5).
Sceptics are concerned with the question of how we can be certain that our judgements about the world or other subjects are actually true, and thus can constitute knowledge. In the following, I will concentrate on the sceptic about other minds. Much of what I shall say, however, as I hope will be easy to see, is relevant, mutatis mutandis, to the problem raised by scepticism about the external world as well.
The Wittgensteinian anti-sceptic claims to refute scepticism about other minds by appeal to a certain notion of a criterion. The sceptical problem he takes himself to have solved can be exposed as follows:
It is clear how I know that I myself am in pain. I know it because I feel my pain. My pain is given to me in experience. But how do I know that another person is in pain? I do not feel her pain. Her pain is not given to me in experience. I must infer that she is in pain from something which is distinct from, but indicates her pain. I infer that she is in pain from the way she behaves. Her behaviour indicates that she is in pain. However, it may well be that she behaves as if she were in pain, but is not. Hence, I cannot, from her behaviour, conclude with certainty that she is in pain. It follows that I am never able to know whether another person is in pain. All I ever experience of another subject is her behaviour, and the way she behaves is only contingently connected to what I want to know, namely whether she is in pain. Nothing of what is available to me in experience (her screaming and groaning, her contorted features, her crying or calling upon me for help) affords me a secure basis for the judgement that she is actually in pain.
The anti-sceptic4 to whom the dissolving therapy responds claims that we indeed possess the kind of knowledge which according to the sceptic we cannot possess. He sets out to show that and how we are able to attain knowledge of inner states to which we have no access in experience. We are able to know that another person is in pain, he argues, because it may be that the behaviour of a person not only indicates that she is in pain, but satisfies the criteria for her being in pain. A kind of bodily comportment is a criterion for pain if and only if such comportment defines the meaning of the concept of pain. It is then part of what it means to be in pain that this state shows itself in behaviour of such and such a sort. The anti-sceptic claims that there must be behavioural criteria for the existence of inner states in this sense, because if there were not, judgements concerning inner states of other subjects would have no content. A certain kind of behaviour, if it is a criterion for an inner state, is linked to this state not contingently, but, in a term of Wittgenstein's, grammatically. The connection between a criterion and the state for which it is a criterion is a priori, and not empirical. And from this, the anti-sceptic believes, there follows an epistemological consequence: when the behaviour exhibited by a person satisfies the criteria for pain, we have a secure epistemic basis for judging that she is in pain. We can, on the basis of the satisfaction of behavioural criteria, know that she is in the corresponding inner state.
Of course, our judgements about the inner states of other subjects can be wrong. We can be wrong even if our judgement is grounded in the satisfaction of criteria, because criterial relations are essentially ‘defeasible’ (Baker 1974: 161–2). One reason why they are defeasible is that they can be feigned.5 When someone feigns pain, criteria are satisfied, but no pain exists. In general, someone's bodily comportment satisfying the criteria for the existence of a certain inner state guarantees that the state exists, only if the behaviour has an expressive character, i.e. is an expression of this state. If it is not, the criterial evidence is defeated. Now, whereas it is epistemically unproblematic to obtain knowledge as to whether a criterion is satisfied or not, it is inherently problematic to obtain knowledge as to whether it has an expressive character. We cannot exclude beyond doubt that someone, for example, is feigning pain and that therefore her behaviour does not have an expressive character (Baker 1974: 162). After all, there can be no criteria which distinguish real pain from feigned pain. If there were, an actor could feign them.6 Hence, according to the anti-sceptic, criterial relations are structurally defeasible, despite their a priori character.
The notion of a criterion as it is developed in the anti-sceptic theory is the notion of a relation of a kind of behaviour to a kind of inner state which is both a priori and defeasible. The anti-sceptic claims that scepticism is false, because inner states have outer criteria in this sense. As external behaviour and inner state are linked a priori, we can know about the state on the basis of behaviour.7 Since the link is defeasible, we nevertheless can sometimes be in error.
The anti-sceptic claims to find in Wittgenstein a notion of a criterion which supplies a refutation of the sceptic's doubt. He refers to citations such as these: ‘Grammar tells what kind of object anything is’ (PI 373), ‘Essence is expressed by grammar’ (PI 371) or ‘An “inner process” stands in need of outward criteria’ (PI 580). But it is just this anti-sceptical interpretation of the notion of a criterion that therapists think represents a fatal misreading of Wittgenstein.
Therapy for philosophy comes into play when it is realised that the anti-sceptic's answer to the sceptic does not so much solve the problem as reiterate it. The anti-sceptic claims that I can know that another person is in pain on the grounds that her behaviour satisfies the corresponding criteria. However, given that criteria are in principle defeasible, this seems to be straightforwardly false. In order to know, I must be able to rule out the possibility of error. The epistemic basis of my judgement must be such as to rule out this possibility. But even when someone's behaviour satisfies the criteria for pain, it still remains possible that she is not in pain. It remains possible, for example, that she feigns pain, and there is no information available to me which would enable me to rule out this possibility. Since criterial evidence is in principle defeasible, even if I know that criteria are satisfied I am left in an epistemic position wherein I cannot rule out that the other person is not in pain. And this means that I do not know that she is in pain.8
The anti-sceptic, in his attempt to explain how our judgements may sometimes be in error, falls back into the sceptical predicament. It turns out that criteria do not better our epistemic situation. Given that they are defeasible, it is of no consequence epistemologically that they are a priori.9
The guiding idea of the dissolving therapy is the following. The anti-sceptic is unable to refute the sceptic effectively because he shares with the sceptic a fundamental premise: our epistemic relation to other subjects is a relation to two spheres which are logically distinct from one another. The outer sphere is the domain of the body and its comportment; to it we have direct epistemic access. The inner sphere is the realm of thoughts, feelings, and sensations. Our access to this sphere is mediated: a person's bodily comportment informs us of what is going on ‘inside’ her. Hence, for anti-sceptic and sceptic alike, we gain access to someone else's inner states in two steps. In the first step, we apply predicates that refer only to the outer sphere. On that basis we then, in a second step, apply predicates referring to the inner sphere.
Dissolving therapists identify precisely this two-step model of our relation to other subjects as the real root of scepticism. In their view, the idea responsible for scepticism is not the idea that the use of predicates referring to the body furnishes an uncertain basis for the use of predicates referring to an inner state; rather, it is the idea that our use of psychological predicates is based upon the use we make of bodily predicates which, for their part, are logically independent of them. In the light of this idea, our fundamental epistemic relation to other subjects is conceived as a relation not to a human being, but to a human body. We refer to another subject fundamentally through concepts of the body. Only then do we come upon the question of what might be going on in her soul and in her mind. Yet this conception, McDowell argues, is by no means ‘compulsory’ (McDowell 1998b: 385). It is the result of a conceptual displacement in the philosophical description of our experience of other subjects. The sceptic and the anti-sceptic remove the concept of a human being from its ‘focal position’ and put in its place a concept of a human body devised by philosophy (McDowell 1998b: 384). They then attempt to clarify the concept of a human being in terms of this philosophical artifice, albeit with a predictable lack of success.
According to the diagnosis of the dissolving therapist, anti-sceptics and sceptics have fallen victim to a philosophical confusion. They fail to consider our relation to others from the point of view of a participant in a human community, and consider it instead from the viewpoint of a cosmic observer. They take on what McDowell calls a ‘sideways-on view’ or a view ‘from outside’.10 Only from this view-point does the concept of a human being appear to be a philosophical riddle; only thence would we first chance upon others as bodies about which we need ask what justifies our applying predicates to them that pertain to their inner states. Sceptics and anti-sceptics do not simply advocate a false thesis about the nature of our knowledge of other subjects. They assume a false perspective on our knowledge.
If scepticism is the expression of a false perspective upon our relations to one another, a therapy of scepticism must above all understand how one can be tempted to such a perspective. We can liberate ourselves from a temptation only if we understand its ground. McDowell believes the temptation of the viewpoint of the cosmic observer results from certain historical and social forces such as the scientism characteristic of our intellectual climate.11 These forces, McDowell believes, can exert power only so long as we have no convincing philosophical articulation of the participant's perspective. His therapy, then, consists in spelling out a conception of our knowledge of others which is implicit in the perspective of a participant in a human community. The conception is therapeutic in the sense explained above: It is not a self-standing theory of knowledge which can be considered apart from its therapeutic intention. Rather, its content resides in how it renders unattractive the distorting perspective upon ourselves which leads to scepticism.
When we articulate the perspective of a participant, we are, according to the dissolving therapist, leading to the following conception of our knowledge of others: the experiential basis for a judgement that someone is in pain is not a behaviour we comprehend through concepts pertaining solely to the body. Rather, human behaviour is marked from the start by its expressive character. When we judge that someone's behaviour satisfies the criteria for pain we do not classify her behaviour in a way which leaves it open whether it is in fact expressive of pain – this was the anti-sceptic's view; rather, we classify her behaviour as pain behaviour, and thereby recognise it as an expression of pain. The behavioural criteria for pain, then, are satisfied if and only if the behaviour in question is expressive of pain.
This entails a disjunctive conception of the experiential basis of judgements about someone's being in pain: correct and incorrect judgements do not have the same basis in experience. If the judgement is correct, then the evidence given in experience consists in the fact that the criterion for pain is satisfied – she is expressing pain. If the judgement is incorrect, it consists in the fact that the criterion only appears to be satisfied – it only looks as though she is expressing pain, but in fact she is not.
When we accept this conception of our knowledge of others, sceptical doubt is no longer intelligible. It can be framed only as long as we adhere to the idea that someone's inner life is not given in experience, and hence, must be inferred from behaviour. The dissolving therapist advances a conception according to which someone else's inner states are not beyond my experience and thus available to me only indirectly. As Wittgenstein writes with regard to someone being in fear:
It is possible to say ‘I read timidity in his face’ but at all events the timidity does not seem to be merely associated, outwardly connected, with the face; but fear is there, alive, in the features. If the features change slightly, we can speak of a change in the fear.
(PI 537)
The features in his face are not indications of his fear, but the fear itself is visible to us, it is ‘alive’ in the features. The content of my experience of another person and the content of my judgement concerning the inner state of this person are one and the same: that such and such is the case regarding him; that he is, for example, in fear.12 In this way, behavioural criteria tell us, and indeed indefeasibly, that such and such an inner state exists.
It is a consequence of this understanding of criteria that I cannot be more certain that the criteria of a certain inner state are satisfied than I am that a corresponding inner state exists. I cannot be certain of the experiential basis of the judgement that someone is in pain without being certain of that judgement itself. The observation that another person is expressing pain and the judgement that she is in pain are not logically independent of each other. I can be certain of the observation's validity only as I am certain of the judgement's validity, and vice versa. We might say that the basis for my judgement concerning another person's pain and that judgement itself form a hermeneutic circle.13
The sceptic considers our relation to other subjects from the point of view of a cosmic observer. We now see that this conception of our epistemic relation to others goes with a certain conception of the certainty which must attach to judgements if they are to constitute knowledge. The sceptic requires that a judgement concerning someone's inner states be justified by an experience whose content is logically independent of those states. The therapist dismisses this conception of certainty: the conception of certainty as a guarantee from without. Criteria provide us with certainty of a kind that does not rest upon a basis external to the judgement itself; its certainty is not warranted from without. The epistemic basis of our judgement is expressed through predicates that already refer to an inner state, and not through predicates that do not contain the idea of an inner state.
The dissolving therapist observes that the anti-sceptic's conception of criteria, far from refuting sceptical doubt, only reiterates it. Now, as we shall see, the dissolving therapist's conception of our experience of inner states as manifested in behaviour, far from dissolving scepticism, only reinstates the sceptical disquiet. In its treatment of the sceptic's doubt who seeks to establish the certainty of her judgements, the disjunctive conception of experience, Cavell claims, achieves nothing. For, according to Cavell, this conception:
can preserve the certainty of the connection between a criterion and what it is a criterion of only at the price of never knowing with certainty that the criterion is satisfied, that what it is of is there […] If I claim that X is the case on the basis of the presence of the criteria of X and ‘it turns out’ that X is not the case, then I can always say, ‘The criteria were only seemingly present,’ or ‘The criteria were only seemingly satisfied’. That something is a criterion of X is now – to appeal to an old thought – necessary because analytic, and therefore empirically empty.
(Cavell 1979: 41–2)
The disjunctive conception of experience must to the sceptic seem a failure. It appears not to respond to her problem at all. The disjunctive conception can explain why criteria are able to provide us with certainty concerning the inner states of other subjects only by simultaneously leaving the question unresolved of how we can be certain that a criterion is given to us in experience. We have no way of knowing whether the situation is such that the criteria for pain are satisfied, or whether it is such that they only appear to be satisfied. But of course, it does not matter where the uncertainty enters: in our inferring from the satisfaction of a criterion that our judgement is true, or in our initial identifying of the criterion itself. The therapy does not resolve the epistemological problem, but rather relocates it.14
The moral that Cavell draws from the failure of the disjunctive conception is that the point of philosophical therapy cannot consist in a dissolution of sceptical doubt; rather, its point will have to consist in bringing us to a proper understanding of that doubt. ‘A formidable criticism of scepticism,’ Cavell writes, ‘will have to discover and alter its understanding of itself’ (Cavell 1979: 38). The therapy, according to this understanding of it, requires a paradoxical reinterpretation of sceptical doubt. Whereas the dissolving therapy aims at leading the sceptic to a conception of our knowledge wherein her doubt no longer has a place, the reinterpreting therapy aims to lead her to a conception of our knowledge that disarms her doubt precisely by affirming it. In affirming it, however, the therapist reinterprets the content of the sceptic's discovery: the groundlessness of our judgements about the inner states of others does not show that our epistemic faculties are limited, but rather reveals that our fundamental relation to other subjects is not epistemic.
The dissolving therapist identifies a premise shared by sceptic and anti-sceptic. It is the premise that judgements about the inner states of someone else must have an epistemic basis which is logically independent of the judgement based upon it. The reinterpreting therapist, in turn, identifies a premise shared by the dissolving therapist and the anti-sceptic. It is the premise that our judgements about the inner states of someone else stand in need of an epistemic basis, and that the satisfaction of criteria affords such a basis. The anti-sceptic and the dissolving therapist agree in the view that bodily criteria are constitutive of the content of corresponding concepts of inner states. And they take it that criteria in virtue of playing this transcendental role also play an epistemic role. Criteria confer content on concepts of inner states and in doing so they provide an epistemic basis for judgements about the existence of these states. According to Cavell, it is precisely this idea, the idea that criteria are ‘the means by which the existence of something is established with certainty – in perhaps the most famous case, that the criteria of pain […] are the means by which we can know with certainty that another is in pain,’ which is the real root of the sceptic's doubt, or more specifically, of the disquiet that moves her (Cavell 1979: 6–7).
Criteria confer content on judgements about inner states of others, but they do not provide an epistemic basis for judgements about these states. In order to see that this is so, reconsider the case where criteria for pain are feigned, as happens, for instance, on stage. How do we describe an actor who, having been struck by a rapier in a duel on stage, collapses and whimpers or cries out? It is beyond question that the actor is feigning pain. According to Cavell, it is precisely this that we comprehend on the basis of the criteria for pain. The criteria we have for pain enable us to describe this scene as one in which pain (as the case may be) is being feigned. That we have to use the concept of pain whenever pain is merely feigned in order to say what it is that someone is feigning, shows wherein the content of criteria has to consist: criteria for pain guide the application of the concept of pain, be it in the context ‘pretends to be in …’ or in the context ‘is in …’. Criteria do not allow us to attain certainty that someone else does not merely feign pain, but actually is in pain. They allow us to identify what it is that someone is feigning or actually experiencing as pain, and not, say, sorrow or rage. Cavell writes:
Criteria are ‘criteria for something's being so,’ not in the sense that they tell us of a thing's existence, but of something like its identity, not of its being so, but of its being so. Criteria do not determine the certainty of statements, but the application of the concepts employed in statements.
(Cavell 1979: 45)
Behavioural criteria are constitutive of the content of judgements about inner states, not in that they allow us to know that some inner state exists, but in that they enable us to identify it properly. We can be deceived as to the existence of an inner state, not because criteria are defeasible, as the anti-sceptic believes, nor because they may appear to be satisfied when in fact they are not, as the dissolving therapist believes. We can be deceived because criteria have nothing at all to say about whether an inner state exists or not. When someone pretends pain, the criteria for pain not only appear to be satisfied. They are satisfied. Compare the case of pretending with a case where we are not certain whether the person is expressing pain, or whether she is clearing her throat. In this sort of situation, we are not certain whether the concept of pain applies; whereas in a situation in which someone is pretending pain the concept unquestionably applies. In the first case we do not know whether it is pain which is at issue, but this is quite certain in the case of someone who is feigning it. When someone is pretending pain, she gives us the criterion for using the concept of pain; and it is only because we know that it is pain that is at issue that we can raise the question whether she is feigning it.
The sceptic claims that knowing that someone is in pain requires that we have an epistemic basis in experience which guarantees the existence of pain. She discovers that behaviour cannot afford such a basis. The anti-sceptic and the dissolving therapist, on the contrary, claim that behaviour can afford such a basis, since it may satisfy grammatical criteria; criteria provide an epistemic basis for our judgements concerning someone's inner states. The reinterpreting therapist disagrees with this understanding of criteria. He claims that criteria for pain are criteria for the use of the concept of pain, but do not guarantee the existence of pain. They do not afford an epistemic basis for judgements of existence. He thus affirms the sceptic's claim that there is no epistemic basis in experience which guarantees the existence of pain.
The reinterpreting therapist affirms that we cannot know that someone else is in pain in the sense that there is no basis in experience which would guarantee the existence of pain. However, as interpreted by the therapist, this discovery takes on a different meaning. Its new meaning develops from a reflection on the being of criteria. It is clear that there must be criteria for the use of psychological concepts if our judgements about the inner states of others are to have content. But what does it mean to say that there are criteria that determine the use of our concepts? According to the reinterpreting therapist, that there are criteria for using a certain concept means that there is a practice of using this concept in judgements. Wittgenstein makes this point when he says that our judgements about the inner states of others have content in virtue of our being ‘master of a technique’ and thus in virtue of our being a participant in an accustomed use of psychological concepts (for instance, the concept of pain) (PI 199). We have mastered the use of a certain concept – we are familiar with its use – as we participate in a ‘custom’ or, as I would prefer to say, a practice.
Our judgements about pain have content in virtue of being phenomena of a practice of making such judgements. This practice, however, is not just a practice of judging that pain exists. It is essentially a practice of responding to the pain of others. This is obvious when we consider how we learn the use of the concept of pain. We learn not simply to perceive that someone is expressing pain. More fundamentally, we learn to respond to someone expressing pain. We learn, for example, that someone in pain needs help or care. We learn to feel pity for her, to console her, to give her medication, to call for a doctor, etc. We learn to use the concept of pain in learning how to respond to someone expressing pain. The phenomena of the practice of using the concept of pain comprise responsive pain behaviour as well as expressive pain behaviour. The pain behaviour constitutive of the content of judgements about pain is of two forms: behaviour on the side of the one being in pain and behaviour on the side of the one judging about the pain, i.e. behaviour expressive of pain and behaviour responsive to pain.
The capacity to use the concept of pain, then, is essentially a capacity to respond practically to the pain of another. Our practical response to someone else is the original mode in which we know that she is in pain. In its basic appearance, a judgement that someone is in pain is implicit in the taking of such a practical attitude towards the other. Independently of a practice of responding to someone in pain, the idea that someone is in pain makes no sense.
There are criteria for pain because and insofar as there is a practice of using the concept of pain. And the fundamental phenomena of this practice are cases where recognition of criteria takes on the form of behaviour responsive to pain and where satisfaction of criteria is behaviour expressive of pain. For conceiving of someone as being in pain is responding to her as to someone in pain, while conceiving of someone as pretending pain is withholding one's response to her. And someone can be said to withhold only a response of which she is capable. Equally, in order to be able to pretend pain, one must have experienced pain. And the human experience of pain includes behaviour expressing one's pain in ways to which others can respond. ‘A child’, Wittgenstein writes, ‘must have developed far before it can pretend, must have learned a law of truth before it can simulate’ (LW 2 42). Hence, the cases where recognition of criteria is responsive pain behaviour and satisfaction of criteria is expressive pain behaviour are fundamental in that without them there would be no practice of using the concept of pain, and, thus, no criteria of its use. It is through these cases that an understanding of the concept of pain is acquired and maintained. They are the normal cases, not in the sense that they are frequent statistically speaking (although they may be that, too), but in the sense that they are the cases with which someone who has mastered the practice is familiar.
The normal cases are fundamental; without them, there would be no concept of pain.15 And in these cases – in the normal cases of using the concept of pain – there is no step from identifying a state as pain to claiming that it actually exists. When, in such a case, one claims that some inner state exists, one does not take a step beyond the initial identification of this state. Such a claim is already inherent in the normal cases of the use of the corresponding concept.16 As we see what it is for there to be criteria for the use of concepts, we arrive at a conception of our knowledge which dislodges the idea that our relation to other subjects is epistemic: the capacity to know about the inner states of other subjects turns out to be a capacity primarily exercised in the taking of a practical attitude towards them. Knowing is responding.
Given this conception of knowledge, we are now in a position to reinterpret the sceptic's discovery. That criteria for pain are satisfied, I have argued, does not establish the existence of pain. It only establishes the identity of a state, pretended or real, as pain. Hence, the sceptic is right in holding that there is no behaviour which would guarantee the existence of pain. However, this implies that we can never know whether someone is in pain only if it is always necessary to take a step from identifying a state as pain to claiming that it exists. But we have seen that there are criteria for the identity of a state as pain only because there is a practice of using the concept of pain. And the fundamental phenomena of this practice, its normal cases, are such that there is no room for such a step. It follows that there is no epistemic grounding of our judgements about pain. However, this does not mean that something of the kind the sceptic is seeking is absent. To say that criteria are only criteria for identifying the inner state of another subject and not for knowing its existence is to argue that claiming its existence is not something we do in addition to identifying it. There is no epistemic grounding of our judgements about the existence of pain because there is no step to be taken from the identification of pain to its existence. As we understand what it is for there to be criteria for the identity of a state as pain, we see that being able to identify such a state cannot be dissociated from being able to know that someone is in pain.
We now see where the sceptic's basic misunderstanding lies. It lies in the way in which she conceives of the impossibility of being certain of someone's pain on the basis of her behaviour. She thinks that it expresses an incapacity of our epistemic faculties – that it has the character of an epistemic failure. But we cannot ‘know’ whether an inner state we have identified on the basis of criteria actually exists or not; not in the sense that we are condemned to fail epistemically in our relations to such states, but in the sense that our relation to other subjects' inner states is not an epistemic one at all.17 Since a case of knowing that someone is in pain is, fundamentally, a case of responding to someone in pain, our knowledge of the inner states of other subjects cannot be founded upon an epistemic basis. We cannot ‘know’ whether an inner state exists or not. But this does not represent a contentful discovery regarding the reach of our knowledge. It is expressive of the fact that knowing, in normal cases, is responding.
I have argued that the reinterpreting therapist pursues the paradoxical strategy of stripping sceptical doubt of its power precisely by showing that it is right: the sceptic is right when she discovers that we can never be ‘certain’ of the pain of other people. But this discovery does not express a surprising truth about the reach of our epistemic capacities, as the sceptic thinks. That there is no epistemic grounding of our judgements about the existence of the inner states of others does not mean that there is something that fails to be grounded, but rather, that there is nothing to be grounded. Our judgements cannot be grounded because being able to use psychological concepts at all means being familiar with normal cases of their use, where there is no step from identifying a state to claiming that it actually exists, and therefore no gap between what is directly given in experience, namely the behaviour, and the inner state of the other. In normal cases, our behaviour responsive to pain does not rest upon evidence which answers the question: is the other actually in pain? Rather, it is itself the answer to this question.
According to this diagnosis, the sceptic's doubt results from a misunderstanding of the truth she discovers. It is by no means a fabrication, but she misunderstands it. But what is the reason for this misunderstanding? Why does she think that knowing the inner states of others requires us to bridge a gap between what is directly given in our experience and what we actually want to know; namely, whether the other is really in pain?
Consider a situation of doubt. A little boy is lying in his bed and crying for help. He tells his mother that he feels sick; that his stomach, his head and just about everything else hurts. He winces and cries and tells her that he cannot get up. However, there is something the mother finds strange in his behaviour this morning. His face is neither pale nor red, his wincing sounds somehow forced, not like it normally sounds when he is in pain. And he is talking so much about his pain, something he also normally doesn't do when he is in pain – almost as if he wanted to convince the mother of his pain. The mother becomes suspicious. Maybe he is only pretending pain because he does not want to go to school today? Isn't there a maths test scheduled for today? The mother withholds her response to the boy.
In such a case of doubt, we do not respond ‘blindly’ to the other,18 but we are confronted with the question whether the pain-behaviour of the other does have an expressive character or not. Hence, a case of doubt is one in which we see our relation to the other as being marked by a gap between that which we experience, namely his pain behaviour, and the fact we wish to know, namely whether he really is in pain.
What will the mother do in such a case? She will look for criteria for pain. In a case of doubt, we bring in criteria. But given that criteria are only criteria for the identity of an inner state, and not of its existence, how can we characterise the function of our appeal to criteria in such a case? It cannot be that we are appealing to criteria in order to establish a basis from which we can securely infer that the other is in pain. In appealing to criteria, we are not appealing to something that would allow us to make an inference to bridge the gap between the behaviour of the little boy and his inner state. We are rather looking for something that will make the gap disappear. Criteria enter in a case of doubt not as a basis for an inferred judgement but in an attempt to relate to the scene as a scene of expression, and thus as a scene where no inference is needed. We are looking for something that helps us to see that the situation – despite its first appearance – is normal. In other words, we are looking for something that will restore our familiarity with the situation.
Now, imagine the same case of doubt, but abstract from the fact that there is a practice of using the concept of pain. If there is no such practice, our appeal to criteria cannot have the function of restoring our familiarity with the situation. For that we are familiar with a situation in which someone is expressing pain means that it is the sort of situation which constitutes the fundamental phenomena of this practice. So if there is no practice, there is no familiarity to restore. The role of criteria now can only be to provide an epistemic basis for an inferred judgement that the other really is pain. If we abstract from the practice of using the concept of pain, we can only understand our appeal to criteria as an appeal to an epistemic basis for our judgement. However, if we understand our appeal to criteria as an appeal to evidence, we cannot avoid the discovery that criteria necessarily fail to provide such an epistemic basis. Any criterion for pain the mother might appeal to will be such that it may be satisfied and it still be the case that her boy is not in pain.
When we abstract from the fact that there is a practice of using the concept of pain, we abstract from the fact that there are normal cases of using this concept: cases with which we are familiar. We can call this the sceptic's condition: the sceptic is taking a point of view on our knowledge of others in which there are no normal cases. From this point of view, there does not seem to be any such thing as a normal case, and thus for the sceptic every case is a case of doubt: one in which a gap appears between the behaviour of the other and the inner state which we actually wish to know. One could also say: the sceptic investigates the case of doubt as if it were the normal case. As we have seen, the fact that the sceptic takes the case of doubt to be the normal case – which means that she does not have the concept of a normal case – has a decisive consequence: she cannot conceive of our knowledge of others as being anything other than a result of having bridged the gap between the behaviour of the other and her inner state. The absence of a notion of a practice leads the sceptic to epistemologise our knowledge of others.19 She cannot avoid conceiving of our knowledge that the other is in pain as being based on an inference from what is given in experience to something which lies behind.
The diagnosis of the sceptic's condition, then, runs as follows: having in her enquiry substituted for the normal case the case of doubt, and not being aware of this substitution, she cannot avoid misunderstanding the real significance of criteria. She cannot help but think that their role is to make our knowledge about others possible by providing the basis for an inference that, as she rightly insists, we can never be justified in drawing.
The sceptic's conception of knowledge as the result of an inference which bridges the gap between what is given in experience and the inner state of another is not simply a prejudice that could be philosophically challenged. It is rather the expression of a denial: a denial of the very practice in virtue of which her concepts do have content, and hence a denial of that which allows her to say anything at all.
The failure of the dissolving therapy, I have suggested, has its ground in the fact that it, too, operates with the premise that constitutes the real root of scepticism, namely, that our fundamental relation to others is an epistemic one. By way of a conclusion, I would like to claim that the reason for this presupposition is an inadequate understanding of the nature of the point of view we take when we engage in philosophical reflection.
The dissolving therapist believes that the sceptic's confusion consists in her attempt to comprehend our relation to other subjects from the vantage point of an observer, instead of placing herself in the position of a participant and clarifying it from within. However, the dissolving therapist conceives of our tenacious inclination to take the observer's vantage point as resulting from historical and sociopsychological, that is to say, extraphilosophical forces. That scepticism is a contingent fact of the philosophical tradition, as the dissolving therapist has it, implies that philosophical clarification is something which can be carried out without taking the vantage point of an observer.
However, I would like to claim that philosophical clarification can only begin in so far as one has taken the observer's vantage point. If this is right, then sceptical doubt does not represent an accidental confusion within the philosophical tradition, but rather has its occasion in the nature of philosophy itself.20 It is the philosophical perspective as such that suggests the idea to us that our knowledge concerning others requires an epistemic basis.
Why is this so? Why does it contradict the idea of philosophical reflection to hold that the participant's position could be clarified without our having to take the perspective of an observer? The reason, according to Cavell, is that philosophy, were it not to assume that vantage point, would have nothing to focus on at all. For then the very idea of specific judgemental practices, whose possibility we can enquire into and which we can seek to clarify, would be unavailable to us. In other words, it is precisely from the observer's vantage point that we are first able to bring the position of the participant into view as such; and it is thus that every philosophical reflection is first provided with its object. The idea that there is a practice of judgement with a specific form represents the beginning of any philosophical enquiry. Without relating to ourselves as subjects who do not merely judge in this or that way, but have a practice of making judgements, one in which, for instance, inner states are ascribed to other subjects, no properly philosophical reflection upon the form of our knowledge of those states could commence at all.
But why are we able to identify and classify such practices from the observer's vantage point alone? It is a fundamental characteristic of practices that a practice exists at all only in so far as we are familiar with it. But if practices have their being in their being familiar, this means that we can identify them as such only when we remove or distance ourselves from them. Or better: to identify as such something that is familiar to us means giving up our familiarity with it, leaving it behind or beginning to find it strange. Identifying and classifying something that has been familiar to us, and the increasing strangeness of what was familiar about it, are two sides of the same process. One already must have lost quite a bit of one's familiarity with some practice if one is to be at all able to identify it as such; yet at the same time it is precisely this identification that accentuates the estrangement from it requisite for identifying it in the first place. The idea of a particular practice of making judgements, such that one can ask philosophically how that practice is possible, necessarily assumes that one is not involved in it, but has stepped back from it. Hence, the question whether in our philosophical reflection we have to assume either the participant's position or the observer's vantage-point is beside the point. That we start to reflect philosophically means that we have already assumed the latter point of view.
This explains why everything might appear to be in question in the sceptical light of the philosophical perspective. For, from the very start, this perspective is one in which we have broken off our familiar relationship with the world and other people, such that they might appear to us as strange. Philosophy has the tendency to cast a disruptive glance at the world; in its view, it seems as though the world and other people have to be, as Cavell puts it, ‘regained every day, in repetition, regained as gone’ (Cavell 1988: 172). The philosophical perspective itself, not only some feature of it, is such that in taking it we already set one foot on the road to scepticism. As Cavell writes, ‘Wittgenstein's claim is that philosophy causes us to lose ourselves and that philosophy is philosophy's therapy’ (Cavell 1979: 34).
Of course, this does not mean that philosophical reflection will necessarily end up in scepticism. Philosophy is only a necessary condition for scepticism, not a sufficient one. However, it does mean that the perspective of philosophy as such will tend to instill sceptical doubt in us, which will seem inescapable to the extent that we fail to understand and take into account the peculiar character of this perspective. It is not the mere taking of an observer's vantage point that leads philosophical reflection into sceptical doubt, as dissolving therapists rashly claim, but rather an inadequate conception of that viewpoint. One becomes a sceptic irrevocably when one conceives of philosophical reflection as explaining the possibility of our practices without making use of the idea of our being familiar with these practices, hence with our being familiar with its normal cases.21 One is then not aware of the fact that philosophical reflection itself is the very reason for the absence of this familiarity, and hence the reason for the very problem that we try without success to solve in the course of our enquiry. It is precisely when one fails to comprehend this that one will understand the point of the criteria we appeal to in cases of doubt to be epistemic; they will then seem to furnish an epistemic basis upon which our knowledge that another is in pain is founded.
In her philosophical enquiry, the sceptic not only steps back from our practice of making judgements in order to obtain a reflective understanding of that practice, but does so unaware of what she is doing. She is blind to the fact that her philosophical reflection on our practice precisely begins with such a withdrawal from our practice. From this blindness towards the character of her reflection, it follows that the sceptic is forced to deny the very knowledge she cannot help but have: the very knowledge that is implicit in her practically responding to the world and other people (Cavell 1979: 104 and 109). By deceiving herself about our relation to the world and other people in this way, the sceptic is deceiving herself about one thing above all: herself (Cavell 1979: 179–80, 207, 222).
The sceptic therefore cannot be freed from her doubt simply by being shown that she is caught in a ‘muddle.’ She is not merely drawing a false conclusion from a confused premise. Rather, she is denying something that she cannot help but have. One can say neither that this denial is necessary nor that it is arbitrary; it is a denial linked to the nature of philosophical reflection itself.
Therefore, the therapy that Wittgenstein devises for the sceptic will not relieve her of all doubt, but will get her to see the denial upon which her doubt rests. Comprehending this denial does not render the sceptic's doubt meaningless but gives it a different meaning – perhaps one she can live with.22
1 PI 153. Concerning the idea of confusion, see also, inter alia, PI 16, 38, 132, 149, and 693.
2 See PI 133, 255.
3 See, for example, McDowell 1994 (especially pp. 175–80), 1998a Chs. 11–12, 1998b Ch. 17, Diamond 1991 Introduction 2, Chs. 1, 7 and 9; and McGinn 1989 and 1998.
4 I am referring to the position shared by authors like Rogers Albritton, Norman Malcolm, Gordon Baker and Crispin Wright, to mention only those who are particularly important in the present context. See Albritton 1966; Baker 1974; Malcolm 1968; and Wright 1982.
5 Another reason for the defeasibility of criterial evidence is the fact that criterial relations are context-dependent. See, for example, Baker 1974: 162 and Malcolm 1968: 85.
6 Compare Davidson's reflections upon Frege's assertion sign in ‘Moods and Performances,’ p. 113, in Davidson 1984.
7 Thus, Baker 1974: 163. See also Malcolm 1968: 84: ‘The satisfaction of the criteria of y establishes the existence of y beyond question.’
8 For this objection see, for example, John McDowell, ‘Criteria, Defeasibility, and Knowledge’, p. 372 (in McDowell 1998b).
9 See Cavell 1979: 41. Dieter Birnbacher, who interprets the concept of criteria in the manner of the anti-sceptics, attempts to evade the predicament that then results with this formulation: ‘So the justification conferred by evidential criteria is a precarious one; it does not furnish a logical guarantee. Any single constellation of the supporting evidence is fallible and can in principle be destroyed by other pieces of evidence which might be produced at some future time . … Evidential criteria guarantee not an objective certainty (in the sense of infallibility) but an objectively justified claim to certainty in practice, which may be fallible but … is justified and acknowledged.’ (Birnbacher 1982: 64) But this description, too, means that I can be entirely justified in believing that x exists even though it could well be the case that x does not actually exist.
10 See McDowell 1994: 34–6, 41–2, 81–2, 168.
11 See, for example, McDowell 1994 (pp. 70ff), ‘Two Sorts of Naturalism’ (pp. 174ff) and ‘Aesthetic Value, Objectivity, and the Fabric of the World’ (pp. 128f) (both in McDowell 1998a).
12 See McDowell 1998b: 385.
13 I am using the idea of a hermeneutic circle in the sense in which Hans-Georg Gadamer has developed it, in Gadamer 1986 (1960]: 270–95.
14 A similar criticism of McDowell's conception is also made in Glendinning 1998: 138 ff. I have discussed McDowell's position in greater detail in Kern 2000a. There, I argue that McDowell's failure stems from his having fused together two different modes of arguing, which, though they do point in the same direction, cannot simply be combined. His strategy is the incoherent attempt to carry through a Wittgensteinian therapy for philosophy by means of Kantian transcendental reflection.
15 For the notion of a ‘normal case’, see, for example, PI 141, 142; and John Austin, ‘Other Minds,’ p. 113 in Austin 1961.
16 See, for example, Cavell 1979: 51.
17 Cavell puts this point as follows: ‘Wittgenstein's appeal to criteria, though it takes its importance from the problem of scepticism, is not, and is not meant to be, a refutation of scepticism. Not at least, in the form we had thought a refutation must take. That is, it does not negate the concluding thesis of scepticism, that we do not know with certainty of the existence of the external world (or of other minds). On the contrary, Wittgenstein, as I read him, rather affirms that thesis, or rather takes it as undeniable, and so shifts its weight. What the thesis now means is something like: our relation to the world as a whole, or to others in general, is not one of knowing, where knowing construes itself as being certain. So it is also true that we do not fail to know such things.’ (1979: 45).
18 As Wittgenstein describes our capacity to master the use of a concept in PI 219.
19 To say that the sceptic epistemologises our knowledge is not to say that she wrongly assumes an epistemic relation to certain ‘propositions’ that make no epistemic claim, as Marie McGinn argues. McGinn argues that Wittgenstein dissolves scepticism's problem by showing that the sceptic assumes an epistemic relation to certain propositions that do not in fact make any epistemic claims but rather embody techniques for describing the world (techniques which we can only master in practice). I take this to mean the following: every judgement constituting knowledge contains not only epistemic elements, that is, convictions that make epistemic claims, and which we, for our part, are able to justify, but also non-epistemic elements, convictions of a sort Marie McGinn calls ‘framework propositions,’ to which we have no epistemic relation at all. The function of these propositions is to render explicit the presuppositions that make the judgement possible in the first place, presuppositions we become aware of only in practice, when we act, and never with the help of reflection alone. When I say that the sceptic epistemologises our knowledge, this means something else: it implies not that our knowledge of others or of the world contains a non-epistemic element, so-called presuppositions that ground our knowledge, but that our knowledge, e.g. of the existence of another's inner state, does not have any ground at all. There are no presuppositions that ground our knowledge that the other is in pain because in normal cases there is no step to be taken from the identification of pain to its existence. This distinction explains why she believes that Wittgenstein's therapy for philosophy consists in dissolving the sceptic's doubt. See McGinn 1989; and also, more specifically with regard to her understanding of the concept of therapy, McGinn 1997, especially pp. 23, 27 and 143–76. With regard to scepticism about other minds, see McGinn 1998 esp. pp. 51f.
20 For that, see, for example, Cavell's essay, ‘The Uncanniness of the Ordinary,’ pp. 170–4, in Cavell 1988.
21 That the criticism of this sceptical conception of philosophy has, at the very least, its roots in Kant's critical philosophy, I argue in Kern 2000b. There I suggest that the significance of Kant's criticism of scepticism can be understood only when one sees that it also corrects the sceptical conception of philosophy's task.
22 Translated from German by Jack Ben-Levi.