Unnatural Doubts.

Unnatural Doubts. – book reviews

John Skorupski

Unnatural Doubts is a big, comprehensive, difficult–and very impressive–book about one of the deepest philosophical questions. Williams has a feeling for the larger intellectual significance of scepticism and for its phenomenology. He attacks the sceptic’s citadel with boldness, determination and strategic sense. He never underestimates its strength, and wastes no time on ineffectual gimmicks. And his treatment of a large range of other writings on the subject (for example, coherentism in Ch. 8) combines sympathy with acute criticisms.

This is a book with a major central thesis. Recent years, Williams thinks, have seen the growth of what he calls a “New Scepticism”. (He mentions Thomas Nagel, Barry Stroud, Peter Strawson and Bernard Williams as in varying degrees inclined to it.) “It represents a powerful reaction to post-Wittgensteinian (or ‘ordinary language’) and neo-pragmatist tendencies to dismiss traditional philosophical problems, particularly sceptical problems, as not real problems at all” (p. xiv). The new sceptics revive Hume’s verdict, that the only remedy for sceptical doubt is carelessness and inattention. Sceptical doubts are natural doubts, to which there is no answer. They are not, to be sure, natural in the sense that we are prey to them in the business of everyday life. On the contrary, the natural attitude of everyday life strongly resists them. But another attitude, equally natural to us, though not so common, is the reflective attitude, in which we meditate on the credentials of our beliefs in general. When we adopt that attitude, we find ourselves forced into sceptical doubt by simple and unavoidable arguments, which, so long as we maintain that reflective perspective, we are quite unable to resist. This, according to the New Sceptics, reveals “something profound and disturbing about the human condition” (p. xv). (Though what that is they never quite make clear.)

Williams is not a New Sceptic, though he shares many of their instincts. He believes he can give the sceptic a definitive answer. His method is wholly novel. It is not a “constructive” response–the kind of response which acknowledges sceptical questions as fully legitimate and attempts to provide a direct answer in the manner of Descartes. It is not “therapeutic”, as is the “post-Wittgensteinian” and “neo-Pragmatist” response. But it is nevertheless “diagnostic”. The diagnosis offered is “theoretical” rather than therapeutic.

This difference between the therapeutic and the theoretical is subtle. The therapist tries to show that the sceptic’s citadel is not really there. The illusion that it is is generated by our own reflection–yet if we go on reflecting hard enough, the apparent citadel dissolves, we find that we were prisoners of our misunderstandings, and that we can walk freely through. In contrast Williams agrees with Hume that when we take up the reflective attitude we find that scepticism is irrefutable. When we inquire in general about what we know, our answer must be that we know nothing. He goes so far with Hume and the New Sceptics. But he does not agree that we can know nothing when we are not inquiring about what we know, but engaging in some other, non-epistemological inquiry. The sceptic confuses “the discovery that knowledge is impossible under the conditions of philosophical reflection with the discovery, under the conditions of philosophical reflection, that knowledge is generally impossible” (p. xx). “He takes the discovery that, in the study, knowledge of the world is impossible for the discovery, in the study, that knowledge is impossible generally. He infers the impossibility of knowledge from what is, at best, its instability” (p. 359). Williams is arguing, it appears, for a kind of dualism of theoretical reason. The sceptic’s citadel is real, but it controls only the path of epistemological reflection. Over the diverse paths of ordinary inquiry, it has no sway.

Why do we fail to distinguish the epistemological from all other contexts with enough care? It is because we accept a certain theoretical doctrine–the doctrine that the content of a proposition determines its “epistemological status”. Williams calls this “epistemological realism”. It is “the thought that propositions themselves determine, in virtue of their content alone, the conditions that must be met if our believing them is to count as knowing” (p. 349), or again, that “there are invariant epistemological constraints underlying the shifting standards of everyday justification, which it is the function of philosophical reflection to bring to light” (p. 113). The denial of this thesis Williams calls “contextualism”. But he spends far too little time explaining what the thesis is or what its denial amounts too. His descriptions of “epistemological realism” are frequent, but metaphorical and murky.

It is plausible enough that the degree to which a person is justified in judging or asserting something depends in part on the context in which he does so. Williams, however, insists that

To adopt contextualism … is not just to hold that the epistemic status of a given proposition is liable to shift with situational, disciplinary and other contextually variable factors: it is to hold that, independently of all such influences, a proposition has no epistemic status whatsoever. There is no fact of the matter as to what kind of justification it either admits of or requires. (p. 119) But while this tells us that Williams means something stronger, it does not tell us what that something is. If it is plausible that a proposition’s epistemic status is partly a function of contextually variable factors it is, by the same token, plausible that “independent of all such influences” a proposition has no epistemic status. This does not show that propositional content does not contribute determinately to fixing epistemic status–which is what at first sight Williams seems to want to deny. To take an analogy, it may be that sense only determines reference in conjunction with context of utterance–but that does not show that sense makes no determinate contribution to fixing reference at all.

Consider the suggestion that there is a function which determines, for every possible context in which a person may consider a proposition, the degree to which he would be justified in asserting or believing that proposition. A subjectivist may want to say that that function can vary from person to person. Or one may want to say that the function is partial–there are triples of person, context, and proposition, for which the function determines no degree of justification. Or one may insist that there is some further relevant variable, other than proposition, person and context. As for the sceptic, he would say that for some or all classes of proposition the value of any such function is zero for all persons and contexts, or zero for some persons in all contexts, etc.

In contrast, to deny that content contributes determinately to epistemic status would be to deny that such a function–if there is one at all–has content as one of its variables. It would be to hold that degree of justification can vary without depending systematically on–perhaps among other things–content. It would be hard to understand such a view–what grip would it leave one on propositional content? However, despite first appearances, I do not think it is Williams’ view. The passage on p. 119 which I quoted above continues in this way:

Thus stated, contextualism implies a kind of externalism, for though appropriate contextual constraints will have to be met, if a given claim is to express knowledge, they will not always have to be known, or even believed, to be met. As this suggests, Williams’ key point is that context can include states of affairs of which the thinker is unaware. Perhaps one should not call this “externalism”, given Williams’ acute criticism of various aspects of what is often called “externalism” (Ch. 2, [sections] 2; Ch. 8); but it is at least quite clear that he is anti-internalist. And he connects this closely with a rejection of foundationalism, which he characterises as the view that beliefs about the external world form a “natural” epistemological kind, whose common property is epistemic dependence on the evidence of the senses: “foundationalism and internalism are two names for the same thing” (p. 323).

So Williams holds that I may be justified in holding that p in a given context even though I do not know or even believe that I am in that context–and indeed could not know or justifiably come to believe that I am. Now to engage with the sceptic this thesis has indeed to be about justification, not just knowledge. And at this point a familiar disagreement sets in. Those of us who are “internalist” about justification will insiste that justification–the notion we have in mind and that the sceptic has in mind–is normative. Specifically, it concerns the normative notion of rationality–I want to be able to assess whether my beliefs are rationally justified or at least permitted and to get rid of those which are not rationally legitimate (i.e. justified or at least permitted). Other people may be interested in other things, such as whether I am, in fact, a reliable believer. But criticism of my rationality in holding a belief must be something fully open to assessment by me. In this project of self-assessment, if I cannot tell whether I am rationally permitted, then I am not rationally permitted and the sceptic wins, in the sense that I have to agree with him (and Hume) that if I believe at all, I believe without rational legitimacy.

The curious thing is that Williams is well aware of this line of thought. It is what leads him to suggest that in the reflective, epistemological perspective we must concede the sceptic’s claim (e.g. p. 99). In effect–to revert to the justification function–he thinks that when the context is fixed as that of epistemological inquiry, then the value of that function is zero for all or most propositions, and all persons, just as the sceptic says; but that nothing follows from this about the function’s value for other contexts–in which indeed it takes determinate non-zero values. This is close indeed to the straight Humean and New Sceptical conclusion that we do believe without rational legitimacy (in particular that we have no good reason to hold that on some externalist notion of “justification” or “knowledge” our beliefs are justified or constitute knowledge). But to take the final, generalising, Humean step would be, on Williams’ diagnosis, to overlook the “instability of knowledge”–to mistake “the discovery that, in the study, knowledge of the world is impossible for the discovery, in the study, that knowledge is impossible generally”. Unlike Hume, Williams means to confine the authority, and not just the power, of scepticism to the study.

Is this doctrine of instability stable? In arguing against the applicability of scepticism outside the study Williams rejects “the doctrine of the epistemic priority of sensory experiences over independently existing things” (p. 93)–“the thought that any belief whatever about ‘external objects’ must in the end derive its credibility from the evidence of ‘the senses'” (p. 116). But this rejection takes place in the study! It is an integral part of Williams’ theoretical diagnosis of the failure of scepticism and as such is offered from within the epistemological context. The whole book, after all, is an epistemological, not historical, psychological etc., project. Williams is aware of the problem and in his closing pages retreats a little from outright endorsement of the idea that knowledge in the study is impossible:

Does Hume really undermine his ordinary knowledge of the world when he enters the study? Or does he simply find himself unable to solve a problem that we can now see to be an artefact of prior metaphysical commitments? It is hard to say. (p. 358)

It seems to me that the strategic position is less puzzling than this suggests. If Williams were right in rejecting “the thought that any belief whatever about ‘external objects’ must in the end derive its credibility from the evidence of ‘the senses'”, then he would have a route to a complete refutation of scepticism–there would be no refuge for the sceptic in the study. The trouble is that that thought is not easy to reject, because the simple point that justification is normative seems enough to ground it in exactly the way the sceptic wants it grounded. (It is enough, for these purposes, that beliefs about “external objects” are not a priori.) If Williams is tempted to argue that knowledge is unstable, it is because he implicitly recognises the epistemological context as a normative one, in which reflection legitimately lifts contextual restrictions on what is in question, and thus leads to a conclusion about the epistemic priority of experience. And he implicitly assumes that when that is done, the sceptic wins. The trouble is that the epistemological context is not just another context. (Any more than the context of repentance is just another context.) It is the context in which we reflect on the rational legitimacy of our reasonings in all contexts. So if the sceptic wins in this context, he wins.

What, finally, of the other “therapeutic” kind of diagnosis? Against this Williams brings the familiar charge that the arcane theories of language on which such therapy depends will always be outweighed by the evident intelligibility of what the sceptic says (pp. 15, 36-7, 153).

I find this blankly contrapositive argument feeble–but I should declare an interest, since I have defended the plausibility of therapeutic diagnosis in an article to which Williams refers. There, as Williams says, I suggested that the intelligibility of scepticism presupposes “the classical pre-understanding of meaning”, contrasting that with “the epistemic conception” (which, I argued, does not entail an epistemic conception of truth). This latter, epistemic, conception is undeniably in the “neo-pragmatist” and “post-Wittgensteinian” spirit which Williams and the New Sceptics reject. But it is worth stressing that the diagnosis it offers does not deny the intelligibility of the sceptic’s so-called hypotheses about dreaming, brains in vats, etc. What it denies is the intelligibility of the particular reading the sceptic has to give them, a reading in which no rational grounds could in principle be advanced for ranking their plausibility against other, more familiar “hypotheses”. We can understand them well enough in their natural sense, and we can even get a feeling for what the sceptic intends by the special emphasis he places on them–it is what induces the often commented on “vertigo”. But this is not to concede the literal intelligibility of what the sceptic intends, which is far from obvious. To deny it strikes me as a good deal less disconcerting than denying “the thought that any belief whatever about ‘external objects’ must in the end derive its credibility from the evidence of ‘the senses'”, as Williams does.

It goes without saying that much more could be said about the contrast between “therapeutic” and “theoretical” diagnosis, but this is not the place to pursue it further. Williams’ anti-sceptical diagnosis is argued with enormous subtlety. It would be no small matter to respond to it with an equal level of subtlety and care–it would certainly take more than a brief review.


Department of Moral Philosophy University of St. Andrews St. Andrews Fife KY16 9AL Scotland

COPYRIGHT 1994 Oxford University Press

COPYRIGHT 2004 Gale Group