1. The Frustrations of Epistemology

1.1. We can distinguish two uses of the term ‘epistemology’. In its broadest use, it refers to philosophical explorations of the nature and possibility of knowledge, an enterprise in which many, if not all, great philosophers have been involved. In its narrower use, ‘epistemology’ refers to a particular branch of Western philosophy, which explores the nature and possibility of knowledge by certain characteristic analytic methods, designed, on the one hand, to establish the conditions that must be fulfilled for someone to possess knowledge (i.e. to define or analyse what knowledge is), and, on the other, to establish, against the sceptic, that we do in fact have knowledge, at least in some domains of inquiry. While epistemology in the broad sense is an inevitable part of any serious philosophical undertaking, the practice of epistemology more narrowly understood is peculiarly frustrating and paradoxical, so much so that some philosophers feel it must be, as Charles Taylor put it, ‘overcome’.Footnote ² John McDowell's Mind and World, for example, is certainly a work of epistemology in the broad sense, but I think McDowell would be insulted if it were mistaken for a contribution to epistemology more narrowly conceived.Footnote ³ Those, inside or outside philosophy, who bemoan the failings of the analytic tradition, tend to invoke epistemology as Exhibit A; hence the not-uncommon belief that epistemology must be transformed or transcended if that tradition is to be redeemed.

The frustrations of epistemology (here and henceforth, I use the term principally in its narrow sense) are no secret to students of philosophy since almost everyone begins their philosophical education with matters epistemological. The rationale for this is not the kind of intellectual-existential crisis which beset Descartes, who could plausibly represent himself as grappling with the fact that almost everything of substance he had learnt in school was false, a predicament that might reasonably provoke a thinker into the project of systematically reforming knowledge. The contemporary rationale is typically more modest: philosophical inquiry in any area aspires to make knowledge claims, hence we should start by exploring the conditions for knowledge. That seems reasonable. How disappointing, then, when we discover that philosophers have enormous difficulty saying what knowledge is and even more difficulty showing that we have any. We are very much better at constructing sceptical arguments than refuting them, and epistemological inquiries resemble the plots of British mysteries, where, typically, the set-up is so much better than the resolution. The puzzle we set ourselves to solve is fascinating, but no solution is remotely as gripping. Of course, there are sometimes periods of optimism when a new approach emerges – externalism, for example – and then much is written in the journals on questions of ever-increasing technical complexity. But such optimism has a habit of waning and then the spilling of ink on technicalities seems otiose.

Knowledge, then, makes a lot of trouble for philosophy. This in itself might seem unremarkable; indeed, to be expected. Philosophy has a hard time with many fundamental concepts, so why should knowledge be any different? However, what is odd is that the concept of knowledge causes us little or no trouble in what philosophers are wont to call ‘ordinary life’. Of course, the acquisition of knowledge causes us problems. Ignorance can be hard to overcome. It is often the case that we think we know when we don't, can't establish what is the case, don't know how to interpret evidence, and so on. But these familiar problems with knowledge never threaten to loosen our grip on the concept of knowledge itself. We are secure in our use of that concept, and of cognate concepts like justification. And it is not that we are simply blind to the kind of considerations that philosophers spin into distinctively philosophical problems with knowledge. For example, in ordinary life, we are alive to the failings of the justificatory strategies presented in the Agrippan trilemma. We are quick to protest against putative justifications that move in a circle, or invite a regress, or make dogmatic assumptions. Such protests can be overheard in conversations in any pub on any evening. But, in such contexts, no-one thinks there is a trilemma, from which there is no exit. We all have a strong grip on what a sound justification is like, and that grip refuses to be shaken no matter how much trouble with knowledge we actually encounter. Engaged in some inquiry, we might recognise that we don't know how to justify some claim, or even how to begin to justify it. But such concerns do not provoke us to throw up our hands about the concept of justification. On the contrary, in such a case we know how far short we are of a justification because we know what a justification would look like.

Such is not true of many other concepts we turn to philosophy to clarify. Concepts such as justice, identity, personhood, agency and responsibility can be really troublesome in ordinary life. And such trouble sometimes threatens to loosen our grip on the particular concept in question. Some years ago, my mother was stricken by a debilitating stroke. Being forced to reflect on the life of someone in such a condition can force one to question one's understanding of consciousness, agency and personhood. Similarly, in ordinary life, we may doubt that we know what justice requires in some situation, and do so in a way that reflects uncertainty about what justice is. Such concepts don't only make trouble in philosophy. But knowledge, it seems, makes a huge amount of philosophical trouble, but precious little trouble in ordinary life.

This, I think, is a puzzle that any serious philosophical account of knowledge needs to reckon with. Let's call it the conundrum of epistemology. How can it be that a concept that presents so few problems in ordinary life makes such philosophical mischief? Answering that question is my first criterion of adequacy for philosophical accounts of knowledge.

1.2. One way of responding to the conundrum would be to argue that the philosophical trouble is made by philosophers setting unduly high standards. We cannot refute sceptical arguments, but that's because they operate with too demanding a conception of knowledge. We need to lower the bar or, to change the figure, move the goalposts. We can drop the notion of knowledge altogether and settle for justified belief. Or we can preserve the notion of knowledge and reconceive its elements, weakening the notions of truth and justification.

Such strategies have the significant disadvantages. Those who forsake the concept of knowledge seek to avoid scepticism by running into its arms. And those who would reconceive truth and justification attempt to alleviate philosophical trouble by doing violence to the very conceptions with which we are so at ease in everyday circumstances. Richard Rorty's particular way of overcoming epistemology, at its most radical, treated ‘true’ as a term of approbation, ‘a compliment we pay to sentences seen to be paying their way’,Footnote ⁴ and construed justification ultimately in terms of communal agreement or ‘solidarity’. But in ordinary life, we have a firm grip on the distinction between being true and being widely endorsed, and the distinction between justification and communal agreement. Philosophy needs to respect those distinctions, not obliterate them. Indeed, the arguments for eliding them are, ironically, sceptical in character. So philosophical trouble is not so much avoided as put in the service of a revisionist account of our ordinary conceptions. The result is only to trade one unsatisfying approach to knowledge for another.

The fact is that if you are to know that p, you must have a true belief that p, and your belief must have the right kind of grounds. Since knowing that p is incompatible with error, you cannot know that p if, for all that you know, p might be false. It follows that the right kind of grounds are grounds that ensure you are not wrong. These ideas are central to our concept of knowledge – that concept, in the use of which we are so secure – and epistemology should take them at face value.

So, a further criterion of adequacy is that philosophical accounts of knowledge should not lower the bar or move the goalposts, but provide an account of the nature and possibility of knowledge, as we know it to be, and of ourselves as knowers, as we know ourselves to be. This means that an adequate account of knowledge cannot represent the reasons we give when, say, we justify our perceptual beliefs, as standing in need of supplementation by philosophy (or anything else). We need an account on which, when I tell you I know there was a magpie in the garden because I saw one there, the grounds I offer, as I offer them, can suffice for knowledge.

1.3. Any such account will obviously have to deal with the sceptic. Notwithstanding my respect for what we might call our ‘ordinary epistemic practices’, a philosophical account of knowledge cannot simply affirm them and scorn all those who would call them into question. It is vital that we properly understand the nature of scepticism and the distinctive allure of sceptical arguments. Such arguments have a peculiar fascination, so much so that the capacity to be seized by sceptical doubt is sometimes taken as a sign of philosophical aptitude. (We tend to write off students who don't see the point of the sceptic's machinations, even if by the end of the course we're trying to convince them that sceptical arguments cannot carry conviction.) At the same time, it is not uncommon to portray scepticism as something pathological, so that propensity for philosophy involves susceptibility to an affliction. Hume famously describes sceptical doubt as ‘a malady, which can never be radically cured; it always increases the further we carry our reflections’.Footnote ⁵ For him, the sceptical impulse is of a piece with philosophical thinking and can be quietened only by the distractions of everyday life. Others hope that reason can somehow heal itself.

What is the disease like? Sometimes it manifests as a kind of obsessive-compulsive questioning of our grounds for belief, a refusal to be satisfied that our putative grounds are what we take them to be – good enough for knowledge. Sometimes it appears as a pervasive anxiety about being cut off from reality, a paranoid preoccupation that things might be not as they seem. The deep question is whether the affliction is internal to reason itself, as Hume thought, or whether the infection comes from without. In any event, we think of susceptibility to this disease as part of philosophical consciousness and insist that indifference to scepticism is an entitlement only survivors can claim. So, my third criterion of adequacy is that philosophical reflection on the nature of knowledge must not only refute or disarm the sceptic. It must also understand the reality of sceptical doubt and its peculiar magnetism.

1.4. Finally, the philosophy of knowledge must meet these criteria of adequacy in a way that brings knowledge into view as a socio-historical reality. We must understand how knowledge is acquired, shared and preserved, and to that end we need a philosophy of teaching and learning that countenances, not just such facts as that one person may transmit knowledge to others, but the cultivation of the power of knowledge itself. We also need to understand knowledge in relation to activity, transcending wooden oppositions between knowing-that and knowing-how, to embrace fully the pervasiveness of knowledge in the life-activity of the human subject, from ‘embodied coping’ to the most rarified and abstract forms of contemplative thinking. One might say that the mark of our life-form is living-with-knowledge, a circumstance that only heightens the irony of epistemology's seemingly unremitting trouble with knowledge. The task, however, is not to find ways of setting aside philosophical confusion so that we may simply acquiesce in the practices of ordinary life. Rather, we must settle accounts with epistemology narrowly conceived in a way that opens up new horizons for the philosophical exploration of knowledge. Epistemology, done properly, should not be preoccupied merely with redeeming what we already know; it should rather strive to increase our knowledge of ourselves and our place in the world.

2. Kern's Sources of Knowledge

2.1. I have set out four criteria a philosophical account of knowledge must meet. It must (i) address the conundrum of epistemology in a satisfying way; (ii) provide an account consistent with what we know knowledge to be, resisting the temptation to drop knowledge in favour of a weaker conception or preserve knowledge only by offering revisionist accounts of truth, justification, objectivity, etc.; (iii) understand the nature of scepticism and its charisma; and (iv) bring knowledge into view as a socio-historical reality.

It might be complained that my characterisation of the frustrations of epistemology is belied by recent developments. Has there not been fascinating recent research in social epistemology, a new synergy between epistemology and cognitive science, and exciting work at the border of epistemology and ethics and political philosophy? Moreover, when one takes into account the increasingly porous borders between the analytic and European traditions in philosophy, things are surely not as bad as I make them appear. This may be. Yet, with some notable exceptions, much recent work either bypasses the traditional problems of epistemology or proceeds in the shadow of structures of thinking that are the source of epistemology's discontents and from which we have so far failed to liberate ourselves. In my view, our best hope for liberation rests with McDowell, who in Mind and World and various papers before and since, comes closest to setting us free. But the quietistic tenor of McDowell's philosophy means that he tends to seek emancipation in the form of a release from the obligation to say anything constructive, and that means he falls short of meeting at least some of my criteria.

It was with admiration and gratitude, then, that I read Andrea Kern's Sources of Knowledge, a work consistent with McDowell's insights – indeed, much inspired by McDowellian ideas – but seemingly more expansive in its aspirations to found a substantive philosophy of knowledge. This is a work that attempts to do much that needs to be done, and does it with style.

2.2. According to Kern, the trouble with knowledge arises from the difficulty of reconciling two thoughts, a difficulty she calls the ‘paradox of epistemology’. The first is the claim I endorsed above: since knowing is incompatible with being wrong, the kind of justification one needs for knowledge must guarantee the truth of what is known. The second is that human beings are fallible. We are prone to error. Moreover, we sometimes make errors we don't know we've made, and make them in spite of taking every reasonable precaution not to. But if knowledge is ‘a self-conscious act’, as Kern affirms, then a subject who has knowledge must know she has truth-guaranteeing grounds. It would thus seem that the subject must be able to rule out the possibility that she only thinks she has such grounds when really she doesn't. How can she do that, if she is fallible? This is the problem that familiar sceptical arguments are typically designed to illustrate. If we can't solve the problem, then we will have to admit that there is no such thing as a truth-guaranteeing ground. The very best grounds we can have are merely appearances that cannot guarantee the reality of what appears: they stop short of the facts, as McDowell might say. But if that is true, we can't do justice to the first thought. So we're in trouble.

Kern rejects the standard responses to the problem. Since there is no future in infallibility, some philosophers elect to settle for something less than knowledge as the first thought defines it. Others choose to deny that knowledge is a self-conscious act and embrace the externalist thesis that grounds sufficient for knowledge can be beyond the cognitive reach of the knower. Both strategies, Kern holds, are versions of scepticism rather than antidotes to it.

The right approach, she maintains, is to argue that we do in fact have truth-guaranteeing grounds for knowledge, which we are aware of as such. Such grounds, she claims, are familiar to us. You ask me how I know that there are blue jays in Kingston and I reply that I can see one in the garden. I ask you if you know what a blue jay looks like and you say, ‘Yes, I remember them in my parents’ garden in Sudbury’. George phones from Seville and asks if the blue jays are nesting in his garden again this year and I, having seen them, tell him ‘Yes, they are’, and now George knows. Perception, memory and testimony all traffic in truth-guaranteeing grounds, or at least perceiving, remembering and informing do. Such acts are ‘factive’: they cannot be done unless what is perceived, remembered or conveyed is true.

If this move can be made to work, then propositional knowledge is what we always thought it was: true belief, justified in such a way as to secure the believer's grip on nothing less than how things are. Hence there is no longer an incentive to lower the bar or move the goalposts, and my second criterion of adequacy is met. But more than this, Kern's position helps resolve what I called the conundrum of epistemology, and thus addresses my first criterion. For, on her view, the familiar reasons we give in the ordinary circumstances of life are just the kind of truth-guaranteeing grounds on which knowledge depends, and, since knowing is a self-conscious act, we must know them to be such. So it should not be a surprise that (i) we have no trouble with knowledge in the everyday (other than the usual, non-philosophical trouble about deciding what is the case), and (ii), as Hume beautifully observed, philosophical scepticism cannot carry conviction in the face of life. The trouble with knowledge is a distinctly philosophical aberration, the result, as a Wittgensteinian might say, of the bewitchment of our intelligence by an overreaction to our fallibility.

Kern's approach, however, is not marked by the kind of quietism sometimes associated with Wittgenstein and McDowell, in which philosophy's role is principally therapeutic. On a number of occasions, she reminds her reader that it is one thing to diagnose weaknesses in an opponent's position, or to show some philosophical stance possible, and quite another rigorously to establish one's view, the latter requiring one shows, not just that one's position is true but that it must be so. So she doesn't just point out that we are familiar with truth-guaranteeing grounds, she acknowledges that she must embrace and defend a number of substantive theses, such as the view that human perceptual experience is essentially conceptual in content, and perceptual judgement non-inferential in character, so that there need be no work for knowers to do upon the deliverances of experience to turn them into truth-guaranteeing grounds for their beliefs. She also maintains that, to combine truth-guaranteeing grounds with fallibilism, we must embrace a disjunctive view of experience. And she further argues that the concept of belief depends upon the concept of truth-guaranteeing grounds, from which it follows that to reject the latter is to undermine the very idea of taking the world to be one way rather than another, thereby wrecking not just the concept of knowledge as traditionally understood, but any epistemically less ambitious substitute.

2.3. All this sets up the critical turn in Kern's argument: the turn to knowledge as a rational capacity. On this view, knowledge and its cognate concepts (sensory impression, belief, ground, truth, etc.) do not form what she calls ‘an analytic unity’, so that knowledge can be analysed without remainder into, e.g., true, justified, belief. The rational capacity for knowledge is rather a constitutive unity, such that the concepts of belief, truth, and grounds ‘are intelligible [only] in virtue of the unity they form in the description of knowledge’.Footnote ⁶ As Kern puts it, ‘a rational capacity is a unity of acts that is logically prior to the several acts that make it up’.Footnote ⁷ So just as we cannot understand the individual acts that comprise a person's playing soccer (taking a corner, shooting at goal, strategically fouling an opponent, making a slide tackle, etc.) independently of the idea of the rational capacity to play that game, so we cannot understand believing, being justified, and so on, except as acts manifesting the rational capacity to know.

Kern persuasively argues that appeal to rational capacities is explanatory – we explain particular acts as issuing from the exercise of the capacity in question. These explanations are normative in character. To see an act as realised by a capacity introduces a standard for assessing the correctness of the behaviour in question: the rational capacity determines not just the identity of the act, but its normative standing. Since this must be appreciated by the agent herself, whose activity is guided by her understanding of the appropriate norms and who will explain what she is doing as manifesting the capacity in question, rational capacities are essentially self-conscious. So, a rational capacity for knowledge, like all other rational capacities – e.g. dancing, reading, skiing, playing hockey, playing the guitar, etc. – is constitutive, explanatory, normative and self-conscious.

There is much here that is controversial. For example, it follows that nothing can be a believer unless it is a knower, and it can be a knower only if it is capable of self-consciously understanding and explaining its actions as falling under norms inherent in the rational capacity for knowledge. This sets a high bar for knowledge that non-human animals do not meet. Not everyone will be comfortable with that conclusion. At the same time, however, Kern's ‘turn to capacity’ has two striking advantages. First, it helps us understand fallibility. Philosophers, if they don't go so far as saying that knowledge itself is fallible,Footnote ⁸ typically understand fallibilism as the view that ‘one can know that p even though one's justification for p is less than conclusive’ (BonJour) and ‘does not guarantee the truth of one's belief that p’ (Reed), or that ‘the level of justification requisite for knowing that p is compatible with p being false’ (Brown), all views that Kern thinks no better than the sceptic's.Footnote ⁹ If we see knowledge as a rational capacity, however, we can say, very plausibly, that it is our capacity for knowledge that is fallible, not its acts. My knowing that it is snowing isn't fallible, nor can I be said to know it's snowing if my justification leaves it open whether it is snowing. What is fallible is my rational capacity for knowledge, the exercise of which can go awry or be thwarted by circumstance, leaving me ignorant, sometimes unknowingly ignorant. But from that it does not follow that, when all goes well, my grounds for belief fail to guarantee that I know what is the case. Similarly, when an expert penalty-taker misses from the spot, we do not conclude that her capacity to score never suffices to explain her scoring. The light these considerations cast on fallibility in turn serves to illuminate the allure of scepticism, insofar as the sceptic misrepresents the significance of the possibility of error, and hence they help address my second criterion.

A further attraction of the capacity approach is that it opens up the possibility of understanding knowledge as a socio-historical reality, thereby satisfying my fourth criterion. The capacity for knowledge is one of numerous rational capacities partially constitutive of the human life-form, albeit the most fundamental insofar as the exercise of any rational capacity presupposes that the subject knows what she is doing. The rational capacity for knowledge is in this sense universal. Thus, Kern's position inevitably brings into view questions of development and education. We cannot but ask how the capacity emerges in the individual and is formed and cultivated by experience, upbringing and tutelage.

3. Questions and Concerns

Let me now gently raise some questions and concerns.

3.1. There is, of course, a precedent for many of Kern's views in McDowell's philosophy, and so we might ask (to put it bluntly): What does Kern give us that McDowell does not? Earlier, I contrasted Kern's philosophical rigorism with McDowell's quietism. McDowell is famous for eschewing ‘constructive’ philosophical theorising, preferring instead to deploy philosophical argument to make space for us to say what it comes naturally to us to say. This is particularly evident in his attitude to scepticism, where he argues that the ‘mere intelligibility’ of the idea of perception affording ‘openness to the facts’ earns us ‘the intellectual right to shrug our shoulders at sceptical questions’.Footnote ¹⁰ As he puts it, ‘[t]he aim here is not to answer sceptical questions, but to begin to see how it might be intellectually respectable to ignore them, to treat them as unreal, in the way that common sense has always wanted to’.Footnote ¹¹ Of course, McDowell is willing to engage in a great deal of subtle philosophical argument on the way to finding peace, but his view of the conceptual content of experience, and his recommendation that we treat veridical and non-veridical experience disjunctively, are not really components of a substantive theory of knowledge, so much as positions we must embrace to protect the conception of knowledge implicit in our ordinary forms of thought and talk. We might say these positions are antidotes to constructive theories rather than theories themselves. His disjunctive account, for example, is the name for the space of possibilities that comes into view when one rejects what he calls the ‘highest common factor’ (HCF) account of experience – the idea, ensconced in the argument from illusion and similar sceptical forays, that the content and epistemic significance of experience can be identical in veridical and non-veridical cases (Kern calls this the ‘common ground-thesis’Footnote ¹²). To reject the HCF account, is to make an adjustment in our thinking that removes certain impediments to saying what we want to say. But I don't think McDowell sees this as an invitation to theorise (e.g. to write books on disjunctivism). It's hard to tell exactly where Kern stands on such methodological questions. She clearly has more sympathy for constructive philosophy than McDowell, but I am uncertain just how far she thinks defending and developing her position would require, say, the development of a substantive conceptualist-disjunctivist theory of experience.

Disjunctivist approaches look as if they need theoretical development because they fly in the face of some deep-rooted philosophical prejudices and hence provoke philosophical quarrels. Finding such approaches counter-intuitive, their opponents suppose that it will take some philosophical heavy-lifting to render them plausible, or even intelligible. It is worth observing, then, that one of the reasons knowledge makes so little trouble in ordinary life is that we are quite at home with broadly disjunctive understandings of epistemically erroneous experience.Footnote ¹³ Consider dreaming, for example. We all recognise that when we are awake, we know that we are. And we know this, not because of some introspective sense of ‘being awake’, but because we know that we are having perceptual experiences of our immediate environment (e.g. I know I'm awake because I'm seeing my laptop) and we know what we up to in that environment (e.g. I know I'm writing a paper on my laptop). Nothing could convince you now that you might be asleep and dreaming you're reading this (not even the dream-like incoherence of the argument), even though you will admit that were you dreaming you were reading a philosophical essay you (typically) would not know that you were dreaming. We treat the two states asymmetrically, and we simply don't allow the acknowledgement of our fallibility to turn malignant. We never think we might be dreaming when we are not, except in certain very specific circumstances, where we find it impossible to believe what is happening, such as in receiving unexpected and tragic news. Otherwise, we are at ease with our fallibility and with the admission that there are circumstances in which we take ourselves to have knowledge when in fact we are confronted with mere appearance. Nothing here rests on the fact that dream ‘experience’ is phenomenologically unlike real experience, since that is something one can acknowledge only if one knows one is awake. And in any case, we are also at ease with perceptual illusions, errors and failures of judgment. We are all too aware that we sometimes get things wrong and, when we do, we do not know this and perhaps cannot discover it, at least on the basis of what we can perceive at the time. But the fact that we sometimes unknowingly labour in the realm of mere appearance doesn't undermine our confidence that at other times reality is within perceptual reach. Sometimes we perceive how things are, sometimes we merely think we do. That's what it is to be fallible. The fact that, in ordinary life, this truth does not make trouble testifies, not to the philosophical naïveté of everyday consciousness, but to the cogency of philosophical views that can countenance and explain this. Kern certainly gives the impression that such an explanation requires us to engage in constructive philosophy of a kind that makes McDowell wary, but just how far and to what degree remains uncertain to me.

3.2. Of course it can be hard to see, when one is in philosophical mood at least, that the disjunctive account really silences the sceptic. In the case of all other capacities, there is no logical obstacle to our discerning when they let us down. But when it comes to knowledge, as we have already said, we have to admit the possibility of cases where we're wrong, don't know that we are, and may be unable to find out. I can mistake a defective exercise of the rational capacity for knowledge for a non-defective one. So now it looks as if, in any putative instance of perceptual knowledge, I can't rule out that this is a case when I only appear to have a truth-guaranteeing ground when in fact I don't. But then, for all that I know I might be wrong, and the disjunctive account leaves us no closer to knowledge. One response, articulated by Sebastian Rödl in Self-Consciousness and endorsed by McDowell, is to grant that when I am fooled I do not (of course) know that I am, but to deny that it follows that when I am not fooled, I do not know that I am not.Footnote ¹⁴ When I know there is a magpie in the garden by perceiving one there, I do so in virtue of my having a truth-guaranteeing ground that I know myself to have (the fact that I see the magpie). When I am fooled, my grounds for belief are not that I see a magpie, since there is no magpie to be seen. There is no perception, just mere appearance. So we can distinguish the situation where I know from one in which I am fooled, because in the two cases my grounds for belief are quite different.

This response tends to divide readers between those who think it conclusive, and those who think it useless. The latter group will likely say: Of course, we can ‘distinguish’ the two cases, the one where (a) our belief is grounded in awareness of how things are, the other (b) where this is not so. The problem is that, in the moment, we cannot tell which case we are in because the two are reflectively indistinguishable by the subject. So we seem to be no further ahead. However, the objection misunderstands the significance of the disjunctive account. The account is designed to show that direct perceptual access to facts about the environment is intelligible, and hence so too is the notion of truth-guaranteeing grounds. That undercuts the HCF account of experience and, with it, the sceptic's contention that the best grounds we can have leave it open whether things are the way those grounds present them as being. This undermines the paradox of epistemology and shows that knowledge is possible. The question of whether, in some or other situation, the subject actually has knowledge is then to be answered by appeal to the ordinary ways we decide such things.

One can fail to see this because one mistakenly takes the objection to show that the subject's grounds for belief are no better than the whole disjunction (‘either (a) I see there is a magpie or (b) it merely appears to me that there is a magpie’), and then the subject looks in need of a further criterion to decide the matter. But the point of the disjunctive account is to that the subject's grounds are (one of the other of) either (a) or (b), and if (a), then she has knowledge. The question of how she decides which it is is not to be answered by some general philosophical argument against scepticism, but, in the case at hand, by the usual ways we discern whether a bird is a magpie.Footnote ¹⁵

Another example. Sebastian claims there is orange juice in the fridge, saying he saw a carton there. But Sebastian has been wrong before. How does he know that this is not one of those ‘bad’ cases? This question might have a place in ordinary epistemic contexts. Being absent-minded, Sebastian occasionally puts the empty carton back in the fridge. That circumstance makes it reasonable to ask him to go have another look. Of course, that second look will be fallible, so we might have to ask him to check again. But there will quickly come a point where doubt is epistemically inappropriate, because entirely unmotivated. Of course, our fallibility means that nothing can rule out the possibility that – even when we're drinking what we take to be orange juice – this just might be a case where we have things wrong in spite of our best efforts, nothing, that is, except it's not in fact being such a case. But so what? That doesn't mean that, when we are not wrong, our grounds are not such as to secure that we know, and know that we do. So all we need do, philosophically, is distinguish the cases as Rödl and McDowell do, and then, in epistemic practice, do our best to ensure that we exercise our rational capacity for knowledge successfully, and live with the occasional frustrating consequences of our fallibility.

There is a species of argument, which Kern rejects, that I think can help halt the slide from fallibilism into scepticism. This is the kind of argument that attempts to show that most of our beliefs must be true. Such an argument is famously offered by Davidson (I will forego the details, which turn on the preconditions of interpretation and mutual understanding).Footnote ¹⁶ Kern thinks the argument no good because it can't show that any particular belief is true. And so it can't, but such arguments can nevertheless play a valuable role in the dialectic. Davidson's, for instance, purports to show that the content of perceptual beliefs is world-involving, a strategy that, like McDowell's, attempts to attack what Kern calls the sceptic's ‘unworlding of grounds’. Moreover, the argument can be deployed, not as a head-on refutation of scepticism, but as a reason to think there is no sceptical case to answer.

For example, at the point where scepticism about Sebastian's claims that there is juice in the fridge ceases to be motivated by what we might call reasonable doubts, the sceptic is driven to the fanciful: e.g., ‘How do you know, Sebastian, that, say, aliens have not replaced the juice with smuice that tastes indistinguishable from OJ, but is actually…’. In my view, one should treat such questions as if they were attempts to express reasonable doubts and reply: ‘Nothing speaks for that, and everything against it’, or we can say, with McDowell at his most obdurate, ‘I know why you think that question is peculiarly pressing, but it is not’.Footnote ¹⁷ Such perverse doubts are just ways of insisting on the HCF conception of grounds, a position disjunctivism has already undermined. One can leave things at that, but if one wants to say more, then it is inviting to challenge the very idea that our thinking might be entirely out of step with reality. If aliens have been at the juice, then, our epistemic situation is askew in this case, but things can't always, or even often, be like that, so we are entitled to restrict consideration of the likelihood of error to the realm of genuine possibility, and the doubter, as Peirce would insist, has an obligation to make the doubting real.

Of course, McDowell and Kern reject the particular argument offered by Davidson for a number of reasons, not least of all for his treating perception, not as the rational apprehension of how things are in the environment, but as a non-rational causal process that yields beliefs standing in need of justification from other beliefs.Footnote ¹⁸ This leaves Davidson embracing a form of coherentism no better than scepticism (‘frictionless spinning in the void’ in McDowell's memorable phraseFootnote ¹⁹). But Davidson's is not the only way to design an argument to the effect that our beliefs cannot be out of touch with reality in the way the evil genius scenario deems possible. My view is that such arguments have a role – even if they cannot show, of any particular belief, that its grounds are truth-guaranteeing – for they target what Kern herself later describes, in her discussion of Michael Williams's view in Unnatural Doubts, as ‘the deepest root of skepticism’, the idea that sensory experience could be just as it is and all our beliefs about the world be false.Footnote ²⁰

3.3. Kern defines knowledge as the outcome of a perfect exercise of the rational capacity for knowledge: ‘Someone knows something just in case her belief is a perfect exercise of a particular rational capacity for knowledge’.Footnote ²¹ I am uneasy about the invocation of ‘perfection’ here. It looks to me either trivially true or false. The claim is trivially true if the exercise of the capacity is perfect just in the respect that it yields knowledge. Otherwise it seems false, because I can surely come to know things by less-than-perfect exercises of my capacity for knowledge, either because the circumstances are not great (I hear the call for help notwithstanding the howling wind), or because I am a little inattentive or otherwise epistemically lazy, dozy, or on Dreyfusian ‘auto-pilot’. Just as the footballer can score with a less than perfectly struck penalty, so I can come to know there is a squirrel in the garden despite the snowstorm and the fact that I am not looking all that carefully. It seems to me that an adequate theory of doxastic responsibility needs to incorporate discussion of such less-than-ideal ways of coming to know, rather than focusing, as Kern does, entirely on cases where hindrance results in ways of failing to know.Footnote ²²

3.4. As I noted above, one great attraction of the capacity approach, for me at least, is the way it brings developmental and educational issues to the centre of inquiries into knowledge. Kern maintains that the rational capacity for knowledge cannot be innate. It must be acquired. But now she confronts a problem. All other rational capacities are, it seems, acquired by ‘practice’, but practice presupposes that the subject knows what she is doing or trying to do. Thus, learning by practice presupposes the prior possession of a rational capacity for knowledge. So now it looks as if the capacity can be neither innate nor acquired.

Kern's way out of this dilemma is to argue it is wrong to assume, as she maintains I do in The Formation of Reason and other writings, that to invoke ‘practice’ as the source of the rational capacity for knowledge is to explain it as emerging out of acts ‘logically more fundamental’ than the capacity they allegedly explain.Footnote ²³ In contrast, she argues that the child develops the rational capacity for knowledge by exercising that very capacity, only at first in ways that are ‘deficient’, partial and inadequate. What is crucial is the role played by another person: the child's initial acts of knowing are ‘mediated through and dependent on the exemplary acts of another subject’.Footnote ²⁴ Of course, everything depends on how this mediation and dependence are understood, and Kern does not elaborate much. It seems crucial that the other subject should not be pictured as merely scaffolding, or otherwise facilitating, the expression of the child's capacity for knowledge. Rather, the other's activity plays a constitutive role, so that the activity of the child initially counts as manifesting her rational capacity to know only in unity with the other. This Vygotskian move enables us to say that initially the child's doings are not in themselves knowings. The character of her epistemic dependence is such that she relies on others, not just for things to know, but for her very capacity for knowledge itself. Her path to full-fledged possession of the capacity is one of the gradual acquisition of the requisite degree of epistemic autonomy. Thus, it can be true that the child, considered in isolation, lacks the rational capacity for knowledge, but, thanks to her striving to become a knower, she possesses it at first in unity with another, and only later self-sufficiently, as a fully epistemically responsible agent.Footnote ²⁵

I see Kern's view, cast in this way, as a friendly amendment to my position in The Formation of Reason, one that provides a framework in which we may speak meaningfully about the acquisition of rational powers, while correcting for the tendency, which I inherit from McDowell and Ilyenkov (and some readings of Wittgenstein), of speaking as if child's development necessarily involves a transformation from a non-rational (‘merely animal’) mode of life to a fully rational one. Better to say that children are within the space of reasons from the get go, not because their rational powers are simply ‘built in’, but because they are orientated to others in ways that enable joint activity that is genuinely expressive of rationality, even if their contribution in itself is not (or is so expressive only in consort with the other). Of course, it is one thing to articulate this insight, another to do developmental psychology in its light. That is an exciting prospect, but one that involves swimming against some strong currents in contemporary thinking.Footnote ²⁶

Moreover, there are many open questions. The idea that there can be deficient, yet still knowledge-yielding, exercises of the capacity for knowledge returns us to the question of whether knowledge should be defined in terms of the perfect exercise of that capacity.Footnote ²⁷ And the rich notion of epistemic dependence introduced in the idea of knowing through the mediation of another might be extended to characterise some dimensions of the condition of mature adult knowers, who, after all, know much only in consort with other subjects, with socially established bodies of knowledge, traditions of thinking, reasoning, and other forms of epistemic practice. It is not as if epistemic self-sufficiency is something we ever attain.

3.5. Finally, let me turn to Kern's treatment of scepticism. I claimed that any satisfying account of knowledge must make sense of the peculiar allure of scepticism. Does Kern succeed in doing so? At the end of her book, she casts scepticism as ‘an unavoidable form of philosophical reflection’ without which ‘we would not develop a philosophical understanding of the kind of activity we engage in when we form beliefs about how things are in the world’.Footnote ²⁸ She follows Stanley Cavell in arguing that a properly philosophical understanding must grasp the truth in scepticism. For her, this resides in the sceptic's insight that knowledge requires truth-guaranteeing grounds, a truth that the sceptic spoils by immediately embracing the falsehood that there are no such grounds. I'm not sure this is enough. On Kern's account, the truth the sceptic grasps is not a distinctive outcome of scepticism; rather, it is implicit in ordinary epistemic practice. What is distinctively sceptical is the subsequent mistake. But that mistake is one that might not have been made. Nothing internal to reason forces us to make it. We make it only if we embrace a host of other assumptions about the relation of mind and world.

Kern's view of the truth in scepticism is not supplemented by the kind of existential themes, which surface in Cavell and Diamond, where, whatever else it may be, sceptical doubt reflects disquiet about our metaphysical loneliness and the elusiveness of reality and other minds.Footnote ²⁹ I'm not sure whether such ideas properly account for scepticism's allure, only that Kern's account falls short of a compelling explanation of why scepticism seems to come so naturally to us – why it grips and excites us – when we are in fact so secure in our epistemic practices.