We refer to the objective world as if it were essentially independent of the fact that we refer to it. This attitude towards the world is described by the concept of objective knowledge, which simply picks out those of our mental states that put us in touch with how things are. By the ‘world’, we usually understand the totality comprising everything that pre-exists our two most basic epistemic activities: acquiring knowledge of what is the case and securing that knowledge against any possible objections we might encounter. The concept of the world is therefore indispensable to our understanding of whatever it is that we know. For when we know something, such that we can both express what we know in the form of an explicit knowledge claim and defend it in the face of critical questioning, what we usually know is simply how the world is.

When, in our more reflective moments, we try to arrive at a more articulated understanding of this idea, we hit upon a concept of the world as the unified horizon of everything that is the case. On this conception, the world is the object of each successful representation of what is the case. Or, to be more precise: the states of the world and precisely not the world itself – that is, the world simply as the world – are the objects of each successful representation of what is the case. This is the intuitive basis of what Bernard Williams called ‘the absolute concept of reality’:¹ from the standpoint of our knowledge claims, the world itself assumes the status of the absolute, of that which is independent of and prior to knowledge. Our knowledge claims, by contrast, are subject to the condition that they must either describe how the world is or undergo revisions when confronted with critical objections. Indeed, our reference to the objective world makes us fallible precisely because the world, in whatever way it ultimately exists, is independent of our acts of referring. Take away this independence, and our reference would no longer be fallible. But that just means it would no longer constitute reference to an objective world: to cognise is just to be in a fallible position.

Cashing out these general reflections in the form of a developed theory is a much more problematic enterprise than it may seem at first glance. For one thing, it is vital to note how we are already operating on two levels. On the one hand, we need to understand first and foremost that it is the world that we know when we possess empirical knowledge. On the other hand, this very assertion already oversteps the boundaries of objective knowledge, of empirical cognition. Indeed, it does so in a twofold sense, because knowing what knowledge is is not itself a piece of empirical knowledge about how the world is, any more than the world simply as the world can ever become an object of empirical knowledge. Otherwise, the putative piece of knowledge whose content is the proposition that the world precedes our knowledge claims would be just as fallible as our knowledge of a determinate state of the world. Yet this seems to be impossible: knowledge of the conditions of knowledge’s fallibility operates (at least prima facie) at a different theoretical level to the fallible knowledge it thematises. And so, we might think, it must enjoy an exemption from fallibility, else we would be fallible in relation to the very question of whether we are fallible.

It seems, therefore, that the standpoint from which we go about trying to render empirical – and that means fallible – knowledge comprehensible has to differ from the standpoint from which we make empirical knowledge claims. Knowledge of empirical knowledge is not itself empirical. So, since minimal insight into the relation between empirical knowledge and the world cannot itself be subject to inductive verification or falsification, we need to draw a distinction between two theoretical levels: between the level of objective knowledge on the one hand and the meta-level, of knowledge of what objective knowledge is, on the other. However unremarkable this distinction may at first appear, the entirety of the following study is devoted to spelling out its far-reaching implications and to making them fruitful for contemporary epistemology – in particular, for the debate surrounding scepticism. As a theory that investigates truth claims, epistemology itself lays claim to knowledge as soon as it pronounces on how the concept of knowledge should be understood.

However, as an engagement with the problem of scepticism soon teaches us, the knowledge claims of epistemology are far from unproblematic. If the sceptic convinces us that we cannot, after all, know what objective knowledge is, let alone how it is possible, then objective knowledge itself threatens to collapse. For without any epistemic grasp of the very concept of objective knowledge, we can hardly ascertain whether there really is any objective knowledge. There is therefore no way around the question of what it really means to know something. The possibility of knowledge is a methodological presupposition of every instance of empirical knowledge ascription, without itself being empirically knowable.

Perhaps the most important methodological insight of modern epistemology since Descartes is that the theoretical stance of epistemology itself is motivated by scepticism. An engagement with scepticism belongs to the very conditions of epistemology: it first makes space for and then sustains the question of what it means to know something. Generally speaking, philosophical questions of the form ‘What is X?’ acquire their determinacy through being distinguished from what they are not (in this respect, they are just like anything else).² And the question of the nature of knowledge is no different: it is comprehensible only given the possibility that there is something we do not know, or perhaps even cannot know. Knowledge, in other words, acquires its particular conceptual contours precisely in virtue of how we come to distinguish it from its constant companion, ignorance.

This point is echoed in the fundamental epistemological insight recently introduced into the debate by Jonathan Schaffer’s contrastivism.³ According to this position, the content of all cases of (propositional) knowledge is respectively determined by its belonging to a class of propositions that is distinguished from a contrast class. Instead of ‘S knows that p’, we should in fact always read ‘S knows that p rather than q’. The contrast class (q) thereby contains all those propositions whose truth implies the falsity of the propositions of the contrast class (p).⁴ So, the proposition that I, Markus Gabriel, can now see my notebook in front of me belongs to the class of perceptual propositions. If it were the case that all perceptual propositions were false, such that no one ever perceived anything (because, say, we are only ever dreaming, or we are brains in a vat, or …), then the class of dream propositions would be a contrast class of the perceptual propositions.

Knowledge generally contrasts with ignorance. Indeed, the nature of error (ψεῦδος) or ignorance has been one of the central questions of epistemology ever since its origins. It was raised by Plato, especially in the Theaetetus and the Sophist, and it can hardly be a coincidence that the question of knowledge arose in the context of pre-Socratic metaphysics, which paradigmatically marked a distinction between being and seeming that was at once ontological and epistemological.⁵ The development from pre-Socratic metaphysics to the Sophists, against whom Plato developed his theory of ignorance, his ‘pseudology’, is an historical expression of the fact that, in relying on the determinacy of knowledge, the epistemologist thereby owes a debt to the possibility of ignorance. Omnis determinatio est negatio thus applies just as much to the determinacy of the concept as to the objects of knowledge. Whoever claims to know what knowledge is generates an ontological space of opposition within which knowledge contrasts with ignorance, and the continual possibility of the latter becomes the wandering shadow of the former.

I want to elaborate and deepen this connection by pursuing the following thought: scepticism is a condition of intelligibility of epistemology. In its continual confrontation with ignorance, epistemology occupies the specific standpoint of a metatheory. From this metatheoretical standpoint, it investigates the question of what (first-order) knowledge is and, in doing so, claims (second-order) knowledge for itself. We have already begun to distinguish first-order knowledge (empirical cognition) from epistemological knowledge by designating the object of the former as states of the world. Since epistemology itself, by contrast, refers both to the successful unity of knowledge claims and the world (knowledge) and to their potential difference (ignorance), it is continually beyond all empirical knowledge. For the world simply as the world is no more an ordinary object than objective knowledge itself. This, as we shall see, is one of the central insights to be won through our confrontation with scepticism.

The entire first part (Part I) of this study is therefore concerned with the function of scepticism in the dialectical economy of epistemology. By ‘dialectic’, I understand a form of reflection on the connection between the motivation of a theory and its execution. The task of this first section will be to work out the foundational methodological function of scepticism in the construction of (modern) epistemology. Chapters 1 to 4 take up the question of how the concept of the world relates to scepticism. In this context, we will need to draw a distinction between three different conceptions of scepticism, which will be crucial for the arguments to follow. Specifically, we must distinguish between negative dogmatism, Cartesian scepticism and methodological scepticism. The concept of negative dogmatism, I will argue, can be won through an engagement with the sceptical foundations of Kant’s transcendental idealism. For, insofar as he attempts to draw boundaries to knowledge, he puts forward negative dogmatic claims: claims to know that there are some things we cannot know.

Specifically, we will have to critically examine Kant’s conception of the world against the background of his Refutation of Idealism. Kant’s drawing of epistemic boundaries, his negative dogmatism, clearly involves a distinction between two theoretical levels: empirical and transcendental. Nevertheless, the way in which he carries through on the sceptical motivation of his theoretical standpoint does not leave empirical knowledge unaffected: in Kant, the world threatens to vanish into cognition, objectivity into subjectivity. In light of this threat, he seeks to defend his transcendental idealism with the help of the Refutation of Idealism and to show that the objectivity of knowledge, far from being undermined by transcendental idealism, is in fact secured by it. In contrast to a subjective idealism à la Berkeley, transcendental idealism is compatible with the assumed existence of things or objects extended in space and time. Nevertheless, Kant’s own idealism leads him to over-extend the thesis of his Transcendental Aesthetic. As we shall see, he is ultimately unable (within the context of the Refutation of Idealism at any rate) to mark a distinction between a spatial representation and the representation of something spatial. Drawing this distinction requires the introduction of a criterion of publicity, and therefore of other subjects in space and time; these subjects must be able both to refer to the same spatial items and to communicate the fact that they can do so.

Throughout the book, I shall be discussing the problematic of idealism in terms of a distinction, due originally to Robert Brandom, between a thesis of the sense dependence of objectivity on subjectivity and the thesis of a reference dependence of objects on subjects. The distinction guides my arguments right through to the final paragraphs of this study. A concept P is sense-dependent on a concept Q, according to Brandom, if and only if we would not understand P if we did not understand Q. Understanding P presupposes understanding Q. By contrast, Brandom explains, a concept P is reference-dependent on a concept Q if and only if there would be nothing which fell under P if there were nothing which fell under Q.⁶ The concept of ‘idealism’ can now clearly be understood in at least two different ways: as a thesis of sense dependence on the one hand and as a thesis of reference dependence on the other.⁷ Sense dependence idealism merely asserts that we would have no concept of objectivity if we did not distinguish it from our subjectivity. This thesis is a second-order assertion about a condition of our access to the world (and so belongs to the metatheory). Reference dependence idealism, by contrast, asserts that there would be no objects if there were no subjects and is therefore a first-order thesis about what there is – or, rather, about the way in which what there is exists. Kant oscillates between the two assertions. It is certainly true that nobody has operated with a differentiation of theory levels as clearly as Kant does – a fact to which the distinction between the empirical and the transcendental bears ample testimony. Yet his handling of this differentiation is ultimately inconsistent, and he consequently ends up a victim of his negative dogmatism (see chapters 1–2). As we shall see, this reveals itself as the Achilles’ heel of the Refutation of Idealism.

Moore’s overreaction to Kant’s negative dogmatism – his naïve ontology of individuals – forms the next topic of discussion. Moore (quite intentionally) undercuts Kant’s reflections and manages to formulate a decisive objection to the Refutation of Idealism, which I shall elaborate in my own discussion of Kant. Yet Moore neglects both Kant’s conception of the world and his distinction between theory levels. Moreover, he is unable to defend the fundamental category of his naïve ontology, the so-called physical object, against objections stemming from the conceptual relativity of our access to the world. This clears the path for a remobilisation of Kant’s conception of the world without violating the common-sense intuition that all true judgements describe the world as it is in itself – that is, how it is independently of the beings who describe it. Specifically, adopting the Transcendental Dialectic’s conception of the world in combination with a rigorous differentiation of theory levels can help dissolve a confusion lying in the concept of the world. In a word: Kant here shows us how we might distinguish between the world as unified horizon and the world as object of our knowledge. It is unfortunate, therefore, that the Transcendental Dialectic’s conception of the world plays no role in the Refutation of Idealism, the latter occurring at a systematically unfruitful point in the Critique.

At this stage, Cartesian scepticism enters the picture (chapters 5–6). In this context, ‘Cartesian scepticism’ admittedly stands neither for a sceptical position that Descartes himself advocated nor for the concrete form assumed by his sceptical reflections in the Meditations. Nevertheless, as the first to recognise the outlines of its logical structure and to deploy it in a methodologically controlled manner, Descartes bequeathed his name to this variety of scepticism. Before Cartesian scepticism can be developed as a general paradox attending the concept of knowledge or the concept of a justifying reason (chapter 6), we need to set out certain methodological provisos (chapter 5). To this end, I introduce a distinction between logical and dialectical analyses of sceptical arguments. Both methods can then be put to use in order to better appreciate the impetus behind Cartesian scepticism. Furthermore, we shall see how Cartesian scepticism can best be analysed as a paradox – that is, as a set containing seemingly acceptable (and well-motivated) premises, seemingly acceptable (and well-motivated) inference rules, and an obviously unacceptable conclusion.

Our confrontation with the paradox or paradoxes of Cartesian scepticism, however, cannot be restricted to attempts at solving (or dissolving) it. It won’t do, say, just to question or replace one of the paradox’s premises in an effort to circumvent it. Merely logical repair mechanisms will not serve our ultimate purpose here. Rather, we must always ask ourselves questions of another sort: in which theoretical context does the paradox arise? Under which theory conditions can it be introduced – i.e. motivated? In contemporary epistemology, such considerations trade under the name of a ‘theoretical diagnosis’.⁸ They concern the function of Cartesian sceptical premises in the context of determinate theories and make up the focus of what I am designating dialectical analysis. We will test out both methods, logical and dialectical analysis, in the course of a critical engagement with Crispin Wright’s ingenious anti-sceptical strategy of implosion. However, it will turn out in chapter 6 that, in those cases where the strategy is successful, implosion not only disables Cartesian scepticism but also threatens the very epistemological standpoint from which the strategy is itself developed. For its own standpoint is motivated by premises that lead to the formulation of Cartesian scepticism. Accordingly, the first part of the study then ends by constructing a general paradox of Cartesian scepticism. Since we need draw on nothing more than certain foundational premises of discursive rationality to generate this paradox, it represents nothing less than an intrinsic threat to our rational cognition quite generally.

However, the threat can be averted. By reaching for a dialectical manoeuvre, we can restabilise discursive rationality by grounding it in human practice, which is essentially tied to fallible, finite knowledge acquisition. Part II, on contextualism and finitude, is occupied with the prospects for this manoeuvre. Specifically, it endeavours to rebuild epistemology on a contextualist discourse theory by drawing upon the attempts by Wittgenstein and Sextus Empiricus to draw boundaries to our knowledge. It transpires that there is a common denominator shared by Pyrrhonism and the strategy pursued in Wittgenstein’s late philosophy, even though the latter is ultimately anti-sceptical and directed against hyperbolic doubt. This common denominator is contextualism, which I understand as a lesson concerning the necessary finitude of objective knowledge. I begin Part II with a systematic outline of Pyrrhonian scepticism (chapter 7), without, however, entering into Sextus’ late Pyrrhonism in all of its historical details. My sole concern is to provide a systematic reconstruction of the fundamental operation of self-application, the περιτροπή, the ploy of ‘self-refutation’ or of ‘turning the tables’. This operation consists in applying our insight into the finitude of knowledge to itself and thereby, however paradoxically, in drawing boundaries to our very insight into the boundaries of knowledge.

Chapters 8 to 10 then reconstruct Wittgenstein’s contextualism as a theory of assertoric content. It will turn out that the possibility of all discursive determinacy – that is, of all assertoric content – presupposes discursive operating conditions, conditions which discourses cannot have in their self-conscious possession in ipso actu operandi. To justify this thesis, I focus especially on the problem of rule-following, while at the same time connecting Wittgenstein’s position with certain basic concepts of system theory, primarily of a Luhmannian provenance. Wittgenstein and Luhmann are in agreement insofar as they argue for the necessary finitude of all operations of observation – that is, of all operations of determinacy: whatever can be something determinate for a discursive community counts as something determinate only under the presupposition of historically variable parameters, which respectively fix what a community can register. The community constitutes a discourse precisely by fixing hinges upon which all individual moves in the discourse turn. In my reading, this is the most important thesis of Wittgenstein’s On Certainty, and I will interpret it as an insistence on the necessary finitude of discourse. This insistence, it should be noted, can already be found in Sextus, albeit in a different context.

Chapters 9 to 12 discuss Wittgenstein’s private language argument in order to provide a more precise evaluation of the finitude of all discursively communicable knowledge. On my interpretation, which takes its cue above all from Crispin Wright, Saul Kripke and Meredith Williams, the argument tries to prove that all discursively communicable knowledge is finite; knowledge depends upon the stability of a discourse, which has to be fixed as the context of knowledge ascription. Nevertheless, the participants in a discourse cannot determine its foundations from within the discourse itself – not without initiating a meta-discourse, which brings along its own presuppositions or operating conditions in turn. One result of this line of thought is that it is impossible to determine when and whether someone knows something from an absolute standpoint. Yet Wittgenstein shows us how we might put this partially sceptical thesis to use in order to develop a contextualism that manages to reconstruct scepticism as a harmless lesson about our discursive finitude.

Chapter 11 then places John McDowell’s disjunctivism in this argumentative context, discussing it as an anti-sceptical strategy. Disjunctivism attempts to develop knowledge or cognition under the conditions of a theory of intentionality, understanding the latter as a cognitive relation of mind and world rather than, as in Wittgenstein, as a necessarily social and, in this sense, discursive phenomenon. Following Wittgenstein, I attempt to show that McDowell’s disjunctivism faces difficulties as an anti-sceptical strategy, since he fails to integrate the socio-semantic dimension of subjectivity into his approach from the outset. He determines the subject as cognitive intentionality and not as a person in space and time (who is always already socially implicated).

Chapter 12 contains a response to the rule-following problem developed by Kripke in his interpretation of Wittgenstein. It uses some of the conceptual cornerstones of Kripke’s suggestion that we might need a ‘sceptical solution’ in the guise of a community view of rule-following in order to understand how anyone can ever actually follow a rule. I turn this slightly desperate diagnosis into a more positive account of the social nature of thought and theorising. In other words, we need to give up methodological solipsism in epistemology and stop thinking of the problem of knowledge in terms of an attempt of a single knower (typically called ‘S’) to put herself in touch with objective matters of fact (typically called ‘p’) by claiming that they obtain. Knowers turn into such individuals only by being singled out from a community of successful knowers. Epistemology should thus return from its self-inflicted Robinson Crusoelike isolation to the actual context of knowledge acquisition. I call this move ‘stage setting’.

Chapter 13 then enquires into the relationship between the concept of representation, which traditionally leads to the familiar sceptical aporia of mental representationalism and Cartesian scepticism. Since Wittgenstein ultimately uses his contextualism to undermine the methodological solipsism of modern epistemology – a position accepted by so many post-Cartesian (as well as ancient) epistemologists – contextualism would seem to offer an escape route from the general paradox of Cartesian scepticism. By uncovering the problem of rule-following and its communitarian solution (which I discuss extensively in chapter 10), the solipsistic I can be shown its way out of the ‘fly bottle’.⁹ The concept of representation thus disappears – but at a price: our subjectivity and our metatheoretical standpoint, both of which have to be invoked when we talk about discourses as such, come under serious threat.

This difficulty becomes especially clear when we take a closer look at Wittgenstein’s liberal naturalism (chapter 14). In order to safeguard the possibility of mutual understanding between different discourses (language games), Wittgenstein introduces ‘second-nature’, which functions as a kind of unifying horizon of all discourses. Humans can communicate with one another because of their shared natural history, and ‘very general facts of nature’¹⁰ ensure that there is ultimately nothing alien within the domain of the human. Indeed, Wittgenstein is also explicit that all knowledge comes about only ‘by the grace of nature’.¹¹ This position, however, is incompatible both with Wittgenstein’s own motivating contextualism and, as we shall see, with the vital sceptical lesson of finitude. Specifically, it is a position that fails to enquire into the operating conditions of the very discourse in which the claim that our nature is constituted in such and such a way enjoys the validity it does. In other words, Wittgenstein does not think to apply his contextualism to his own reflections (περιτροπή).

As a response to this shortcoming, I endeavour to push Pyrrhonian scepticism through to its logical endpoint while still conceiving of the discourse about finitude as a finite discourse. This attempt throws up the obvious problem that the resultant metatheory has to concede its own revisability and contingency. To be sure, this does not mean that it has to be false or self-undermining (in the manner of a performative contradiction). But it does mean that it has to be distinguished from an alternative of which it necessarily remains ignorant; it cannot know its own contrast class. Given the principle that all determinacy in logical space generates binary oppositions, applying our insight into the finitude of objective knowledge to itself means erecting a boundary between this knowledge and an ignorance that, however indeterminable, is an indispensable assumption.

Lastly, by examining Brandom’s interpretation of Hegel, chapter 15 ventures a final attempt to rescue the concept of the world from the contingency of the metatheory. We will see that Brandom’s interpretation does not take adequate account of the absolute idealism adumbrated in the Science of Logic and, instead, operates with a concept of the world that Hegel wants to supersede. To bring out this point, it will pay to play off some of the basic moves of Hegel’s logic of reflection against Brandom’s version of an objective idealism. This will finally bring into view the possibility of using Pyrrhonian scepticism to develop a discourse theory whose task consists ‘merely’ in examining the dialectical consistency of given theories. Once absolute idealism has been liberated from certain absurd prejudices (a task to which this book can unfortunately make only a small contribution), one can understand such an approach as delivering on a Hegelian thesis. Taking our cue from Hegel helps provide a basis for a systematic project: of setting out a methodology for epistemology that does not so much seek to solve supposedly grave sceptical problems as merely to delineate the dialectical topography of their possible solutions. With such a method, we might then evaluate their respective prospects for success.

I see this model of philosophical methodology, generated via reflection on the finitude of knowledge, as a means of inheriting Hegel’s talk of ‘absolute knowledge’, particularly in his Phenomenology of Spirit. I therefore conclude with an attempt to show, at least in outline, that the paradigm of philosophy as the unity of method and object – that is, of self-thinking activity in the medium of pure thought – makes an unexpected return within the contemporary scepticism debate in a way that lends renewed impetus to Hegel’s programme of absolute knowledge. In this connection, it is already worth noting, even at this early stage, that absolute knowledge is not infinite knowledge. Absolute knowing too has to fail on account of its higher-order immediacy, its own finitude – which is why, of course, absolute knowing does not represent the last word even for Hegel.¹²

Given the length and breadth of what is to follow, the above will have to suffice as a promissory overview. Yet, as an aid to the reader’s orientation, it is worth adding that Part I contains a largely destructive discussion of the concept of knowledge and possible anti-sceptical strategies, while Part II pursues a more constructive project, building contextualism on the ruins of Part I, but without raising objections against the truth of Pyrrhonian scepticism and its lesson regarding our necessary finitude. Epistemology confronts us with our discursive finitude, a finitude we cannot transcend. But, as I try to show elsewhere, this does not rule out metaphysical theories of infinity, provided that these theories conceive their task aright.¹³

A final word on my approach: I work on the understanding that it is legitimate to combine strategies from ‘analytic’ philosophy – primarily within epistemology, though also from philosophy of mind – with traditional questions of metaphysics and the theory of knowledge. In any case, it is hardly self-evident that the still widespread practice of opposing analytic to continental philosophy is worth sustaining. It would be wrong to classify the arguments that follow by trying to ascertain their respective debts to particular schools; indeed, they do not even presuppose that overgeneralising titles such as ‘analytic’ and ‘continental’ allow us to draw anything like adequate distinctions between different methods and schools of thought. The excessive professionalisation of the practice of philosophy in our time does not mean that philosophy itself is a professionalised science that divides neatly into different disciplines, each marked out by its given characteristic methods. For the purposes of classifying the business of modern academic institutions, a strict organisation of philosophical discourse may be acceptable, even advantageous, to a certain extent. But it has the potential to become orthogonal to philosophy as a discursive practice of freedom. For this reason, in trying to determine the function of scepticism in epistemology, I have tried to develop my own position by taking up the approaches of highly diverse thinkers, without worrying about whether I was philosophising within the accepted house style of any particular school.