ISSN 1393-614X Minerva - An Internet Journal of Philosophy Vol. 6 2002.


E.P. Brandon  


Michael Williams has frequently considered and rejected approaches to "our knowledge of the external world" that see it as the best explanation for certain features of experience.

This paper examines the salience of his position to approaches such as Mackie’s that do not deny the presentational directness of ordinary experience but do permit a gap between how things appear and how they are that allows for sceptical doubts.

Williams’ main argument is that, to do justice to its place in a foundationalist strategy, the external world as hypothesis must offer an explicandum that does not invoke concepts of objects but is rather purely experiential. He next claims that no coherent regularities are available at such a level so there is nothing to be explained. Coherence only comes with objects, not as something objects could explain.

Confronting this with Mackie’s Lockean theory of perception, we find that Mackie decisively rejects the first claim about the nature of the explicandum, since he sees ordinary perception as involving intentional objects which are distinct from the persisting objects they present. He is also committed to rejecting Williams’ line on purely experiential regularities, though this plays a subordinate role in his general position.

The crucial issue then becomes the tenability of Mackie’s intentional object analysis and the extent to which it might yet tilt the argument in favour of realism against a global sceptic.

In formulating his own epistemological strategies Williams might appear to countenance a version of Mackie’s view divorced from foundationalism. But while Williams’ contextualism in its minimal version might do so, in practice it retains the lessons derived from his skirmishes with scepticism and thus disallows certain types of enquiry.

I conclude by contrasting Mackie’s response to scepticism with that of Williams in his diagnostic vein.

"But though subject to what strikes me as decisive criticism, the thought that belief in the existence of the external world can be defended by some kind of quasi-scientific inference dies hard." So Michael Williams in a footnote in his Unnatural Doubts (1996, 367).

I intend here to set out Williams' "decisive criticism"—mainly, as we shall see, variations on the theme that an external world hypothesis has nothing independently specifiable to explain—and examine its salience to approaches such as Mackie's that do not wish to deny the presentational directness of ordinary experience but do permit a gap between how things appear and how they are that allows for sceptical doubts.

I — Williams' Criticisms

The footnote I quoted is attached to Williams' discussion of Ayer's classification of the pattern of anti-sceptical strategies, in particular to Ayer's third, "scientific approach". Williams characterises this as a form of inference to the best explanation: the external world or other minds are justifiable as the best explanations of "the kind of experience we have or the way we see others behave" (1996, 54).

Williams' main critical question is: what are these hypotheses supposed to explain? The standard Humean answer is the coherence of experience:

Our experience is not chaotic but follows definite and repeated patterns, a fact best explained by our interacting with a world of stable, physical objects. Note, however, that this answer is given in the context of a response to scepticism that accepts the general epistemological priority of experiential knowledge over knowledge of the external world. It must therefore be possible to know that experience is coherent without leaning on knowledge of objective reality. That is to say, the coherence of experience must consist in its conforming to purely experiential regularities (1996, 54).

So it is besides the point to follow Hume in telling us about the coherence and regular dependence of the things in his chamber after a brief absence—he should tell us about the coherence of his impressions, but he tells us about books and a fire, what Williams allows may be described as "dependent experiential regularities: regularities in the appearance of external things under relevantly similar conditions of perception" (1996, 54). But what we need are "independent, hence purely, experiential regularities" (1996, 54).

But these are not forthcoming, since we cannot find them "unless we are already in a position to allow for the objective conditions of perception" (1996, 54). "[E]xperience is coherent only within certain objectively fixed limits, having to do with the environment and the perceiver: purely experiential regularities collapse with every blink of the eye and turn of the head" (1996, 55). Without the appeal to physical objects that is to be justified, there is no reason to suppose that experience is coherent in any relevant way, so "we subtly beg the question against the sceptic, helping ourselves to the very knowledge he has called in question" (1996, 55).

Before looking at an earlier discussion that Williams invokes to defend the claim about the lack of coherence in pure experience, or the absence of "purely phenomenal laws" (1986, 304), let us note a point he makes in a review of John Watkins' appeal to the same strategy in a Popperian context:

there seems to be no hope of putting the hypothesis of an external world under the negative control of experience, for any experience, no matter how phantasmagoric, can be explained by reference to external events and conditions—influence of drugs, unusual circumstances, and so on. So, by Watkins' own standards, the hypothesis of an external world is not a scientific theory (1986, 304).

This criticism seems, however, to overlook the comparative worth of competing explanations that is a crucial element in inference to the best explanation. We are meant to prefer explanation A to explanation B because it is a better explanation; we need not presume that B cannot be fudged up into an explanation at all. The negative control of observation in rejecting a putative scientific theory coexists with the possibility of reconciling those observations with the rejected theory by invoking sufficiently baroque additional elements.

To return to Williams' main line of criticism, let us see what he says in Groundless Belief, chapter 4, to show that there are no purely phenomenal regularities to be found. Both here and in the review of Watkins, Williams insists that it is obvious that claims about objects do explain how things look to perceivers—we need more from a reply to scepticism and this more is the epistemological priority of perceptual knowledge (1999, 133). He argues that coherent experience must be such as to allow one to formulate inductively confirmable generalisations.

Since ... 'coherent' experience is a precondition of its being possible to apply concepts from the conceptual framework of physical objects—or at least of there being any point to applying such concepts 1 —the order to be found within experience must be describable without any help from concepts belonging to that framework (1999, 134).

He then insists that if we do try to do this, the meagre resources we are left with are quite inadequate for the job: the sequence and collocations of coloured patches vary with every change of circumstance, including turning our head or shutting our eyes, as Hume noted. Williams sums up by saying that "predicates of a purely experiential language would not be projectible" (141).

In this discussion Williams considers two possible replies. The first is that his insistence on pure experiential regularities requires too strong a contrast between observation and theory. It does; but he claims the Humean project requires it if it is to have a philosophical point. The second is that Hume and his followers only claim a limited constancy and coherence among our experiences. Hypothesized objects give us greater coherence. To this Williams replies by reiterating his point that pure experience is not at all coherent, and he notes a probable inconsistency in Hume's claims about the nature of his own experience.

The crucial criticism then includes two elements: the external world as hypothesis demands an explicandum in terms of purely experiential content, to give philosophical interest to the project, but there is no such explicandum to be found.

II — Mackie's Version of the Hypothesis

The Mackie version of the external world as hypothesis that I want to consider occurs in the context of his defence of a Lockean account of perception (originally in Problems from Locke [1976], with a later discussion in Logic and Knowledge [1985]). 2 It decisively rejects Williams' first point and presumes his second also to be mistaken. In his later discussion Mackie is at pains to show that his account is sufficiently Lockean to be understood as a representative theory and to permit the formulation of sceptical doubts, doubts he claims can be set to rest primarily by the hypothetical strategy. 3

The bare bones of his account of perception involve a three-way distinction between (a) how things are, (b) how we normally perceive them to be, and (c) how we can attempt to perceive them by stripping away normal and automatic interpretations. Mackie allows that, whatever may have been the case when we were first learning to interact with the world, the way the world is perceived by normal adults incorporates "a recognition of persisting solid objects in a three-dimensional space" (1985, 215). These "how it looks" are to be interpreted as intentional objects, "existing as the contents of states of awareness, and therefore literally mind-dependent" (1985, 214). But it is crucial to his arguments that this is not how they present themselves—"What we perceive, as an intentional object, is mind-dependent; but mind-dependence is not part, let alone a non-detachable part, of what we perceive" (1976, 61) so we should say that "even when we allow for these intentional objects perception is ordinarily judgementally direct though causally mediated" (1985, 216). Mackie stresses that "this account leaves room for the suggestion that our ideas, in the sense of these intentional objects, may either resemble or fail to resemble the outside things" (1985, 216) and thus leaves scope for variations on a primary/secondary distinction. More to our present concern, it leaves room for scepticism about that very external world which is built into ordinary perception.

Within the usual formulation of the sceptical problem, Mackie distinguishes two issues: a question of meaning and a question of justification. He appeals to a constructive theory of meaning to allow that we can give meaning to sceptical hypotheses, and must then face the question of whether anything can be said against them.

The rationale for this confrontation with Williams is that, at this point, Mackie invokes the parallel with scientific inference: we do not need to establish facts about a cause independently of its effects, because we can introduce causes and the causal laws that link them to effects as a package. What would be arbitrary done singly can be reasonable when "a systematic explanation of a whole body of effects" (1976, 63) is to be offered. So we can "argue that the real existence of material things outside us is a well-confirmed outline hypothesis, that it explains the experiences we have better than any alternative hypothesis would, in particular better than the minimal hypothesis that there are just these experiences and nothing else" (1976, 64). In the context of this outline realistic hypothesis of an external world, Mackie says "it fills the gaps in things as they appear, so producing continuously existing things and gradual changes where the appearances are discontinuous. Its resulting merit is a special sort of simplicity, the resolving of what would, on the rival, phenomenalist, view, be quite unexplained coincidences" (1976, 64).

III — Discussion

In this section I want to focus on some points of contrast between Mackie and Williams. In the following I shall look at how Williams might be seen as coming closer to Mackie when he considers hypothetical world strategies without the foundationalist context he is so insistent on in his critical remarks.

To return to the bare bones of Mackie's account, one can say that whereas Mackie thinks scepticism arises and can be answered while dealing with (b), we have seen Williams insisting that the explicandum must be formulated in terms of (c). This is, as I have indicated, a consequence of Williams setting the issue within a foundationalist approach to knowledge as he understands it.

The sceptical problem Mackie wanted to address arises with (b), but it might be instructive to look more closely at what little he has to say about the more primitive level. That little is in the context of the problem of meaning: when how it looks is characterised in (b) terms, there is no problem of meaning, but what happens when it is characterised in (c) terms? Mackie says we have to consider this "because if we deny, as Locke denied, that we have innate ideas of space and time and substance and causation, we must suppose that at an early stage in our lives we started with experiences whose contents were only of the uninterpreted sort and somehow developed the interpreted sort out of these" (1985, 219-20). He espouses here a "plausible" doctrine of meaning to the effect that "the meanings of any claims we make must be built up somehow out of elements which figure within the contents of our experience" (1985, 219), but concedes that we need not deny innate propensities to arrive at particular concepts. To safeguard the intelligibility of getting to (b), Mackie says that the crucial point is that "[e]ven an uninterpreted how-it-looks is a cluster of features seen simply as being there" (1985, 220) without being given as appearances or as mind-dependent. 4

One may yet wonder what is the force of the initial denial of innateness. The mechanisms whereby I come to have the experiential content I have are innate (assuming Mackie's Lockean primary/secondary distinction); so what isn't innate is simply the having of them—that requires interaction. A particular visual percept (my first view of Lund, for instance) is not innate in the sense that it has been sitting somewhere waiting to be had/perceived. But it is the result of interaction between contingent physical input and innate physiological mechanisms which might have yielded something quite different, so its being the way it is depends on innate mechanisms. When, in particular, it does not resemble those physical inputs, the inclination to label it as basically innate is surely considerable.

An analogy may help to clarify the point here. Different software programs may output very different content when processing the same sequence of binary code. To the extent that what happens on the computer screen is due to the program, it is innate; to the extent that it depends on the particular sequence of code being processed it is non-innate. Mackie's picture seems to want us to construct a large part of the software as well as to process input with it, but these are very different enterprises and it is not obvious that a realistically inclined empiricism needs to have them run in parallel.

Returning to Mackie's main arena, the (b) level, what is the force of Williams' claim that in characterising the sceptical question in this way we have begged the question? One point here is that, if Mackie is right, we can construe claims about experience as not committing us to an external world but as reports on the intentional content of the experience. I shall return to this issue later. But even if it is allowed, it does seem prima facie to beg the question by invoking external world concepts to characterise the intentional content. In reply one might take other examples of the observational/theoretical contrast where we find it permissible to defend hypotheses of a certain sort by reference to observations that have been interpreted in the light of those hypotheses. It is often said that we could in such cases move towards less contentiously interpreted data if necessary, but once the hypotheses have been endorsed we do not bother to do so. But let us suppose that such de-interpreting is impossible. It is still not obvious that the use of a hypothesis to "clean up" data or characterise them in some way will inevitably beg the question whether those data better support it or some rival. At a commonsense level we may assume a persisting sun but still consider whether there is anything to be said for the view that it is new each day. Or at a higher level of sophistication one might refer to the use of Newtonian assumptions to filter astronomical observations: such observations eventually include features like the perihelion of Mercury that help to reject Newtonian theory. 5

IV — Williams' Concessions?

As we saw, in offering his "decisive criticism" Williams insists on seeing the world-as-hypothesis strategy within a foundationalist Weltanschauung. In a more recent discussion (2001), there is an apparent wavering. Here Williams considers someone suggesting that we cannot get purely experiential regularities because "we need to think in terms of an objective world in order to appreciate the ways in which experience is stable and regular" (2001, 115). He again invokes the foundationalist context but then allows that "[o]nce we treat our belief in an external world and the recognition of certain regularities in experience as part of a single explanatory package—each component gaining credibility from the other—we have abandoned foundationalism in favour of some form of coherence theory" (2001, 115). Coherence theory, however, is convicted of hidden foundationalism so nothing has really changed.

Williams' own preferred epistemology is a form of contextualism, but notoriously this ism is not easy to specify with any exactness. At one point Williams gives a Sellarsian characterisation of contextualism that seems to me to capture the logically underwritten common ground shared among Popperians and other theorists of the nature of empirical and scientific knowledge:

Empirical knowledge, like its sophisticated extension, science, is rational, not because it has a foundation, but because it is a self-correcting enterprise which can put any claim in jeopardy though not all at once (Sellars quoted in Williams,
2001, 179)

Williams points out that this avoids the foundationalist assumption that some claims are specially designed for deciding such questions and the coherentist assumption that everything fits together in one total system—though one wonders how long Sellars would expect a self-correcting enterprise to tolerate gross inconsistencies between different successful sub-units.

Williams immediately goes on to make some remarks about how things look that also take him some way closer to Mackie's position. His first set of points concerns the meaning of language about how things look: simply that "'[r]ed' in 'looks red' means exactly what it means in 'is red'" (2001, 179). To talk of 'looks' is to talk guardedly, not to talk of some different stratum of reality. Williams points out, however, that this, if true, is still only about talk about looks, what of the looks themselves? His Sellarsian answer is that these are to be understood on analogy with theoretical entities; they have an explanatory role; "[s]ensations are causal intermediaries between events in the world and our seeing and reporting that things are thus and so" (2001, 184). In a sentence highly reminiscent of one of Mackie's I quoted earlier, "perception is ordinarily judgementally direct though causally mediated" (1985, 216), Williams says "[o]bservational knowledge is causally mediated but epistemically direct" (2001, 184).

But Williams is not inclined to Mackie's metaphysical assertion. Mackie's gloss on how it looks was, as we saw, that it is a matter of a mind-dependent intentional object, however much we may normally concentrate on the objects we perceive. So when I take off my glasses and tell the ophthalmologist that I see a blurred T I am here to be interpreted as doing two things: referring to a real-world T that I am interacting with and about which I do not have any weird beliefs that it has suddenly exhibited macroscopic quantum blurring, and saying, of the experiential content I am aware of, that it is blurred. Most of the time the scope of the explicit or implicit "looks" will be qualitative, not existential, in the terms Williams uses (2001, 180): a real object is said to look a certain way rather than it being said that it looks as if something is so. But this is consistent with the view that something in the characterisation is intentional. 7

But Williams does not go beyond securing the benefits of direct realism, so his account of looks, even though he attempts to incorporate our ability to report them and our relatively privileged access to them, does not amount to conceding the appropriateness of Mackie's handling of scepticism, though it is an admission that there can be more to talk of looks than mere guardedness. Williams wants us to remember the lessons learned from his earlier diagnoses of scepticism. There are assumptions we are not forced to make, there are general questions we do not need to confront in going about ordinary enquiry as historians or physicists or doctors. We are better off not raising them.

By way of an ad hominem argument against Williams here, I want to point out how prone he is to sheer assertion of somewhat contentious philosophical views. One case in point concerns the status of norms. Ordinary practices, cognitive and other, tend not to raise too many general normative questions. One may have to apply norms to find the solution to one's current problem, but little or nothing prompts one to ask general questions about norms themselves. Except, of course, philosophical enquiry. But that is distinctly dubious, the home of unnatural assumptions and equally unnatural answers.

So when Williams confronts the general question of the status of norms, what does he do? When faced with conflicting norms, to ask which norm is true,

is to proceed as though there were some fact of the matter—some fact about what the correct standards of epistemic justification are, or ought to be—that holds quite independently of what we take them to be. This is not how things are. Norms, including epistemic norms, are standards that we set (2001, 170).

As far as I can see, Williams nowhere offers any argument to support this line. It is no doubt the conventional wisdom of all card-carrying naturalists, and I certainly do not wish to dispute it, but I do want to observe that it is simply presented to us on a plate, take it or leave it (though Williams acknowledges that similar disputes are "about the shape of our culture and so, in the end, of our lives" [2001, 12]). What is conspicuously lacking is the kind of argument, from what we need for adequate explanation, that Mackie offers in a related context (1977, ch. 1)—precisely the kind of argument that Mackie has offered with respect to the external world. To be fair to Williams, he does mention Harman's version of an explanationist case against moral facts (2001, 126 and 246) though with little obvious endorsement.

To indulge a little in diagnosis à la Williams, I am inclined to think that a key difference between Williams and Mackie lies in their motivations for deciding how to use the very simple Sellarsian picture of rational enquiry. At the level of ordinary enquiry, there is presumably little problem of principle about deciding what to do next, which claims to put in jeopardy this month. One might think that is quite sufficient—let us get on with what we can do.

But one might also want to use the tool in more amorphous and synoptic enquiries. Here there may seem to be a question of a kind of Aristotelian priority. Mackie insists that we start off with views about an objective world and only later get ourselves into a state in which we can raise a sceptical doubt. If one's use of the tool mirrors this sequence, scepticism (or phenomenalism) appears fairly late in the day as a logically possible contender to explain experience and collapses pretty immediately, despite its ontological economy. Equally easy victories can be won over other sceptical hypotheses. It is a bit like refurbishing phlogiston as a theory of combustion. 8 But if one insists—presumably as a foundationalist will—on starting with the sceptical problem, it can seem a lot more daunting. At that point we need some of the anti-sceptical diagnosis that Williams has provided, to deny the possibility of wiping the slate clean epistemically. (This might suggest a compromise: Williams and Mackie both reject the possibility of starting from scratch in any way, including the Cartesian; their disagreement is on the pertinence or intelligibility of a later question about explaining the contents of experience.)

There also appears to be a temperamental difference. Williams notes that an appeal to explanation can support some sort of primary/secondary quality distinction or a rejection of moral facts, but he wants to stress rather that contextualism is happy to treat facts and values in the same ways, so why make an issue of it? Why raise a philosophical question, instead of going about our ordinary business? I think Mackie's answer is because we want a kind of synoptic understanding, as well as particular answers to ordinary questions. No one need deny that we reason about scientific facts and moral values in parallel ways, just as we reason about chess or legal obligations. But Mackie wants to put some of these large bodies of claims in jeopardy, as a whole body of thought, when Williams seems content to just live and let live. He seems to lack Mackie's sensitivity to and distaste for what Mackie called the "queerness" of various putative entities.

A final point against Williams is likewise inconclusive. We have seen that he allows that external objects explain in an unexciting, non-philosophical way the looks of things. But he denies that we can invoke that fact as any kind of epistemic justification for postulating them. But one might proffer as a basic principle that if X explains Y better than currently feasible alternatives (and Y is assumed to be true) then that explanatory fact always does provide some justification for believing in or accepting X and its concomitants. Atoms contributed significantly to explaining various data about substances; and that explanatory success was surely part of the justification we had for postulating them as real factors. Of course, Williams might say that if we have already got X then it doesn't need any further support of this sort, but that is to deny the salience of the intentional object side of the story. It seems to me that Williams' "decisive" reason for rejecting the hypothetical world must depend upon the rejection of that intentional object account and that is not something for which he seems to have argued.


1. This is Williams' response to an objection similar to that which I urged in the previous paragraph.

2. A version published slightly earlier (1969) presents it as the major reason to reject phenomenalism. In this context it was criticised by Ward (1973) on the grounds, as I see them, that coincidences only exist relative to further assumptions (in this case, that the only possible external world is one very much like the one we have). Ward illustrates his case by suggesting that being dealt a hand of 13 cards that are all spades would be coincidental on the assumption that we are playing bridge, in which it would be the most powerful hand, but it would not be coincidental in other contexts, where its contents gave one no particular advantage or disadvantage. But this seems to me wrong: shuffling is a process intended to lead to lack of obvious order; having 13 cards of the same suit is very obvious order; so it appears odd and unexpected in any such context. To use Mackie's own example of Ptolemaic astronomy, even if the world really did work that way, there would be an unexplained 365-day epicycle for each planet and the sun, which would be more odd than the other epicycles, even if we had simply to accept the whole lot as inexplicable. As Ward says, Mackie's position does seem to assume that the persistence of one thing is less in need of explanation than the recurrent creation and disappearance of many similar things. Mackie's fondness for seeing continuities underlying change perhaps testifies to this quasi-rationalist view, but I suspect it is one we almost all share.

3. Mackie also invokes the important idea that some of what we extrapolate from experience is already supported within experience itself (so some phenomenal physical laws are carried over into the unobservable realm) but this can only happen once we have justified the outline realistic position itself, so the crucial issue depends, on his view, on the hypothetical strategy Williams condemns (v. 1976, 64-5).

4. In his earlier anti-phenomenalism piece, Mackie floated the thought that we may never have gone through (c), "having inherited, as a sort of original sin, the instinctive tendency to interpret what we perceive" (1969, 121). I find it hard to see what is being asserted in claiming that we do not have an innate idea of causation but we do have a propensity to arrive at the, or a, concept of causation.

5. Of course, if the filter had only allowed observations through that completely coincided with what the theory expected, perhaps it would not have been rejected. But is the intentional object case any different? No one supposes we always get perceptions that reliably reflect the objective world.

6. If I may be allowed to quote one of my earlier formulations of what was presented as the tentative results of philosophy for our practice of teaching:

Our position is that any claim to knowledge can be examined, but that it is not possible to test all of them at once. To examine a claim you must take other claims for granted, for the time being (1987, 46).

In filling this out I would first want to insist on the central motivation of achieving explanatory understanding as a way of setting a direction for such a tool. Recalling Harman's distinction between rules of argument as principles of implication and procedures for belief-revision (1986), the key point to be made is that the Sellarsian tool doesn't do anything on its own, it needs to be employed by people who have aims to achieve (remembering also that self-correction can always be done in more than one way).

7. In addition to talking both of ordinary external objects and of how they look, we can also, as Harman maintains, distinguish talk about the experiencing from talk about the intentional objects experienced: "When we talk about after-images or referred pains, some of what we say is about our experience and some of what we say is about the intentional object of that experience" (1999 [1990], 252). Harman's whole argument in this paper depends on denying that we experience "any features as intrinsic features of [our] experience" (251), and thus that we have any awareness of whatever it is that underlies the representations contained in it, though we certainly report on the intentional or relational features of that experience. More generally, the line of thought from Anscombe that Mackie and Harman both share is that our language offers us two constructions with many verbs, one that requires glossing as invoking an intentional object, the other that demands the real existence of a real object. These are not distinguished linguistically in themselves, though it was a burden of Mackie's anti-phenomenalism paper that the former reveals itself pretty quickly by the logical oddities of the intentional objects it invokes.

8. One can see this "unreal" quality of the issue in Vogel's working through of Cartesian scepticism (1990). But it is worth remembering that, in Mackie's terms, it is only an outline real-world hypothesis that has been endorsed so far; there remain conflicting versions within that (Lockean primary/secondary qualities; Russellian structuralism;...) to be fought over.



Brandon, E.P. 1987. Do Teachers Care about Truth? Epistemological Issues for Education (London: Allen and Unwin). Available at

Harman, G. 1986. Change in View (Cambridge, MS: MIT Press).

Harman, G. 1999. Reasoning, Meaning, and Mind (Oxford: Clarendon Press).

Mackie, J.L. 1969. 'What's Really Wrong with Phenomenalism?' Proceedings of the British Academy 55, 113-127.

Mackie, J.L. 1976. Problems from Locke (Oxford: Clarendon Press).

Mackie, J.L. 1977. Ethics: Inventing Right and Wrong (London: Penguin).

Mackie, J.L. 1985. 'Locke and Representative Perception' in J. & P. Mackie (eds.) Logic and Knowledge (Oxford: Clarendon Press), 214-224.

Vogel, J. 1990. 'Cartesian Skepticisim and Inference to the Best Explanation,' Journal of Philosophy 87, 658-666.

Ward, A. 1973. 'What's Not Really Wrong with Phenomenalism,' Australasian Journal of Philosophy, 51, 245-252.

Williams, M. 1986. 'Review of John Watkins, Science and Scepticism,' Philosophy of Science 53, 302-305.

Williams, M. 1996. Unnatural Doubts (Princeton: Princeton University Press).

Williams, M. 1999 [1977]. Groundless Belief (Princeton: Princeton University Press).

Williams, M. 2001. Problems of Knowledge (Oxford: Oxford University Press).

A Version of this Paper was read at the Fourth European Congress for Analytic Philosophy, Lund, June 14-18th, 2002

Copyright © 2002 Minerva.
All rights are reserved, but fair and good faith use with full attribution may be made of this work for educational or scholarly purposes.

Ed Brandon, by training a philosopher, now works in a policy position at the University of the West Indies in Barbados.

Mail to: Ed Brandon

Return to Minerva (Volume 6) Main Page                          Go to Top of This Page