8   Leaps in the dark

Epistemological skepticism in Kripke’s Wittgenstein

Hannah Ginsborg

I. Introduction

In his famous discussion of Wittgenstein on rules and private language, Kripke describes Wittgenstein as having “invented […] the most radical and original skeptical problem that philosophy has seen to date” (60).1 He presents the problem by imagining a skeptic who challenges an interlocutor’s certainty both about the correctness of her present use of an expression and about what she meant by the expression in the past. The problem as first introduced appears on the face of it to be an epistemological problem on the model of skepticism about the external world or about other minds. What is called into question is our knowledge of the meanings of our expressions and of how we ought to use them. But Kripke goes on to draw a conclusion that is metaphysical rather than epistemological.2 The upshot of the skeptical argument is not just that we don’t know what we mean by our expressions, but that there is no such thing as meaning.

Many commentators have held that Kripke’s skeptical problem is exclusively, or at least fundamentally, metaphysical or constitutive. Kripke, it is thought, presents this problem in the guise of an epistemological problem, or using the language of epistemological skepticism, but the weight of it rests on the metaphysical question of what constitutes the fact of someone’s meaning something by an expression, or of the expression having a meaning. Some commentators take epistemological considerations to figure as constraints on answers to the metaphysical question, but, at least on one influential line of interpretation, epistemological considerations play no role in the skeptical argument at all. And even where it is thought that epistemological considerations do play a role, the basic thrust of the argument is still understood as metaphysical or constitutive: the skeptic who casts doubt on your knowledge of meaning is seen as a dispensable and potentially misleading device for questioning the existence of meaning.

Although I grant that the skeptic’s conclusion is metaphysical rather than epistemological, I believe that the skeptic is much more than an expository device and that the thrust of Kripke’s argument is much closer to standard skeptical arguments than has previously been acknowledged. I shall try to show in this paper that, on the most charitable understanding of the argument, its core is epistemological: the skeptic’s challenge to your knowledge that you meant addition is indispensable to the metaphysical conclusion we are supposed to draw from the argument. My interpretation, if successful, is important not just because it corrects a misreading of the skeptical argument, but because it forecloses a certain seemingly straightforward response to it. Many philosophers who have taken the argument to be fundamentally metaphysical have answered it on correspondingly metaphysical grounds by arguing that Kripke is either unduly restrictive regarding the range of facts that are candidates for constituting meaning, or unjustified in supposing that meaning facts have to be constituted by more primitive facts at all. If, on the other hand, the argument is fundamentally epistemological, then that response is no longer effective. A satisfactory response to the argument has to contend with the skeptical challenge to our knowledge of which uses of an expression are correct, a challenge that cannot simply be met by broadening the range of acceptable answers to the metaphysical question of what meaning consists in.

My reading of the skeptical argument aims, then, to show it in a better light than most of Kripke’s critics have allowed: it is, I argue, stronger and more interesting than commentators have typically given him credit for. But, while my reading allows the argument to avoid one widely accepted response, it opens the argument to a different and less familiar line of objection that, as I argue in conclusion, renders it ultimately unsuccessful.

II. The skeptical challenge: metaphysical or epistemological?

I begin by rehearsing the skeptical challenge as Kripke initially presents it. Suppose that you have never before added numbers larger than 57 and are now asked “What is 68+57?” You answer “125,” and you do so with the conviction that your answer is correct, not just in what Kripke calls the arithmetical sense—that is, not just in the sense that “125” is the correct answer to the question “What is 68+57?”, with “+” understood as meaning addition—but in a different sense, which he calls “metalinguistic,” and which has to do with the relation of your answer to your previous linguistic usage. Roughly, you are convinced that your saying “125” is correct in the sense that it accords with how you used the “+” sign or the word “plus” in the past. Now a skeptic challenges your conviction as to the “metalinguistic correctness” of “125” as an answer to the question. He does so by raising the skeptical hypothesis that you are now misinterpreting your past usage of “+.” You now think that, when you used that term in the past, what you meant by it was addition, whereas what you in fact meant was quus or quaddition, defined so that x quus y is the sum of x and y where x and y are less than 57 and otherwise 5. On that hypothesis, he says, you ought—if you are to accord with your previous usage—to respond to the query with “5” rather than “125.” In order to defend your original conviction as to the “metalinguistic correctness” of “125”, you must come up with “some fact about [your] past usage which can be cited to refute [the skeptic’s hypothesis]” (9). If you are not able to cite such a fact, then your answer of “125” is revealed as an “unjustified leap in the dark” (10).

Kripke goes on to argue that this challenge cannot be met. His first step is to imagine that you try to meet it by appealing, not to the finite list of your previous uses of “+”, which (you concede) is compatible with the hypothesis that you meant quaddition, but to the instructions you had in mind when you responded to “+” questions as you did. This appeal fails, he says, because the skeptic can cast into doubt what you meant by the words that figure in those instructions, just as he challenges what you meant by “+”. Any rules or instructions you say you gave yourself for using those words in the future must themselves be put in words that themselves require interpretation, and so the skeptic can always claim, as he did for “plus,” that, in invoking them to answer the skeptical challenge, you are misinterpreting them. In the end, your present beliefs about what you meant by your words in the past must depend on appeal to a finite number of past uses, and it is always possible for the skeptic to challenge your justification for those beliefs by suggesting quus-like hypotheses that are consistent with hose uses (15).

The passages on which I have been drawing suggest that the challenge is epistemological. It looks as though the skeptic has challenged your conviction as to the “metalinguistic correctness” of “125” by putting forward the skeptical hypothesis that you meant quaddition and that you are now deceived about your previous meaning. Your task, it appears, is to rule out that hypothesis by justifying your belief—on which your conviction as to the correctness of “125” depends—that you meant addition. Even though the challenge is not to the reliability of your memory—you are granted complete recall of all your past behaviour and mental states—it is still epistemological: the skeptic demands your justification for believing, given knowledge of your past responses to “+” questions and the mental states accompanying them, that it is addition rather than quaddition that you meant. But after initially presenting the challenge in this seemingly epistemological way, Kripke goes on to describe it in metaphysical rather than epistemological terms. The skeptic questions “whether there is any fact that [you] meant plus, not quus” and an answer to him must “give an account of what fact it is […] that constitutes my meaning plus, not quus” (11). Epistemological considerations are indeed mentioned, in this formulation of the challenge, but only as constituting a constraint on answers to the metaphysical challenge: “any putative candidate for such a fact must […] show how I am justified in giving the answer ‘125’ to ‘68+57’” (11). And much of the argumentative work that follows is couched in correspondingly metaphysical terms. The bulk of the chapter in which the skeptical problem is presented consists in the consideration and rejection of various putative facts in which your having meant (or meaning) addition might be thought to consist: the fact of your being disposed to give the sum, the fact of your possessing a unique introspectible quale associated with the addition function, or simply the fact of your meaning addition, understood as a sui generis fact that is not available to introspection and does not admit of reduction to any more basic facts. Although Kripke does consider one response to the skeptic which apparently assumes the challenge to be epistemological, since it defends the hypothesis that you meant addition on the grounds that it is simpler than the hypothesis that you meant quaddition, the primary reason he gives for rejecting it is precisely that it misunderstands the challenge as purely epistemic rather than constitutive (38–40). In fact, Kripke expressly denies that the skepticism is epistemological:

The problem may appear to be epistemological—how can anyone know which of these [plus or quus] I meant? Given, however, that everything in my mental history is compatible both with the conclusion that I meant plus and with the conclusion that I meant quus, it is clear that the sceptical challenge is not really an epistemological one. It purports to show that nothing in my mental history or past behaviour3—not even what an omniscient God would know—could establish whether I meant plus or quus. But then it appears to follow that there was no fact about me that constituted my having meant plus rather than quus. How could there be, if nothing in my internal mental history or external behavior will answer the sceptic who supposes that in fact I meant quus?.

(21)

Commentators have disagreed about the relation between epistemological and metaphysical elements in the skeptical problem. Colin McGinn, one of the first commentators to broach the issue, presented the central skeptical problem as constitutive or metaphysical, but claimed that Kripke’s skeptic also raises a distinct epistemological problem about the justification of my linguistic inclinations, and that the solution to this problem is independent of the solution to the primary, metaphysical problem.4 Anscombe also identified two different problems: the epistemological problem “how do I know I meant plus?” and the “more interesting […] [metaphysical] problem with which Kripke is implicitly confronting himself: what is the fact that he knows, namely that he meant, and means, plus?”5 For Crispin Wright, however, the epistemological considerations raised by Kripke did not motivate an epistemological problem distinct from the metaphysical one:6 rather Kripke’s skeptic was a “mere device” for presenting a constitutive thesis.7 Other early commentators also either specifically asserted that the problem was metaphysical rather than epistemological,8 or simply described Kripke’s problem in exclusively metaphysical terms.9 Warren Goldfarb offered a clear statement of the metaphysical approach: “the skeptical challenge is not intended to operate epistemologically...rather, it rests on ontological considerations […] it is the notion of fact, of ‘everything there is,’ that is to provide the ground of the challenge.”10 Paul Boghossian, in his influential 1989 discussion, makes a point of presenting the problem without reference to the skeptic or the quaddition hypothesis, thus demonstrating what he takes to be the exclusively constitutive character of Kripke’s skepticism.11 And subsequent philosophers, whether or not they explicitly engage the issue of whether the skepticism is fundamentally epistemological or metaphysical, have tended to agree in construing the problem in metaphysical terms as a matter of how to identify facts in which meaning consists.12

There has indeed been disagreement, within this broad consensus, about whether epistemological considerations are relevant to the skeptical challenge insofar as they impose constraints on possible answers to the metaphysical question. Among early commentators, Wright held that they do,13 while McGinn and Boghossian viewed them as entirely irrelevant: for Boghossian the temptation to see epistemological considerations as playing a role in the skeptical dialectic results from a “distortion […] induced by the dialogic setting.”14 But whether or not epistemological considerations are seen as figuring in the argument, there is broad agreement that Kripke does not present an argument that is skeptical in the traditional, epistemological sense. The seemingly epistemological challenge presented in the opening pages of Kripke’s text is, rather, a “device” for presenting a metaphysical argument,15 a “guise” for constitutive skepticism,16 or the result of a misleading “tendency to slip into the language of epistemological skepticism.”17

I find this approach to the argument unsatisfying. First, it is hard to believe that, without the skeptic and the quaddition hypothesis, Kripke’s challenge would have been able to engage the interest of so many philosophers. Although this is not itself decisive, it seems reasonable to suspect that the skeptical dialectic is not just novel packaging for a familiar product, but corresponds to something philosophically distinctive about Kripke’s challenge, something that is lost if we read the challenge in a way which makes the skeptic dispensable.18 Second, and perhaps more importantly, the metaphysical reading allows Kripke’s challenge to be too easily answered. It can simply be pointed out that Kripke is operating with an unwarrantedly narrow conception of fact, or that he is refusing to accept that meaning facts could be primitive and irreducible. Many interpreters who see the problem as metaphysical have indeed offered just this kind of response. Thus Goldfarb follows up his claim that “it is the notion of fact which is to provide the ground of the challenge” with an argument to the effect that the challenge can be turned aside by broadening the range of acceptable facts to include, for example, facts as construed by Frege.19 And McGinn accuses Kripke of an unargued reductionism, a charge which is taken up by many other interpreters, including Wright and Boghossian.20

Against the first of these points it might be replied that, while Boghossian’s reading dispenses with the (epistemological) skeptic, many broadly metaphysical readings do not. According to Wright,

the overarching thought behind [the skeptical dialectic] is...that claims of a certain kind cannot be supposed to deal in matters of real fact if someone could know all possible facts which might conceivably constitute the truth of such a claim yet be unable to defeat a sceptic concerning his knowledge of its truth.”21

So even though the skeptic is a “mere device” he does have a role to play. Your inability to meet his challenge to your knowledge that you meant addition, in a context where you are assumed to have complete knowledge of your past behaviour and past mental states, is a stepping-stone—granted the “overarching thought” Wright describes—to the non-factuality of the claim that you meant addition. However, the “overarching thought” might seem questionable to those lacking verificationist sympathies. Why should your inability to justify your belief that you meant addition, even given the knowledge that Kripke’s skeptic allows you, entail that your belief lacks factual content? Moreover, as Wright makes clear, this reading still leaves Kripke open to the charge of reductionism, since it restricts the class of potential truth-makers to facts about behaviour and mental states non-intentionally construed, failing to allow that meaning facts could be irreducible.22

Now for Wright and others who have pressed it, the charge of reductionism represents an objection, and indeed a decisive objection, to Kripke’s argument. But as already indicated, I take the ease with which it can be made to constitute, rather, an objection to the metaphysical reading.23 In what follows I want to offer an account that improves on the metaphysical reading with respect to both of the points I have mentioned: it assigns an essential role to the skeptic, and it defuses the charge of reductionism.

III. Reading the argument as epistemological

I will begin by drawing attention to two features of the argument that I mentioned in my overview, but which have often been ignored or given insufficient weight. The first is that the primary focus of the skeptic’s challenge in the opening phase of the argument is not your putative knowledge that you meant addition rather than quaddition by “+” but rather your confidence about what Kripke goes on to call your “present particular response”, that is, about how you ought to respond to the specific “68+57” query. (“Ought,” that is, in a sense corresponding to ‘metalinguistic correctness’: more on that qualification shortly.) Kripke describes you at the outset as “confident that ‘125’ is correct […] in the metalinguistic sense” and the skeptic as challenging your certainty about that answer: “Perhaps […] as I used the term ‘plus’ in the past, the answer I intended […] should have been 5!” It is only subsequently, following your complaint that the skeptic’s suggestion is “insane,” that the skeptic backs up his initial challenge by proposing that you previously meant quaddition. And the idea that it is your knowledge of the correctness of your “present particular response” that the skeptic is primarily concerned to challenge is borne out by Kripke’s going on to offer formulations of the challenge that focus on your justification for responding “125”, often with no direct reference to your putative knowledge that you meant addition. What the skeptic challenges is your supposition that “in computing ‘68+57’ as I do, I do not simply make an unjustified leap in the dark” (10); the skeptic “holds that my present response is arbitrary” (11); the problem with the various philosophical theories that you might invoke in responding to the skeptic is that they “fail to give a candidate for a fact as to what I meant that would show that only ‘125’, not ‘5’, is the answer I ‘ought’ to give” (11); the skeptic “doubts whether any instructions I gave myself in the past compel (or justify) the answer ‘125’ rather than ‘5’” (13); “the skeptic argues that when I answered ‘125’ to the problem ‘68+57’, my answer was an unjustified leap in the dark; my past mental history is equally compatible with the hypothesis that I meant quus, and therefore should have said ‘5’” (15).

This suggests that if we are going to read the opening phase of the skeptic’s argument as epistemological, on the model of skepticism about the external world, we should understand its target not as knowledge of meaning, but as knowledge of what to say in particular cases like that illustrated in the example. Your knowing to say “125” in response to “68+57”, or, as in Kripke’s later “tabair” example, your knowing to apply “table” to a table at the base of the Eiffel Tower, is a “best case” of knowledge, on the model of Descartes’ knowledge, in the First Meditation, that he is sitting by the fire holding a piece of paper.24 It is a particular example of knowledge from which the skeptic can go on to generalize to cast doubt on all our putative knowledge about the correctness of our uses of expressions. The quaddition hypothesis does indeed cast into doubt your belief that you meant addition, but this is comparable, on the present suggestion, to the way in which Descartes’ dreaming hypothesis casts into doubt his belief that he is awake. According to the dreaming argument, Descartes has to justify his belief that he is awake rather than asleep in order to justify his belief that he is holding a piece of paper, but it is the belief that he is holding a piece of paper, not his belief that he is awake, that is the target of the skeptical doubt he raises. By the same token, according to Kripke, you have to justify your belief that you meant addition rather than quaddition in order to justify your belief in the “metalinguistic correctness” of “125”, but the quaddition hypothesis is primarily aimed at undermining the second of these two beliefs.

The second of the two points has to do with Kripke’s specification of the correctness in question as “metalinguistic.” The skeptic challenges your confidence that “125” is the correct answer “in the metalinguistic sense that ‘plus,’ as [you] intended to use that word in the past, denoted a function which, when applied to […] ‘68’ and ‘57’, yields the value 125” and that “as [you] used the symbol ‘+’, [your] intention was that ‘68+57’ should turn out to denote 125” (8). The metalinguistically correct response to the question “What is 68+57?” according to these formulations, is the response that fits the meaning or intention associated with your previous uses of the “+” sign. In other passages, Kripke characterizes the correctness at issue more briefly, in terms of agreement or accordance with past usage, and without reference to meaning or intention. The skeptical hypothesis is described as the hypothesis that “I have changed my previous usage” (11); it is agreed that, on the assumption I meant plus, “then unless I wish to change my usage, I am justified in answering 125” (11); “[the skeptic] questions whether my present usage agrees with my past usage” (12); if I meant addition rather than quaddition, then “to accord with my previous usage I should say ‘125’” (12). Either way, however, what the skeptical hypothesis calls into question is your knowledge not of what you ought to say given what you mean by “+” now—something that presumably coincides with what is arithmetically correct—but of what you ought to say in order to conform to something about you in the past: whether the meaning with which you used “+”, your intention in that past use, or that past use simpliciter.

Kripke’s seeming indifference between talk of past meaning, talk of past intention and talk of past usage makes it hard to pin down further what he means by “metalinguistic correctness”, and, as will emerge in Section 5, it reflects a point of confusion that—I believe—ultimately undermines the skeptical argument. For now, however, I will simply propose that, to arrive at the most charitable reading of Kripke’s argument, both in its own right and as reflecting Wittgenstein’s concerns in the rule-following considerations, we should privilege the formulations in terms of past usage. The belief that the skeptic challenges is your belief that, in saying “125”, you are responding to the expression in a way which fits how you—and, we might add, your parents and teachers—responded to it in the past: naively put, it is your belief that you are responding “in the same way” as you did previously. You think that you are, to use a somewhat Wittgensteinian turn of phrase, “going on as you ought” from your past use of “+” and that of your teachers.25 The skeptic calls your justification for that belief into question by casting into doubt what you meant in that past use: if you cannot rule out that you meant quaddition, he says, then you are not justified in believing that you are according with your previous use. So the sense in which your saying “125” is, on the quaddition hypothesis, an “unjustified leap in the dark” (10, 15) is that you lack justification for taking yourself to be responding to “+” questions in a way that is consistent with your past responses. You are making a leap in the dark in the same sense in which someone, asked to continue a series of numbers but lacking any idea of what the principle of the series is supposed to be, makes a “leap in the dark” by just making a random guess as to what comes next. The “leap” is from your past uses of “+” to the use of it now manifested in your utterance of “125”. Before the skeptic came on the scene, according to Kripke, you believed that your choice of “125” as, so to speak, the correct next step in the series—as the thing to say in the light of how you used “+” in the past—was guided by your knowledge that, in your past use of “+”, you meant addition. The skeptical hypothesis, in undermining your claim to knowledge of your past, reveals that guidance to be illusory, so that in now “going on” from your past uses to your present use you are going on blindly.

With these two points in view we can reconstruct, from the opening moves of Kripke’s discussion, an argument with the following steps at its core:

  1. If you do not know that you meant addition rather than quaddition, then you do not know that you now ought to respond to “68+57” with “125” rather than “5” in order to accord with your previous uses of “+”.

  2. You do not know that you meant addition rather than quaddition.

  3. You do not know that you now ought to respond to “68+57” with “125” rather than “5” in order to accord with your previous uses of “+”.26

Because your conviction that you ought to say “125” is a “best case” the argument generalizes: you can never know, of any use of language, that it accords with your previous usage—where your previous usage can be understood to include not just your own utterances of and responses to linguistic expressions, but all uses that you have observed, including those of your parents and teachers. You can never know that you are “going on as you ought.” And the generalization, of course, goes further: none of us is ever in a position to know, of any use of a linguistic expression, that it accords with previously observed uses. Each use of language is, in the sense explained above, a leap in the dark.

The argument I have outlined is on the model of traditional skeptical arguments regarding our knowledge of the external world, and its conclusion is epistemological: we can never know, in any use of any expression, that we are according with previous uses of that expression. But—and this is a disanalogy with the case of external world skepticism—we can go on to draw from it a further, metaphysical conclusion. For, if none of us can ever know that we are now according with our, or anyone else’s, previous uses of any expression—if each new use of an expression is a leap in the dark—then it seems that there can be no such thing as understanding our own or anyone else’s expressions, or, correlatively, as meaning anything by them. Take away our confidence, in each use of a familiar expression, that our use of it fits previous uses, that we are “going on in the same way” we and others were going on before, and what remains no longer looks like meaningful use: it appears as though, in Kripke’s words, meaning has “vanish[ed] into thin air” (22). For Kripke this metaphysical conclusion follows because he assumes that meaning something by an expression at any one time is a matter of grasping instructions telling one what to do in the future to conform to one’s use at that time. So if, at a later time, one is not in a position to know that one’s use conforms to earlier uses, it follows that, in those earlier uses, one did not mean anything by the expression. But we do not need to accept Kripke’s rather strong conception of meaning as a matter of grasping instructions in order to see the plausibility of the view that the phenomena of linguistic meaning and understanding require us to be able to recognize ourselves as conforming to previous uses. We need only accept that there is such a thing as meaning because there is such a thing as meaningful use of language, and that meaningful use in turn requires “knowing how to go on” in the use of an expression, where such knowledge in turn requires the capacity to recognize whether a given use is or is not in accord with previous uses.

On this reading of the overall argument, then, it does yield a metaphysical conclusion: there is no fact of your meaning or having meant addition by “+.” But the conclusion does not rest on your inability to say what such a fact consists in, but rather on your inability to justify your belief that “125” conforms to your previous uses of “+”. The consideration and rejection of the various possible facts that might constitute your having meant addition should not be understood as direct support for the claim that there is no such fact, but rather as supporting the epistemological claim—corresponding to Step (2) in my sketch—that you lack justification for believing, and hence do not know, that you meant addition. That claim in turn leads to the metaphysical conclusion that there is no fact that you meant addition, but not by way of a quasi-verificationist move from absence of justification to non-factuality. Rather, the move from your lack of knowledge that you meant addition to there being no fact that you meant addition goes via a claim about your lack of justification for taking your present and future uses of “+” to accord with your previous uses, which amounts to the claim that all your uses are and have been “leaps in the dark.” In contrast to a reading of the argument that sees the non-factuality of meaning as following immediately from your inability to specify a fact in which your meaning consists, the argument as I read it is not vulnerable to the objection that Kripke fails to recognize that meaning facts could be primitive. To counter the argument we have to meet the skeptic’s epistemological challenge to your belief that you ought now to say “125” to accord with your previous use of “+.” And, if Kripke is right that this requires justifying your belief that you meant addition rather than quaddition, then it is not enough to cite a fact in which your having meant addition consists (by saying, for example “The fact that I meant addition”): we have to show that you are justified in believing that you meant addition in the face of the competing hypothesis that you meant quaddition.

Now a critic of Kripke who accepted the epistemological–skeptical reading I have offered of the argument might here propose an epistemological variant of the charge of reductionism. Perhaps Kripke is failing to recognize, not just that there can be primitive facts of meaning, but that we have a primitive capacity to know that such facts obtain.27 In that case you can respond to the skeptic’s challenge “How do you know you meant addition?” by saying simply “I just know I meant addition; no explanation or justification necessary.” But the availability of this response, unlike the original charge of reductionism, is not a weakness in Kripke’s argument specifically but rather a feature of skeptical arguments generally. We might equally well respond to the external-world skeptic’s dreaming or brain-in-a-vat hypothesis by saying “I just know I’m awake now” or “I just know I have a body.” To respond this way is not to answer the skeptical challenge but to refuse to take it seriously, and it is no objection to Kripke’s skeptical argument that it is open to this kind of rejection.

IV. Issues with the epistemological interpretation

Can this interpretation be squared with Kripke’s own insistence on the metaphysical character of the skepticism? I believe it can. First, when Kripke explicitly describes the skepticism as metaphysical rather than epistemological (21; see also 38–39), his aim is to distinguish the metaphysical conclusion that there is no fact of your meaning addition from the epistemological conclusion that you do not know that you meant addition, not to distinguish the metaphysical conclusion from the epistemological conclusion that you do not know that “125” conforms to your previous usage. This is consistent with my view that the argument proceeds by first establishing the latter conclusion and then going on to draw the metaphysical conclusion as a consequence. Second, Kripke is clear that the metaphysical claim follows from your inability to answer an epistemological skeptic, as in his summary of the argument:

This, then, is the sceptical paradox. When I respond in one way rather than another to such a problem as ‘68+57’, I can have no justification for one response rather than another. Since the sceptic who supposes that I meant quus cannot be answered, there is no fact about me that distinguishes between my meaning plus and my meaning quus.

(21)

The context makes clear that the sceptic is demanding a justification for saying “125” in light of the hypothesis that you meant quus: it is your failure to answer him that leads to the conclusion that there is no fact about what you meant.28 Third, although much of the argument for your not knowing that you meant addition—Step (2) of the core argument sketched in Section 3—is cast in metaphysical terms as an argument that you cannot specify a fact in which your having meant addition consists, Kripke also invokes epistemological considerations, suggesting that the real issue is your inability to justify your belief that you meant addition in the face of the skeptical challenge to your knowledge of how to go on. Part of his objection to the proposal that your meaning addition is the simplest hypothesis is that “[if] I can only form hypotheses as to whether I now mean plus or quus […] then in the future I can only proceed hesitatingly and hypothetically, conjecturing that I probably ought to answer ‘68+57’ with ‘125’ rather than ‘5’” (40). To answer the skeptic, you have to show that you know you mean addition with the same level of certainty with which you know you ought to say “125” in the light of your previous uses. And in his very brief discussion of the proposal that meaning addition is a primitive, sui generis fact, he says that we have to be “aware of it with some fair degree of certainty whenever it occurs. For how else can each of us be confident that he does, at present, mean addition by plus?” (51). Although much of the argumentation regarding the various candidate facts is not specifically epistemological—for example Kripke goes on to say that an “even more important” difficulty for the sui generis proposal is the “logical difficulty” of how a finite state can somehow contain an infinite number of applications—all of it is compatible with the epistemological framework suggested on my reading, on which the interlocutor proposes these facts as justifications for her claim that she meant addition (and hence is justified in her belief of the metalinguistic correctness of “125”) rather than as responses to a self-standing metaphysical challenge. For it is open to the skeptic to reject these justifications on metaphysical as well as epistemological grounds, insofar as they have unacceptable metaphysical implications as to the kind of state that meaning could be.

Now, the presence of epistemological considerations in the arguments that Kripke gives against the various “candidate facts” for meaning addition—in my view, facts that can be cited to the skeptic to justify the claim that you meant addition—might well be seen as indicating not that the argument is fundamentally epistemological, but only that candidates for the fact of your meaning addition must satisfy epistemological constraints. This is, for example, Wright’s reading of the argument: Wright thinks that the argument is fundamentally metaphysical but that successful candidate facts must satisfy the constraint that they can be known to obtain in a first-personal, non-inferential way.29 Others have emphasized that it is a constraint on meaning facts, for Kripke, that their holding amounts to one’s being justified, in an internalist sense, in one’s uses of an expression: that your having meant addition by “+” amounts to your now being told or instructed that you should say “125.”30 Why isn’t it enough, to do justice to Kripke’s argument, that we treat the skeptic as making a metaphysical challenge that incorporates both of these epistemological constraints, calling on us to identify a plausible candidate for the fact of someone’s meaning addition, which can be known to obtain in an immediate, first-personal way and whose holding puts the subject in a position to recognize that she ought to say “125?” Kripke’s seemingly epistemological dialectic, with the skeptic’s appeal to the quaddition hypothesis to question your knowledge that you meant addition and hence your knowledge of what you ought to say, would be an unnecessary elaboration of a fundamentally metaphysical challenge.

The answer is that, when the challenge is framed in this metaphysical way, it is all too easy to respond by appealing to the existence of irreducible meaning facts with just the epistemological properties specified: we can say that it is simply in the nature of meaning facts that they are both first-personally knowable and sources of guidance or justification for our uses of expressions. It is only when we are presented, in the skeptical scenario, with a hypothesis that leads us to question a particular one of those uses, that the difficulty implicit in this notion of irreducible meaning facts is brought into the open.31 In undermining our confidence that “125” is correct in the light of our previous uses of “+”, the quaddition hypothesis undermines our confidence that we have the kind of knowledge of meaning that can serve to guide us in the use of expressions, and so undermines the conception of meaning facts on which the response to the metaphysical challenge depends.

V. An answer to the skeptical problem

I have described Kripke’s argument as having two premises: that if you do not know you meant addition by “+”, then you do not know that “125” conforms to your previous uses of “+”, and that you do not know you meant addition by “+”. These premises lead to the epistemological conclusion that your saying “125” is a “leap in the dark,” and the generalization of this epistemological conclusion leads to the metaphysical conclusion that there are no meaning facts. So understood, I have claimed, the argument is not vulnerable to the charge of unargued reductionism. How, then, are we to avoid the conclusion? I believe that we should reject the first premise. Kripke is wrong to assume that, in order to be confident that “125” is ‘metalinguistically correct’—that it conforms to your previous uses of “+”—you must know that you meant addition rather than quaddition in those previous uses.

I have defended this view elsewhere,32 so I will be brief here. I think that Kripke has conflated the idea of conformity to past usage with the idea of past meaning. He thinks that the question of whether you are “going on in the same way” in your use of an expression is the same as the question whether you are according with what you meant by that expression. But these two questions are distinct. It is possible for you to hold that “125” is the appropriate response to “68+57” in the light of your previous history of responding to “+” questions irrespective of what, if anything, you meant when you used the expression previously. So you can concede to the skeptic that you meant quaddition, and hence that, in saying “125”, you are failing to accord with what you meant in the past, and still maintain that “125” accords with your previous uses of “+”. Regardless of what you meant when you used the “+” sign in the past, the appropriate way to go on from the sequence of your past responses to “+” questions is to respond to “68+57” with “125” and not “5”.

Now the skeptic might at this point demand your justification for insisting that “125” rather than “5” is the appropriate thing to say in the light of your previous usage. But it is important to see that, once you have pointed out the irrelevance of what you meant to what conforms to your previous use, he has no way of motivating this demand. The original challenge to the correctness of “125” was compelling because it was backed by an argument: your belief in the correctness of “125” is based on the belief that you meant addition; the quaddition hypothesis shows that belief to be in need of justification; you cannot supply the required justification. But if that argument has been undermined, then you are perfectly within your rights to maintain your belief in the correctness of “125.” Your dialectical situation is quite different from that of the interlocutor at the end of Section 3 who insists, in the face of the skeptical challenge, “I know I meant addition.” That interlocutor is simply rejecting the purported reason for doubt—the quaddition hypothesis—that the skeptic has offered. You, on the other hand, have pointed out that it is not a good reason for doubt, leaving you entitled to hold on to your original conviction that, in saying “125,” you are going on as you ought.

The idea that we can make sense of a notion of conformity to previous use that is independent of conformity to previous meaning—and, more generally, of conformity to a rule grasped in one’s previous use—is unorthodox and requires more defense than I can provide here. But one reason to accept it is precisely that it gives us a way of avoiding the skeptical problem. If, as I have argued, the skeptical argument does not fall to the charge of reductionism, we need some other way of defending our ordinary confidence in the correctness of our linguistic behaviour and in the meaningfulness of the expressions we use. The idea in question, a special case of what I have elsewhere called “primitive normativity,” fills that need. It makes it possible to recognize the prima facie force of Kripke’s skeptical argument while escaping its devastating conclusion.33

Notes

1 Page numbers in the text refer to Saul Kripke, Wittgenstein on Rules and Private Language (Cambridge, MA: Harvard University Press, 1982).

2 For convenience I speak of the argument as Kripke’s, although he ascribes it to Wittgenstein and does not himself endorse it.

3 The text has “mental history of past behaviour,” but I believe that “or” was intended.

4 Colin McGinn, Wittgenstein on Meaning (Oxford: Blackwell, 1984), 149–150 and 176–180.

5 G.E.M. Anscombe, “Review of Wittgenstein on Rules and Private Language,” Ethics 95 (1985): 351. A similar approach was taken by Paul Coates, “Kripke’s Sceptical Paradox: Normativeness and Meaning,” Mind 95 (1986): 80.

6 Crispin Wright, “Critical Notice of Colin McGinn, Wittgenstein on Meaning,” Mind 98 (1989): 290.

7 Crispin Wright, “Kripke’s Account of the Argument against Private Language,” Journal of Philosophy 71 (1984): 762; the same point is made in his “Critical Notice,” 290 and his “Wittgenstein’s Rule-Following Considerations and the Central Project of Theoretical Linguistics,” in Rails to Infinity (Cambridge, MA: Harvard University Press, 2001), 186n27.

8 E.g., Graeme Forbes, “Skepticism and Semantic Knowledge,” Proceedings of the Aristotelian Society 84 (1984): 223.

9 E.g., Paul Horwich, “Review of Wittgenstein on Rules and Private Language, by Saul Kripke,” Philosophy of Science 51 (1984): 164–165; Christopher Peacocke, “Review of Saul Kripke, Wittgenstein on Rules and Private Language,” Philosophical Review 93 (1984): 263.

10 Warren Goldfarb, “Kripke on Wittgenstein on Rules,” Journal of Philosophy 82 (1985): 474.

11 Paul Boghossian, “The Rule-Following Considerations,” Mind 98 (1989): 515

12 See e.g., Ruth Garrett Millikan, “Truth Rules, Hoverflies, and the Kripke-Wittgenstein Paradox,” Philosophical Review 99 (1990): 324–325; Alex Byrne, “On Misinterpreting Kripke’s Wittgenstein,” Philosophy and Phenomenological Research 56 (1996): 339; Scott Soames, “Skepticism About Meaning,” Canadian Journal of Philosophy Supplementary Volume 23 (1997): 212; Jussi Haukioja, “Soames and Zalabardo on Kripke’s Wittgenstein,” Grazer Philosophische Studien 64 (2002): 162–165; Martin Kusch, A Sceptical Guide to Meaning and Rules: Defending Kripke’s Wittgenstein (Chesham, Bucks.: Acumen, 2006), 14–15; Arif Ahmed, Kripke (London: Continuum, 2007), 101–102; Anandi Hattiangadi, Oughts and Thoughts (Oxford: Oxford University Press, 2007), 20; Alexander Miller, “Rule-Following Skepticism,” in The Routledge Companion to Epistemology, ed. Sven Bernecker and Duncan Pritchard (Oxford and New York: Routledge, 2011), 455; James Conant, “Two Varieties of Skepticism,” in Rethinking Epistemology, Vol. 2, ed. James Conant and Guenter Abel (Berlin: De Gruyter, 2012), 23n21; Allan Gibbard, Meaning and Normativity (Oxford: Oxford University Press, 2012), Ch. 2; Paul Horwich, Wittgenstein’s Metaphilosophy (Oxford: Oxford University Press, 2012), 145.

13 Wright, “Kripke on Private Language,” 773ff; “Critical Notice,” 292ff. Others who have emphasized the relevance of epistemological constraints are mentioned in Note 30.

14 Boghossian, “Rule-Following Considerations,” 516.

15 Wright, references in Note 7; Boghossian, “Rule-Following Considerations,” 515; Alexander Miller, Philosophy of Language (London and New York: Routledge, 2007), 185; Conant, “Two Varieties,” 23n21. Note that Kripke himself describes the quaddition hypothesis as a “dramatic device” (10), but the problem it is described as dramatizing seems to be the epistemological problem that my saying “125” is unjustified, and not the metaphysical problem that there is no fact about what I meant.

16 Boghossian, “Rule-Following Considerations,” 515; Kusch, Sceptical Guide, 14.

17 Goldfarb, “Kripke on Wittgenstein,” 474. Conant refers similarly to Kripke’s “alternation between Cartesian [roughly: epistemological] and Kantian [roughly: metaphysical] idioms” (“Two Varieties,” 23n21).

18 We might even wonder if a problem can properly be called “skeptical” if it does not involve a challenge to knowledge or justification; Conant argues vigorously that it can (“Two Varieties”).

19 Goldfarb, “Kripke on Wittgenstein,” 474.

20 McGinn, Wittgenstein on Meaning, 150ff; Wright, “Critical Notice,” 292ff and “Wittgenstein’s Rule-Following Considerations,” 176–177; Boghossian, “Rule-Following Considerations,” 540–542. The objection is implicit in those interpreters who have proposed non-reductionist solutions to (or dissolutions of) the skeptical problem, for example Peacocke, “Review of Kripke,” and John McDowell, “Wittgenstein on Following a Rule,” in Mind, Value and Reality (Cambridge, MA: Harvard University Press, 1998) and “Meaning and Intentionality in Wittgenstein’s Later Philosophy,” Mind, Value and Reality.

21 Wright, Rails to Infinity (Cambridge, MA: Harvard University Press, 2001), 82; for an earlier statement, see Wright, “Kripke on Private Language,” 763. A similar line is taken by Miller, “Rule-Following Skepticism,” 456.

22 Wright, Rails, 83.

23 Gary Ebbs also holds that this objection stems from a misreading of the skeptical argument (see Rule-Following and Realism (Cambridge, MA: Harvard University Press, 1997), 27–31); space limitations preclude discussion of the differences between our alternative readings.

24 Here I disagree with Conant, for whom the “best case” on an epistemological reading of the argument would be your knowledge that someone means addition (“Two Varieties,” 23n21).

25 I defend this reading in my “Going on as One Ought,” presented as the 2016 Jacobsen Lecture at the University of London.

26 The details of the formulation are up for debate: as with other skeptical arguments of this general form, we might replace “do not know” with “are not in a position to know” or “are not justified in believing”, and Step (2) might be qualified as referring to knowledge of a kind that does not depend on the kind of knowledge specified in Step (3). I do not think that clarity on these details is essential for the present paper.

27 The idea that we have such a capacity is part of McDowell’s non-reductionist approach (see Note 20) and is explicit in Brian Loar’s description of the “dyed in the wool Brentanian” as maintaining that “we have direct access to the irreducibly intentional properties of our thoughts […] and that the non-arbitrariness of our applications of our concepts is connected somehow with that accessibility” (“Review of Wittgenstein on Rules and Private Language,” Noûs 19 (1985): 276). See also Jose Zalabardo, “One Strand in the Rule-Following Considerations,” Synthese 171 (2009): 513n7.

28 This is also clear in the passage from p. 21 quoted in Section II (152).

29 See references to Wright in Note 13.

30 See for example E.H. Gampel, “The Normativity of Meaning,” Philosophical Studies 86 (1997); Kusch, A Sceptical Guide, 62–64; Ahmed, Kripke, 103–107; Claudine Verheggen, “Semantic Normativity and Naturalism,” Logique et Analyse 216 (2011); Jason Bridges, “Rule-Following Skepticism, Properly So Called,” in Varieties of Skepticism, ed. James Conant and Andrea Kern (Berlin: De Gruyter, 2014), 256–258. Jose Zalabardo, “Kripke’s Normativity Argument” (Canadian Journal of Philosophy 27 (1997)) emphasizes both of these constraints, and his reading of Kripke’s normativity argument against dispositionalism, like my reading of the skeptical argument overall, centers on a requirement to justify the correctness a particular application of the “+” sign. However there is no indication that Zalabardo takes the argument, as I do, to constitute a form of epistemological skepticism, nor that—to mention a point which is central to my interpretation—he sees the relevant correctness in terms of “going on” appropriately from previous use.

31 Wright recognizes that there is a difficulty in explaining first person authority for facts about meaning (“Critical Notice,” 292–294)—in fact, one of his major criticisms of McGinn is precisely that McGinn finds the non-reductionist response too easy—but does not recognize the role of the (epistemological) skeptical dialectic in motivating it. I think that the depth of the difficulty is obscured by the fundamentally metaphysical reading of the argument that he and McGinn share.

32 Hannah Ginsborg, “Primitive Normativity and Skepticism About Rules,” Journal of Philosophy 108 (2011).

33 I would like to thank Martin Kusch, Alex Miller, José Zalabardo, and the editors of this volume for helpful comments and discussion.

Bibliography

Ahmed, Arif. Kripke. London: Continuum, 2007.

Anscombe, G.E.M. “Review of Wittgenstein on Rules and Private Language.” Ethics 95, no. 2 (1985): 342–52.

Boghossian, Paul. “The Rule-Following Considerations.” Mind 98 (1989): 507–49.

Bridges, Jason. “Rule-Following Skepticism, Properly So Called.” In Varieties of Skepticism, edited by James Conant and Andrea Kern, 249–88. Berlin: De Gruyter, 2014.

Byrne, Alex. “On Misinterpreting Kripke’s Wittgenstein.” Philosophy and Phenomenological Research 56, no. 2 (1996): 339–43.

Coates, Paul. “Kripke’s Sceptical Paradox: Normativeness and Meaning.” Mind 95 (1986): 77–80.

Conant, James. “Two Varieties of Skepticism.” In Rethinking Epistemology, Vol. 2, edited by James Conant and Guenter Abel, 1–76. Berlin: De Gruyter, 2012.

Ebbs, Gary. Rule-Following and Realism. Cambridge, MA: Harvard University Press, 1997.

Forbes, Graeme. “Skepticism and Semantic Knowledge.” Proceedings of the Aristotelian Society 84 (1984): 223–37.

Gampel, Eric H. “The Normativity of Meaning.” Philosophical Studies 86 (1997): 221–42.

Gibbard, Allan. Meaning and Normativity. Oxford: Oxford University Press, 2012.

Ginsborg, Hannah. “Primitive Normativity and Skepticism About Rules.” Journal of Philosophy 108, no. 5 (2011): 227–54.

———. “Going on as One Ought.” Presented as the 2016 Jacobsen Lecture at the University of London. Draft.

Goldfarb, Warren. “Kripke on Wittgenstein on Rules.” Journal of Philosophy 82, no. 9 (1985): 471–88.

Hattiangadi, Anandi. Oughts and Thoughts. Oxford: Oxford University Press, 2007.

Haukioja, Jussi. “Soames and Zalabardo on Kripke’s Wittgenstein.” Grazer Philosophische Studien 64 (2002): 157–73.

Horwich, Paul. “Review of Wittgenstein on Rules and Private Language, by Saul Kripke.” Philosophy of Science 51, no. 1 (1984): 163–71.

———. Wittgenstein’s Metaphilosophy. Oxford: Oxford University Press, 2012.

Kripke, Saul. Wittgenstein on Rules and Private Language. Cambridge, MA: Harvard University Press, 1982.

Kusch, Martin. A Sceptical Guide to Meaning and Rules: Defending Kripke’s Wittgenstein. Chesham, Bucks: Acumen, 2006.

Loar, Brian. “Review of Wittgenstein on Rules and Private Language.” Noûs 19, no. 2 (1985): 273–80.

McDowell, John. “Wittgenstein on Following a Rule.” In Mind, Value and Reality, 221–62. Cambridge, MA: Harvard University Press, 1998. Originally published in Wittgenstein on Following a Rule, edited by Steven Holtzman and Christopher Leich (London and New York: Routledge and Kegan Paul, 1981), 141–162.

———. “Meaning and Intentionality in Wittgenstein’s Later Philosophy.” In Mind, Value and Reality, 263–78. Cambridge, MA: Harvard University Press, 1998. Originally published in Midwest Studies in Philosophy, vol. 17: The Wittgenstein Legacy, edited by Peter A. French, Jr. Theodore E. Uehling and Howard K. Wettstein (Notre Dame: University of Notre Dame Press, 1993), 40–52.

McGinn, Colin. Wittgenstein on Meaning: an Interpretation and Evaluation. Oxford: Blackwell, 1984.

Miller, Alexander. Philosophy of Language. Second edition. London and New York: Routledge, 2007.

———. “Rule-Following Skepticism.” In The Routledge Companion to Epistemology, edited by Sven Bernecker and Duncan Pritchard, 454–63. Oxford and New York: Routledge, 2011.

Millikan, Ruth Garrett. “Truth Rules, Hoverflies, and the Kripke-Wittgenstein Paradox.” Philosophical Review 99, no. 3 (1990): 323–53.

Peacocke, Christopher. “Review of Saul Kripke, Wittgenstein on Rules and Private Language.” The Philosophical Review 93, no. 2 (1984): 263–71.

Soames, Scott. “Skepticism About Meaning.” Canadian Journal of Philosophy Supplementary Volume 23 (1997): 211–49.

Verheggen, Claudine. “Semantic Normativity and Naturalism.” Logique et Analyse 216 (2001): 553–567.

Wright, Crispin. “Kripke’s Account of the Argument against Private Language.” Journal of Philosophy 71, no. 12 (1984): 759–78.

———. “Critical Notice of Colin McGinn’s Wittgenstein on Meaning.” Mind 98 (1989): 289–305.

———. “Wittgenstein’s Rule-Following Considerations and the Central Project of Theoretical Linguistics.” In Rails to Infinity, 170–213. Cambridge, MA: Harvard University Press, 2001. Originally published in Reflections on Chomsky, ed. Alexander George (Oxford: Blackwell 1989), 233–64.

———. Rails to Infinity. Cambridge, MA: Harvard University Press, 2001.

Zalabardo, José L. “Kripke’s Normativity Argument.” Canadian Journal of Philosophy 27, no. 4 (1997): 467–88.

———. “One Strand in the Rule-Following Considerations.” Synthese 171 (2009): 509–15.