DeRose and Black make similar criticisms of Williams; they both argue that Williams has missed some important sceptical arguments in his analysis.

DeRose contends that Williams' anti-sceptical arguments focus mainly on evidentialist arguments. Evidentialist arguments, as DeRose characterizes them, are a variety of sceptical argumentation that aims to show that knowledge of the external world cannot be derived from experiences. These sorts of arguments require foundationalism, because only if we put epistemological emphasis on experiential beliefs over other beliefs does the argument that experiences cannot bring us knowledge hold weight. So if the anti-sceptic simply determines that we do not need foundationalism, and rejects it, then she has made a strong anti-sceptical case.

But DeRose stresses that Williams has only made a strong anti-sceptical case for evidentialist arguments; the important task, argues DeRose, is to show that non-evidentialist arguments also require foundationalism. We may grant this, and begin a search for other sceptical arguments that are non-evidentialist, so that, if we can find any, we may see if there is a way to expose them as having foundationalist presuppositions. Nonetheless, DeRose's criticism of Williams seems to be about scope, rather than about any flaws in argumentation: Williams' anti-sceptical arguments do work, they just have not been shown to apply to all sceptical arguments, or perhaps the most important ones. We will need to survey some non-evidentialist arguments to see if this criticism holds.

With this in mind, DeRose focuses on Williams' anti-sceptical arguments against broad sceptical hypotheses. These hypotheses, in which PE is a proposition about the world, such as "I have hands," and HS is a sceptical hypothesis, such as "I am a handless brain in a vat," are of the form:

I know that PE entails not-HS;
I do not (cannot) know that not-HS;
Therefore I do not know that PE. (Williams, Unnatural Doubts 80)

Williams' contention is that the second premise, that one cannot know that one is not a brain in a vat, only holds if we accept that experiential beliefs have epistemological priority over other beliefs; that is, if we are foundationalists. If the second premise requires foundationalism, then we reject foundationalism and the argument fails.

DeRose contends that there exists an intuitive interpretation of this sceptical hypothesis, and that it does not in fact require foundationalism to succeed. DeRose asks us to suppose that we have never had nor ever have taken ourselves to have had any sensory experiences. Instead, suppose our beliefs about the world are formed completely by events in our brains, without any sense experiences, so that we only have beliefs about the world, and no beliefs about our sensory experiences. Further, we know that this is the case, and take ourselves to know what we believe about the world through perception. We would have no inclination towards believing foundationalism; we could not, because we would hold that we had never had any sensory experiences, and so would hold that sensory experiences are obviously unimportant to knowing. As DeRose says, "Suppose... that we have never had, nor did we ever take ourselves to have, [emphasis added] any sensory experiences." (DeRose 605) Next, someone in this situation makes a sceptical argument of the brain-in-a-vat sort that we mentioned earlier, and knowledge of the world is successfully doubted.

The difficulty is that DeRose attempts, through brute force, to construct a situation in which we cannot possibly hold foundationalist sympathies, so that he may make a non-foundational sceptical argument. He does this by straightforwardly inserting in his argument the premise that as would-be knowers we do not take ourselves to have any sensory experiences. The next step is to simply make a sceptical argument, and we have a sceptical argument from people who do not, and cannot, hold any foundationalist presuppositions; that is, we supposedly have a non-foundationalist sceptical argument. Unfortunately, the price DeRose pays for the straightforward introduction of this foundationalism-avoiding premise is contradiction; we must hold that we do not ever have any sensory experiences, yet we also hold that we know what we believe through sensory experiences.5 As DeRose says,

Suppose... that we have never had, nor did we ever take ourselves to have, any sensory experiences. Rather perceptual beliefs about the external world were produced directly by neural events, without any accompanying experiences. And suppose that, realizing this, we still took ourselves to know what we came to believe through perception [emphases added]. (DeRose 605)

DeRose's sceptical arguer holds beliefs that contradict one another; he holds that we have no sensory experiences while also holding that we know our beliefs through sensory experiences that he, by his own admission, does not even believe we have. We may take for granted that there are psychological accounts of belief that hold that we can and do hold contradictory beliefs, but we nonetheless have an argument based on people who argue from contradictory beliefs; and if someone with contradictory beliefs about knowledge comes to make a sceptical argument that requires beliefs about knowledge, then we have no reason to accept the argument, because the arguer contradicts himself in making his case. If our would-be sceptic believes that he gets knowledge of his beliefs through sensory experiences, then he can make his brain-in-a-vat sceptical argument, but if he also believes that he does not have any sensory experiences, then he cannot reasonably make his sceptical argument. The brain-in-a-vat argument requires believing we have sensory experiences, so both believing that we do and do not have sensory experiences makes it impossible to make such an argument. The result is that even if he could hold such contradictory beliefs, the arguments produced from them would be nonsense. And in order for DeRose's non-foundationalist sceptical argument to succeed, his sceptical arguer must do just that; he must argue from contradictory beliefs, in which case we have no reason to accept the sceptical argument that follows, and no reason to accept that DeRose's argument has presented us with a non-foundationalist scepticism.

But none of this takes into account that the argument presented does seem to be foundationalist anyways. Let us take for granted that the argument goes through despite the contradiction, and that we have a reasonable sceptical argument. DeRose asks us to suppose that we take ourselves to know what we believe through perception (DeRose 605). If this premise simply means that we are caused to know what we believe through sense experiences, then it supports no sceptical threat. Williams discusses the sceptically innocent nature of the contention that beliefs are caused by sense experiences, in order to avoid confusion arising from not recognizing the distinction between sense experience as cause and sense experience as grounds (Williams, Unnatural Doubts 69-70). As Williams explains:

This much is perhaps a truism: that without functioning sense organs, I would never form any beliefs about the external world and so would never come to know anything about it either. But all this shows is that possessing functioning sense-organs is a causal precondition for possessing knowledge of the world: it establishes nothing whatsoever about the general evidential basis of such knowledge, not even that it has one. Consequently, it offers no inkling as to how any such supposed basis might be inadequate. (Williams, Unnatural Doubts 69)

But this does not seem to be what is meant by the premise in question. The premise says that we take ourselves to know what we believe through the senses; that is, we have beliefs, but we know them in light of our sense experiences. DeRose seems to imply that we take ourselves to know our beliefs to be true just because they are beliefs of a sensory kind. Perhaps this is unfair interpretation of what was meant by this premise, but the only other explanation that presents itself is the causal one, which is sceptically irrelevant. And it is hard to see how knowing our beliefs through the senses can be interpreted in any way but this; we have beliefs for whatever reason, but we know those beliefs through sense experience.

This sounds dangerously foundationalist; if we take ourselves to know beliefs through the senses, then the senses are what we take to be central in forming knowledge. If we take the senses to play a central role in forming knowledge, then it is difficult to see how to avoid the contention that sense experiences take a priority over other beliefs in bringing us knowledge. We take ourselves to know our beliefs through the senses, but this seems to boil down to saying that we take sense experiences to be more important than other beliefs in forming knowledge. Williams makes a similar point in response to Stroud's claim that, "What we gain through the senses is on Descartes's view only information that is compatible with our dreaming things about the world and not knowing anything about that world. How then can we know anything about the world by means of the senses?" (Stroud, The Significance of Philosophical Scepticism 12-13) Williams replies:

This takes a lot for granted.... the assumption that we know about the world "by means of the senses" is simply shorthand for a foundational view of knowledge and justification. This is an example of what I suggested is a recurrent pattern in supposedly intuitive arguments for scepticism: a seeming truism (all empirical knowledge is in some sense dependent on the senses) serves to introduce a contentious epistemological doctrine (all knowledge of the world must be derived from more basic, experiential knowledge). (Williams, Unnatural Doubts 69)

It seems that DeRose may have made the same familiar, but perfectly understandable, mistake in identifying an argument as intuitively sceptical yet lacking in foundationalist presuppositions.

If this is so, then it is obvious that DeRose's sceptical arguer can make his case, but only because he is a foundationalist; he holds foundational beliefs by believing that we know our beliefs through the senses, and so he can make brain-in-a-vat arguments that threaten our knowledge. But this is not the non-foundationalist sceptical argument DeRose was looking for. If anything, the case that sceptical arguments require foundationalist presuppositions has been strengthened. It seems that DeRose's strongest criticism of Williams is that he does not address the sceptical arguments that matter, that is, the sceptical arguments that are non-evidentialist yet still intuitive, but we have yet to see such an argument.

Carolyn Black also contends that Williams has failed to address some crucial sceptical arguments. She mentions that all of the sceptical arguments that Williams addresses are very general in their scope, and are only brought forward outside of everyday life situations. Williams' arguments would be strengthened by discussing sceptics who trade in specific questions and doubts, rather than theories and generalities (Black 742).

Williams is of course concerned with sceptical arguments with general focus and scope because they are the radically sceptical arguments, the ones which impugn all of our knowledge; they are the arguments that conclude that we cannot have any knowledge whatsoever, by showing that if knowledge fails in a best-case scenario, then it must fail in all cases of knowledge (Williams, Unnatural Doubts 135).

Without further example of what sceptical arguments that trade in particular doubts are, it is hard to say exactly how such arguments might look. But it sounds reminiscent of the argument from error. For example, if I say, "I know that is my friend down the street," someone may challenge me with a particular doubt and ask, "How do you know that is your friend? She is very far away (or the lighting is bad or you haven't seen her face, and so on.)" A knowledge claim is made, and it is challenged by particular doubts. The difficulty with presenting such an argument as significantly sceptical is that particular doubts may be answered with particular evidence: we may get closer or get into better light or look at the person's face in order to confirm that she is in fact my friend.

Particular doubts do not amount to radically sceptical arguments because they do not rule out the possibility of overcoming the doubts and confirming your knowledge. Radically sceptical arguments, by contrast, due to their generality, rule out the possibility of ever having knowledge, which is a deep epistemological problem. Non-radically sceptical arguments challenge our knowledge, but do not make nowledge impossible, and so do not seem to be deeply epistemologically worrisome (Williams, Unnatural Doubts 47-51). Black's contention sounds as if it may be very similar to the argument from error, and the argument from error is not sceptically interesting due to its lack of scope.6 As Williams says, "... it would not be reasonable to base a general distrust of the senses on the fact of their 'deceiving' us in certain special circumstances: for example, when the objects we are interested in are small or far away or observed under poor conditions." (Williams, Unnatural Doubts 136) It seems that if a sceptical argument is going to trade in particular doubts, then it naturally is not a radically sceptical argument, and so does not present a strong enough problem to give us a great deal of worry. Williams intentionally avoids discussing arguments that trade in particular doubts, because their very lack of generality makes them a weak example of sceptical argumentation.

Black also contends that Williams fails to take account of Pyrrhonist arguments. Such arguments raise numerous genuine doubts about claims or beliefs about one's own experiences (Black 742).

A classic Pyrrhonist argument is the Five Modes of Agrippa (Empiricus 110-112). We have referred to part of the Five Modes as Agrippa's Trilemma, because the 3 modes of infinity, circularity, and assumption are what give the Five Modes their sceptical teeth. Williams does discuss this, and uses it to establish the neutrality of experience in order to show that foundationalism is not forced upon us (Williams, Unnatural Doubts 65-66). But Black must have something different in mind from this; after all, she mentions genuine doubts being raised about one's own experiences.

The Pyrrhonist approach is to not assent to any claim, because any claim may be legitimately doubted, by Agrippa's Trilemma for example (Empiricus 255-256). Perhaps this is what Black had in mind when she mentioned particular doubts. It is hard to say exactly what was intended by introducing the claim that even our beliefs and senses can be doubted. Agrippa's Trilemma shows that the senses can be doubted, which paved the way for contextualism by showing that foundationalism cannot get its hooks in, so to speak, by giving priority to sense experience. Pyrrhonist arguments can give reason to doubt the external world just as much as the senses; and the neutrality of experience then follows.

But perhaps some important questions are raised by this approach. If anything can be legitimately doubted by Pyrrhonist arguments, without any sort of foundationalist presuppositions, why does the sceptic not just play by her own rules, objecting that ordinary everyday contexts don't matter because of Pyrrhonian sceptical reasons, and put forth her own sceptical contexts all the time?


The question we are presented with is: why don't we just do sceptical epistemology all the time? Why not bring up sceptical contexts at every turn? The Pyrrhonist arguments seem to give us reason to bring up sceptical questions whenever we feel like it. If we are still able to ask sceptical questions all the time, then it would seem that scepticism still looms strong. If we use only beliefs about sense experiences along with epistemologically realist principles, such as foundationalism, as the methodologically necessary beliefs in a context, then we do not know if external objects exist. But if we do this all the time, as suggested, then we have just adopted foundationalism; we may have no particular reason for holding foundational theories, but nothing is stopping us from bringing them up all the time.

Of course, we may just as well use beliefs about external objects as our formally foundational presupposed beliefs of context, and say that we do have knowledge of many things. If a sceptic always ask questions about the existence of external objects, forcing us to prove the existence of them by using sense experiences, we can respond in the other direction, asking the sceptic to show us why beliefs about sense experiences are any more dubitable than any other beliefs. Agrippa's Trilemma shows us that we have no good reason to adopt one context's methodologically presupposed beliefs over another's. To force one direction over the other so that sceptical conclusions are unavoidable seems to be foundationalism by force rather than the tacit foundationalism we dealt with earlier. There would be no grand conclusion about all of our knowledge to draw from questioning the existence of external objects all the time, any more than there would be a grand conclusion in the other direction.

We may still have misgivings about all of this. How does contextualism get rid of the problem of scepticism? It seems we can still easily ask sceptical questions, so why do they not have any effect? In fact, it seems that sceptical worries have a sort of broad applicability that most common conclusions in everyday inquiry do not. For example, we may study some particular facts about Lincoln, taking for granted the fact that he existed as a methodologically necessary presupposition for doing history about Lincoln. We may find some novel facts about Lincoln in the process that will affect other closely related forms of inquiry, perhaps history about American politics. Suppose we find out that Lincoln was not the sixteenth president of the US. Our beliefs will presumably change and, realizing we were mistaken, we will take ourselves to have different knowledge from before. These changes may have an effect on how history about Lincoln and American politics and other historical fields is done; perhaps we will be forced to revise history about the civil war, and so on. Taking ourselves to have this new piece of knowledge about Lincoln, it would seem reasonable to use it as a methodological presupposition in other relevant contexts, such as history about the civil war, perhaps. But it will not affect math or logic or quantum physics in the least; for inquiries into these subjects will presumably share no or few important presuppositions with history. Likewise for discoveries in math and physics: these will almost definitely not affect history in any appreciable way. So it is reasonable to suppose that conclusions in some contexts can affect other contexts only depending on their applicability to those contexts.

But suppose we decided to take an epistemological line of inquiry, and ask a sceptical question: how do we know that objects exist? Whether the answer is positive or negative, this seems to automatically have application in a great deal many of other contexts; we must presuppose that Lincoln exists to do history about him, so we must presuppose that at least some objects exist. Likewise for geography, astronomy and so on: we must presuppose the existence of objects to study the solar system, or the terrain of Earth. It would be reasonable to use the conclusions of the epistemological, sceptical question as the presuppositions for a great deal many other fields, so that changing the subject does not seem to alleviate the question of scepticism -- doing history does seem to depend on the outcome of sceptical questions.

If scepticism leads to a negative assessment of all of our knowledge about external objects, then our presupposition that Lincoln existed seems to become an illegitimate assumption; we cannot presuppose it because there are good reasons to think we are not justified in holding it true. How would a skeptical argument like this go? Presumably we would make a claim, for example, "Lincoln was the sixteenth president." The always alert skeptic gets straight to the point and asks, "How do you know Lincoln existed?" Here we must cite sensory evidence: I have seen photos, letters, and so on, that show that Lincoln existed, perhaps. We have changed the subject from history to epistemology, but there is nothing prohibiting this, and we have no other evidence to cite. But if we are challenged again by the sceptic, we are left without any idea of what to cite as evidence; we can probably only say that we saw some evidence. We move to the epistemological context in order to escape the sceptic's challenges, because we run out of normal evidence to cite. But in making this move we ground our knowledge in an ultimate source: the senses. If we wish to defend the claim that we saw photos of Lincoln, we either refuse to defend it, thus making our claim a groundless assumption, or we say that we just saw it, which is circular reasoning.7 Here we run into Agrippa's Trilemma once more, which is perhaps what Black had in mind when she contended that Williams must address Pyrrhonian arguments. We have a skeptical argument from Pyrrhonian considerations in Agrippa's Trilemma. And if it is shown that we cannot know that Lincoln existed, why should we be able to presuppose it?

In Problems of Knowledge Williams addresses these issues. He distinguishes two approaches to justification: the Prior Grounding Requirement, and the Default and Challenge approach (Williams, Problems of Knowledge 24-25). The Prior Grounding Requirement says that being epistemically responsible in believing a proposition depends directly on one's belief being based on adequate evidence. This view makes evidential justification fundamental for all of justification, since being epistemically responsible depends crucially on having evidence; epistemic responsibility just is a matter of having good evidence. The Default and Challenge approach, on the other hand, holds that one is entitled to his belief by default, but that one's entitlement is "always vulnerable to undermining by evidence that one's epistemic performance is not up to par" (Williams, Problems of Knowledge 25). If one's epistemic responsibility in holding a belief is reasonably challenged, then evidence in favour of one's belief or epistemic reliability must be produced in order to hold on to entitlement. In this way, beliefs need not necessarily be derived from evidence, but must be defensible; beliefs may be presumed, but still require defense in the face of discrediting evidence.

An important aspect of the Default and Challenge approach to justification is that challengers share justificatory obligations (Williams, Problems of Knowledge 150-151). One does not always hold the entitlement to nakedly challenge a claim, unlike what the Prior Grounding Requirement would allow; rather, challengers must earn reason to challenge a claim by either finding specific reasons why the claimant may believe falsely, or by having reason to question the claimant's entitlement to hold her belief. As Williams explains:

Appropriate defeaters cite reasonable and relevant error-possibilities. There are two main types. Non-epistemic defeaters cite evidence that one's assertion is false: this evidence might be purely negative, or it might be positive evidence for the truth of some incompatible claim. Epistemic defeaters give grounds for suspecting that one's belief was acquired in an unreliable or irresponsible way. Here the objector concedes that his interlocutor's claim or belief might be true but denies that it is well-grounded. (Williams, Problems of Knowledge 149)

Challenges, as much as claims to knowledge, have some presupposed default entitlements, and so challenges need to take place in a justificatory context as well, so that naked challenges need not be legitimate defeaters for a knowledge claim; a blank challenge of "How do you know?" may be reasonably met with a counter-challenge, by asking, "What do you have in mind in asking how I know?" If the challenger can give no justification for their challenge, then no response is required, because the challenger has not made a challenge in any sort of context and so has no methodological presuppositions, or any presuppositions, and so inquiry cannot logically proceed. That is, the challenger may put forth a challenge that questions the reasoning used in forming the belief, or by presenting a conflicting belief that is also plausible, but he may not nakedly ask "Why?" For a challenge to be legitimate, it must be justified just as a claim must be justified. This requirement gets its inspiration from Wittgenstein, when he says, "Can we say: a mistake doesn't only have a cause, it also has a ground? I.e., roughly: when someone makes a mistake, this can be fitted into what he knows aright" (Wittgenstein 75). The Default and Challenge approach is of course the driving force behind contextualism.

So the sceptic has challenged our claim to knowledge: I say I know something about Lincoln; the sceptic asks me how I know this, demanding evidence. We can see that the sceptic has taken for granted the Prior Grounding Requirement conception of justification: the sceptic does not challenge our claim with particular evidence showing us to be irresponsible or mistaken in believing, as would be the case under the Default and Challenge approach, but rather sees any claim as challengeable, because all beliefs necessarily require evidence to be held in the first place. Assuming the Prior Grounding Requirement gives the sceptic a generic reason to challenge any knowledge claim: if evidence is required before a belief may be justified, then demanding evidence is always an option if one wishes to question another's grounds for believing. The sceptic has reason to bring up the skeptical question at any turn, threatening all of our knowledge.

So the sceptic takes for granted the Prior Grounding Requirement, meaning her challenge to our knowledge claim needs no particular explanation; the sceptic may generically ask, "How do you know?" For example, "How do you know that Lincoln existed?" We would most likely answer by saying, "I've seen photos, and letters and so on that prove his existence." But given that the sceptic is committed to the Prior Grounding Requirement and may indefinitely challenge us for evidence, she may ask again, "How do you know that you have seen photos, letters, and so on?" Here we probably just do not know what to say, other than to say that we saw the photos and letters. We quickly run out of evidence, and are forced into an epistemological context; we commit ourselves to citing sensory information as the ultimate evidence for our claim. If we accept the sceptic's first challenge to our knowledge, we accept all the following challenges. It seems as if we have been forced into accepting a foundational view of knowledge, and so skepticism, but it is only for the Prior Grounding Requirement that we are in such a situation.

Rejecting the Prior Grounding Requirement means that the sceptic is not entitled to blankly and generally challenge any claim. The Prior Grounding Requirement is internalist (Williams, Problems of Knowledge 148), in that grounding a belief requires first having evidence, and is thus foundationalist or coherentist (Williams, Problems of Knowledge 154), so it seems reasonable to reject it in favour of the Default and Challenge approach and the externalist contextualism that follows.

There does not seem to be reason to suppose that we can only draw conclusions about the rest of knowledge from sceptical epistemology, and not draw conclusions about epistemology from the rest of our contexts of inquiry. We have been given reason to think that no belief takes priority over any other, so we may just as easily argue that we know a lot of facts about Lincoln, so he must have existed, so that our conclusions from history would bear on conclusions from epistemology. To prioritize sceptical conclusions about epistemology is a form of foundationalism.

We are reminded of Moore's argument and his conclusions that two hands exist, so objects exist; why not argue in such a way? To disallow it is to prioritize one sort of belief over another. At worst, the conclusions from history and from sceptical epistemology are on equal grounds; neither is more widely effectual than the other. The problem with Moore's claims was that they clashed with scepticism's conclusions, but we have been given good reason to think that sceptical contexts are no more important than any others.8 Moore's conclusions seemed so blatantly obvious that they were almost not worth mentioning. So what was wrong with his obvious pieces of knowledge? The answer may be that nothing is wrong. There is just as much reason to say that Moore knows there is a hand in front of him as there is to say that Moore knows he sees a hand in front of him. Moore's conclusions perhaps seem strange because they act as methodological presuppositions in many contexts, and so we almost never argue for their truth (except in epistemological, foundationalist contexts, which do not take priority over other contexts.) The existence of objects is often presupposed by arguments in particular contexts, so that pointing out that a tree exists seems strange; but we may say "of course it exists, if it didn't, I could not see it right now." Such a statement is no more or less credible than the claims of the foundationalist sceptic.

But perhaps there are good reasons for choosing the Prior Grounding Requirement and its foundationalism over the Default and Challenge model and its contextualism. Skorupski and McGinn broach such subjects.


McGinn characterizes Williams in the following manner (McGinn 211-214): According to Williams' contextualism it is possible that in some, maybe even many, contexts one does know that one is not a brain in a vat; it is a methodological presupposition of claiming to know a certain fact in some contexts that I am not a brain in a vat, and since Williams does not need to deny the principle of closure, the particular fact that I come to know in that context entails that I also know that I am not a brain in a vat. It is Williams' externalism that allows him to make this claim, embodied in the situational constraint; there are objective states of affairs about the world that put conditions on whether or not we have knowledge, but these objective facts do not necessarily need to be known to obtain in order to have knowledge. According to externalism we do not need to know that we know in order to claim knowledge; if I am not a brain in a vat, then I can know that I am not a brain in a vat, even though if I were a brain in a vat I could not know anything (McGinn 213).

In certain contexts then, it is possible to know I am not a brain in a vat. But Williams also points out that knowledge may be unstable; a change of context may result in our losing knowledge we previously had in a different context. For example, a claim to know that one is not a brain in a vat changes the context to an epistemological one in which one's belief that one is not a brain in a vat comes into question. But the only context in which it is reasonable to question such a thing is one in which foundationalist propositions are methodological necessities. The skeptic's conclusion has been cut off from our everyday knowledge, and may only seriously affect our doubts in inquiry involving a philosophical, epistemological, and foundationalist context. The skeptic now just shows not that knowledge is impossible, but only unstable; the sceptic only affects us in one context, and so his conclusion is confined.

McGinn's contention is that Williams' externalism is too weak to defeat the skeptic in this way (McGinn 214). It can only propose that if my experiences are causally related to an objective world, as we take them to be in everyday situations, then I have knowledge of the world, but if the world is not this way, for example if I am a brain in a vat, then I cannot have knowledge of the world; my methodological presupposition that I am not a brain in a vat will be false by externalist standards, so that any knowledge derived from it will also be false. We do not have unconditional certainty about our judgments about our environment (McGinn 215). As McGinn explains, "To rely on contextualism at this point, and to attempt to ground these certainties in the idea of methodologically necessary assumptions, also seems to make our relation to them qualified and therefore too weak." (McGinn 215)

Skorupski conducts his discussion of Williams in terms of epistemological functions,9 but his position seems to be akin to McGinn's. Skorupski explains that Williams' externalism is what allows contextualism to stave off the problem of having no account of propositional content in justification; to deny that a proposition has any content outside of a context would be to hold that degree of justification can vary without depending on content, leaving us in the hard-to-understand position of having no grip on propositional content. In order to remedy this, Williams' externalism allows context to include states of affairs of which the knower is unaware (Skorupski 402).

Next, contextualism's instability of knowledge staves off objections about rational assessment of beliefs. Most internalists will hold that a crucially important facet of justification is rational assessment of one's own beliefs. This view seems to parallel the Prior Grounding Requirement: in order to be justified we must first rationally assess our beliefs, where rational assessment is presumably a matter of giving good evidence and reasons for our beliefs. But if we are externalists and need not know the objective facts that ground our knowledge, then it may be the case that we need not rationally assess our own beliefs in order to have knowledge. For the internalist this unassessed knowledge is not knowledge at all, because "... 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" (Skorupski 403). Externalist "knowledge" allows that I need not know that I know, but knowing that I know is rational assessment, which is the important part of justification for internalists. As such, Williams explains that this sort of rational justification of knowledge arises in an epistemological context in which we have no knowledge, but which also requires foundationalist presuppositions and which does not affect other, non-epistemological and foundationalist contexts; the authority of scepticism is confined to foundationalist contexts.

Skorupski's contention is that Williams' instability of knowledge argument is insufficient to avoid the sceptic; justification of our beliefs through rational assessment is normative, which is enough to ground internalism in the way the sceptic wants (Skorupski 403). The epistemological context is a normative one, and so reflection on our knowledge in sceptical ways legitimately lifts contextual restrictions on our beliefs, leading to a conclusion about the epistemic priority of experience. The epistemological context is not the same as other contexts; instead it is the context in which we reflect on the rational legitimacy of our reasoning in all other contexts (Skorupski 404). The epistemological context takes priority over other contexts, meaning if the sceptic wins in his epistemological context, he wins in all contexts.

If externalism is correct, then rational assessment of beliefs is not necessary. But Skorupski contends that the epistemological context is normative, so Williams' externalism is overridden by the internalist's requirement for rational assessment of our beliefs. Knowledge is not unstable; instead, it is nonexistent, because a negative result in the epistemological context bears on all of our knowledge, given that the internalist context of rational assessment is more fundamental than the externalist. This criticism seems to parallel McGinn's: externalism is too weak because it allows for the possibility that we are brains in vats but do not know it; if externalism is correct, and we need not know that we know in order to claim knowledge, then if we are brains in vats we may claim knowledge about external objects when there is no objective fact about such external objects. Without some sort of internalist-style rational assessment, externalism allows for the possibility that we "know" things about the world when we are in fact brains in vats; as such, internalist contexts of rational assessment of our beliefs take priority.

The theme seems to be that there is something legitimate to the internalist concept of rational assessment of our beliefs over the externalist concept, perhaps because it is a part of our concept of knowledge, whether that is because it is normative or simply because externalism is too weak to deal with scepticism and an alternative is required.


Skorupski argues that externalism is required to deal with the problem of propositional content. But this seems strange. Williams' reason for adopting externalism is to account for objective facts; and once externalism is adopted, objective facts can play into Williams' theory, meaning it is not cut off from the world. But to say that the problem of propositional content is solved by externalism is to say that there is a problem with contextualism not accounting for the objective nature of propositional content; it is to say that there is an objective fact of the matter about a proposition's content outside of any context, which must then be accounted for, like facts about the world, by externalism. Why would we need an externalist account to deal with the content of propositions unless the content of a proposition were an objective matter? The only reason we have to think that there is an objective fact of the matter about what a proposition's content is, outside of context, is if we are epistemological realists. Further, if epistemological realism is to result in scepticism about the external world as Skorupski would have it, it must be realist about propositions about perceptual beliefs, and thus be foundationalist. But Williams adopts an externalist position to account for objective facts about the world, not objective facts about propositions. It seems that Skorupski holds marked foundationalist leanings in responding to Williams.

It is also claimed that internalist rational assessment of beliefs results in the epistemological context being "not just another context," (Skorupski 404) that is to say, that the epistemological, sceptical context takes priority over other contexts so that its conclusions reflect badly on all of our knowledge. But if internalism is just another way of being a foundationalist (Williams, Unnatural Doubts 323), then by giving the internalist context priority we are giving priority to foundationalist theories. Again, this is a position we ought to adopt if we are foundationalists, but we have been given no good reason to be foundationalist over contextualist. If anything, foundationalism's sceptical leanings give us reason not to adopt it.

McGinn likewise mentions that under Williams' externalist contextualist position we do not have unconditional certainty about the external world. But why assume we need unconditional certainty? Again, to ask that our knowledge be unconditionally certain seems to ask that a proposition be justified independent of any context; that it be justified despite any conditions. If I believe that my car is blue, then for my belief to be unconditionally justified, it must be the case that "my car is blue" is justified independently of any context. This is of course to give propositions or beliefs objective, intrinsic content, which is foundationalism.

We seem to have positions that butt heads, so to speak; if foundationalism is true, then we have no knowledge, so contextualism is proposed by Williams as the alternative. But for those with foundationalist sympathies contextualism is not convincing. If one is a foundationalist, then one sees foundationalist contexts as having greater priority than others, and if one is a contextualist, then one does not hold this view. This may come down to straightforward sympathies or personal preference. But there may also be good reasons for adopting one or another theory.

The important point raised by McGinn and Skorupski seems, then, not to be just that foundationalist contexts still allow for sceptical argumentation, but that there is some good reason for being a foundationalist; that we have reason to believe that rational assessment of our beliefs is the preferable approach to justification. McGinn and Skorupski contend that there is still reason to be anti-contextualist and adopt foundationalism, overriding all contexts with sceptical considerations.

Skorupski contends that the reason the internalist context of rational assessment of our beliefs takes priority over other contexts is that it is normative. But what is it for something to be normative? If it is just that a procedure that is normative takes priority over other, non-normative procedures, then Skorupski's claim that rational assessment is normative seems to be shorthand for claiming outright that rational assessment of beliefs, thus internalism, thus foundationalism, just takes priority over other contextual beliefs. This is of course not a position we are forced to take if we are not persuaded by foundationalism's worth. We will have to look at reasons for why foundationalism and contextualism may be considered normative or not. A discussion of the normativity of foundationalism and contextualism seems to be in order.

Williams discusses just this topic (Williams, Problems of Knowledge 153-157). His contention is that contextualism better represents our everyday practices, evaluations and assessments of knowledge, and so better reflects our epistemological norms than foundationalism. Williams' discussion is conducted in terms of the Default and Challenge approach to justification and the Prior Grounding Requirement, but the implications are directly applicable to Skorupski's and McGinn's discussions; the Default and Challenge approach is externalist, contextualist and anti-sceptical, whereas that Prior Grounding Requirement is internalist, foundational and sceptical, just as Skorupski's rational assessment of beliefs is.

The Default and Challenge approach and the Prior Ground Requirement are competing views of justification, both of which seem plausible; if the Default and Challenge Approach is correct, then we have a great deal of knowledge, and if the Prior Grounding Requirement is correct, then we have no knowledge, and both approaches have supporters with strong arguments for adopting one over the other. The tie-breaker would seem to be normativity: if one or the other seems to better represent our normative practices, then it is the more reasonable one to adopt. We have not been given reasons for why Skorupski believes his rational assessment, Prior Grounding, internalist, foundationalist, sceptical approach is more representative of our normative practices, but we can assess how well Default and Challenge contextualism reflects our norms in comparison to it.

Williams isolates three ways in which the Default and Challenge, contextualist approach holds normative advantages over the Prior Grounding Requirement, foundationalist approach (Williams, Problems of Knowledge 154).

First, in contextualism, discussion of epistemic entitlement and assessment of beliefs as knowledge allows that both knowledge claimants and challengers share the burden of justification. This meshes with our everyday attitudes and practices, because normally a challenger must be able to justify his challenge, and to give reasons why it is relevant and damaging to the claim at hand. Blank, outright challenges do not hold any weight in everyday discourse; just asking a physicist or historian "Why?" or "How do you know?" does not further our discussion or lead to significant challenges to knowledge claims. The challenger must have in mind a particular way in which the claimant goes wrong, and so must carry some justificatory burden. Skorupski's approach, and the Prior Grounding requirement, on the other hand, do not put any burden on the challenger; the claimant is expected to first justify her belief with rational assessment or evidence, and the challenger may at any time may challenge this assessment. All justificatory burden is placed on the knowledge claimant, who is left with no way to challenge the challenger. But in our practices and assessments of knowledge, we do allow for knowledge claimants to ask for a reason why a challenger disagrees, and we do allow for claimants to show that a challenge is not significant or relevant.

Second, under contextualism, justifying in light of an objection is a matter of explaining away counterarguments and objections, but being justified in the first place is not just a matter of going through a prior process of justification. We may be default entitled to certain beliefs that are methodologically necessary for a particular form of inquiry. We can be justified in holding beliefs, without having to first explicitly give reasons for justification. This seems to reflect our everyday practices: if someone says, "Tom Selleck is in town," and a challenger asks "how do you know?" the claimant may say that he read in the newspaper about how he is filming in town for his next movie. This is acceptable as an explanation; the claimant is default entitled to the belief that the newspaper is a reliable source of information, and the claimant need not first explicitly justify this. Of course, if one has good reason, one might challenge the belief that the newspaper is reliable, but one would have to have good reasons, and in doing so other default entitlements would come into play.

For foundationalism, on the other hand, there is an evidentialist bias that is not obviously present in our knowledge practices and assessments; our everyday assessments of knowledge obviously include production of evidence sometimes, but not every belief need be justified by first going through a process of justification, as is shown by our acceptance of default entitlements. If every belief needed first to be justified before it could be held, then providing the newspaper as evidence for a belief would not be possible.

The foundationalist may respond to this line of reasoning; he may argue that what we have just described is only justification for all practical purposes. The Default and Challenge approach to justification only gives an account of what we call 'knowledge,' but the sceptic will argue that this falls short of what our standards actually ask of us (Williams, Problems of Knowledge 154-155). A difficulty with this objection is that the point of describing knowledge in this practical way was to square off our theories of justification with our normative practices; whichever of the two approaches to justification better resembles our normative practices is the reasonable one to adopt. If the sceptic's claim is that resembling our normative practices is just 'knowledge for all practical purposes,' and that the sceptical approach has the correct concepts of knowledge down, then the sceptic has just conceded that his theory is less representative of our normative practices than the contextualist Default and Challenge approach. If the only reason to make this argument is to avoid refutation of the foundationalist, sceptical approach, then our reasons for adopting the contextualist, Default and Challenge approach is strengthened: the contextualist approach does not require ad hoc supplementation to fit with our everyday notions of knowledge, whereas the foundationalist who accepts the Prior Grounding Requirement must contend that they have the correct view of justification, no matter what our practices are.

Third and finally, normally we are able to attribute knowledge to others because we can defend their reliability. Williams takes his inspiration for this point from Brandom (Brandom 895-908). In this way we can inherit knowledge from experts, by distributing justification among others; if an expert makes a knowledge claim, and we have good reason to take her to be a responsible and reliable knowledge claimant, then we too may hold her claim as knowledge even if we cannot first explicitly give justification for the belief. Many of us reasonably and legitimately believe that electrons and protons exist, even though we have never seen them, and do not know or even understand the experiments that led to their discovery, but trustworthy and reliable experts have done the experiments. Foundationalist conceptions of justification, on the other hand, tie knowledge directly to an individual's ability to first cite evidence before he may claim knowledge. This is at odds with the way knowledge is socially distributed in everyday practices.

So we have good reason to believe that the contextualist approach to knowledge is better representative of our norms than the foundationalist approach. But foundationalists will most definitely argue that there are many good reasons why their theory represents our normative practices as well. Let us grant that there are some convincing arguments for foundationalism well-representing our normative practices. Which are we to choose now, foundationalism or contextualism, when both have made convincing cases for being representative of our normative practices?

Contextualism still has one over-arching benefit over foundationalism (Williams, Problems of Knowledge 153). Foundationalism represents our everyday practices of knowledge assessment and claiming as self-defeating; none of our knowledge claims are ever justified by foundationalism's own standards. We obviously do not take this to be the case; we take our practices and assessment of knowledge to result in many instances of knowledge. If foundationalism insists that we have no knowledge when we take ourselves to, then it does not fit our norms well at all; foundationalism represents what we take to be working knowledge-practices as unworkable.

Under a contextualist approach norms are after all meant to be something we establish and follow;10 if a theory such as foundationalism proposes that our norms do not work when we are the ones making the rules, so to speak, then that theory would seem to be misinterpreting our norms (Williams, Problems of Knowledge 157). This is not meant to imply that they are arbitrary, because we make up the rules; the rules of games are normative as well, but they are not arbitrary. What counts as a goal in hockey is determined by a norm we have decided upon, but that does not mean that rule is arbitrary; if the net were bigger too many goals would be scored, if the net were smaller, no goals would be scored. We may change the rules as we go along, but there are practical limitations. Hockey is a workable game made up of rules we have established and follow according to our norms, just as knowledge is workable and attainable according to the norms that we have established. To insist upon knowledge being unattainable, as a foundationalist would insist, would be like insisting that a hockey net be infinitely small; it would be a game we could not win or lose or do anything in. The foundationalist may insist on his particular conception of knowledge, but it is clearly not representative of our normative practices if it makes knowledge unattainable; our norms allow that knowledge is readily available.

In this way, contextualism clearly wins the battle of normativity: if our norms say that we have knowledge, and foundationalism insists we have none, then it seems very difficult to then insist that foundationalism is better representative of our normative practices than other theories. To insist that we cannot have knowledge even though knowledge is a normative practice is like insisting that we do not know the rules of hockey; we made the rules of hockey, and any theory that says there are no rules or that we do not understand them is in direct conflict with our norms, and so cannot be normative. Contextualism fits well with our everyday practices and assessments of knowledge, and so seems to be more clearly representative of our normative practices.

If norms are standards we set, like rules in a game, then it does not seem to be fitting to talk about a belief being justified independently of any of our practices and attitudes, just as it does not make any sense to talk about a move in hockey being a goal outside of the rules we established for hockey. If this is the case, and if we are going to argue that knowledge is a normative practice, then we seem to have further support for the contextualist approach to knowledge: no beliefs are intrinsically justified, as in foundationalism, because outside of the context of our assessment and practices and norms, beliefs do not seem to have any significance; outside of context, there is no fact of the matter about what a belief or proposition means or whether it is knowledge, because the institution of knowledge is a normative one, dependent on the norms we set and follow in practice and context.

Williams identifies this as a pragmatic approach to knowledge, which is at odds with the epistemological realist approach, which as we have seen holds that there are objective facts about certain types of propositions that fix our epistemological position (Williams, Problems of Knowledge 170). The epistemological realist position leads to foundationalism, and then scepticism, and our position now is: so much the worse for epistemological realism. Epistemological realism is not representative of our norms and leads to scepticism, so it seems reasonable to choose the pragmatic approach and the contextualism that follows. But there are authors who argue that this view is mistaken in some ways.


Stroud makes a few criticisms of Williams. First, he agrees that the causal truism that we would not have any knowledge without senses does not lead to the discovery that epistemological realism in the form of foundationalism is the proper view of knowledge; but he does think that we can nonetheless get a sceptical conclusion from the causal truism alone. Once the sceptic makes his arguments, based only on the fact that experiential knowledge is causally necessary for knowledge, the trouble begins (Stroud, Epistemological Reflection on Knowledge of the External World 351).

Stroud also agrees with Williams that, given the neutrality of experience, the logical gap between experience and knowledge of the world does not itself bring us the epistemologically realist position that experiential knowledge is epistemically prior to other knowledge. But he contends that Williams is wrong to think that the sceptic is simply assuming epistemological realism from the start. Rather, the sceptic uses a particular line of reasoning, using no epistemologically realist assumptions, to come to a sceptical conclusion. At this point the sceptical arguer puts forward the realist doctrine that experiential knowledge takes priority over knowledge (Stroud, Epistemological Reflection on Knowledge of the External World 352). Presumably the argument continues in such a way that once experiential knowledge is shown to be too weak to give us knowledge we are stuck with a radically sceptical epistemology.

This sounds like a familiar line of reasoning that we have covered: the sceptic does not assume anything; rather, epistemological realism in the form of foundationalism is a discovery made out of sceptical considerations. As Stroud says, "The priority of 'experiential knowledge' over knowledge of objects is in that sense a kind of 'discovery' or outcome which we are led to by applying familiar everyday concepts and distinctions in the course of what is admittedly a special philosophical reflection on our knowledge of the world as a whole" (Stroud, Epistemological Reflection on Knowledge of the External World 354). I think we have shown many reasons why this is most likely not a plausible line of sceptical reasoning, so Stroud's contention is not a particularly worrying one at this point. This claim is supported by Stroud's contention that it is best-case sceptical arguments that bring us to epistemological realism. As Stroud says:

To investigate our knowledge of the world in general we cannot investigate each particular item of knowledge of the world, or each occasion on which we came to know something from sense-perception, on its own, one by one. The philosopher considers one such occasion which can be regarded as optimal for gaining perceptual knowledge of the world. He carefully scrutinizes what goes on in that case, and lets it serve as representative of what goes on on all those occasions on which we take the senses to be operating at their best under conditions we regard as best for the acquisition of knowledge from sense-perceptions. (Stroud, Epistemological Reflection on Knowledge of the External World)

This sounds like a familiar approach to which we have responded: the philosopher can only analyze all of our knowledge at once, in general, outside of any context, if he is first an epistemological realist of the sort that supports foundationalism, and the best case scenario only reflects badly on all of our knowledge if we are foundationalists and must accept that there is one ultimate source of knowledge in the form of the senses, so that showing those senses to be defective impugns all of our knowledge. Further, the sceptic requires foundationalism to make those sceptical arguments; otherwise 'scrutinizing' a particular piece of knowledge does nothing to hurt the rest of our knowledge. So foundationalism is not any sort of discovery out of sceptical reasoning. These points have been discussed at length in chapter 2 and in response to DeRose and McGinn and so on. It seems Stroud is likely to have made the familiar but understandable mistake of making sceptical arguments without seeing the epistemologically realist, foundationalist assumptions at play.

But Stroud does not make this argument blindly. He acknowledges Williams' contention that epistemological realism must be assumed for sceptical arguments to take shape. He recognizes that, only in certain contexts does scepticism arise, but he is wary of the contention that epistemological realism specifically must be assumed for such contexts of sceptical inquiry to take place. He still feels the full force of sceptical argumentation, but does not think that epistemological realism must be adopted for the context of sceptical inquiry to begin (Stroud, Epistemological Reflection on Knowledge of the External World 356). Perhaps Stroud's best-case argument is not the most effective way of displaying how epistemological realism need not be assumed for scepticism to be effective, but Stroud may be on to something important in questioning the significance of epistemological realism. He thinks that it is implausible that epistemological realism is the main problem in scepticism.

The general idea seems to be that Stroud is happy not to accept epistemological realism in any form, but still sees scepticism as a threat. Williams' argument is of course that realist presuppositions are often tacit in sceptical arguments, and so even if one does not explicitly accept the epistemological realist doctrine, one may still be unknowingly depending on its principles. But the general point Stroud makes seems to be worth addressing.

There are two ways to interpret Stroud's claim. One is that he is arguing that we may be able to make some sort of sceptical argument, which leads to the acceptance of foundationalism, without any epistemologically realist assumptions. This I believe is the argument that we have addressed already; I think we have shown that foundationalism is required to make sceptical arguments. Foundationalism is at least some kind of epistemological realism, so assuming foundationalism requires some sort of epistemological realism to be in the picture. I think we have shown that there is no reason to adopt epistemological realism, as opposed to a pragmatic, contextualist approach, that comes out of any sceptical arguing, so assuming foundationalism seems to require assuming its particular brand of epistemological realism as well.

But Stroud might also be taken to be arguing that epistemological realism, broadly speaking, is not responsible for scepticism. Only the type of realism that leads to foundationalism, and thus scepticism, is worrisome; epistemological realism is not the main culprit in sceptical arguments, just one type of epistemological realism. Foundationalism is necessary for scepticism, but epistemological realism in general is not. This is worth examining further, and in the next section we will look at an argument by Graham that concludes just this.


Graham gives an incredibly insightful and organized analysis of the issues of realism, foundationalism and scepticism.

Graham identifies a series of epistemic principles that govern various ways of forming and holding beliefs. If an epistemic principle is 'true,' then beliefs formed or held according to it are justified, where different epistemological theories hold different principles to be true (Graham 5). The principles are seven processes of justification, including: a priori insight, introspection, deduction, memory, enumerative induction, inference to the best explanation, perception and testimony. (Graham 8)11 A strict epistemological theory will only allow that a few of these principles are true, whereas a more lenient theory will hold that more of the principles are true. Further, the order I presented the principles in is representative of how likely a strict theory is to accept them, so that the principles for a priori insight, introspection and deduction are accepted by all theories, but the principles for perception and testimony are only accepted by the most lenient theories.

Next Graham defines three kinds of foundationalism, each holding different principles to be true (Graham 8):

  • Reactionary Foundationalism:
  • A Priori, Introspection, Deduction
  • Conservative Foundationalism:
  • A Priori, Introspection, Deduction
  • Induction, Inference to the Best Explanation, Memory
  • Moderate Foundationalism:
  • A Priori, Introspection, Deduction
  • Induction, Inference to the Best Explanation, Memory
  • Perception

Graham also identifies four theories about the nature of justification: Cartesianism, Reliabilism, Intuitionism, and Pragmatism. These are distinguished by their loyalties to one side or the other of two distinctions: the Actual-Realist/ Proper-Aim distinction and the Fundamentalist/Non-Fundamentalist distinction.

The Actual-Realist and Proper-Aim views are about the relation between justification and truth: the Actual-Realist view is that if a belief is justified it is objectively more likely to be true than an unjustified belief, whereas the Proper-Aim view does not require that a belief is more likely to be objectively true if justified, only that "justification properly aims belief at truth insofar as truth is the aim or norm" (Graham 5).

The Fundamentalist and Non-Fundamentalist views are about the epistemic status of the epistemic principles listed earlier. The Fundamentalist holds that epistemic principles are a priori necessary truths, whereas the Non-Fundamentalist holds that they are empirical contingent truths (Graham 5).

Cartesianism is an Actual-Result, Fundamentalist theory, and holds that "justification supervenes upon necessarily reliable belief forming and holding processes," so that "A principle is true because it is a priori known that the psychological process it governs makes beliefs more likely to be true in all possible worlds" (Graham 5). Reliabilism is an Actual-Result, Non-Fundamentalist theory, which holds that "a principle is true because it is empirically known that the psychological process it governs makes beliefs more likely to be true. The process need only be contingently reliable, reliable in the circumstances of use" (Graham 5). Intuitionism is a Proper-Aim, Fundamentalist theory, which holds that "epistemic principles are true because they are a priori truths, whether or not the processes governed are de facto reliable" (Graham 6). And Pragmatism is a Proper-Aim, Non-Fundamentalist theory, which holds that, "The principles (what Williams calls 'norms') are true because we accept them; we make them the norms; they are not norms simply in virtue of the way things are, independent of our conversational practices" (Graham 7). Pragmatism is Williams' epistemological position, which we have canvassed in this and the last chapter.

We are now in a position to see which theories hold which epistemic principles. Cartesianism and Intuitionism are the most relevant for our discussion. Cartesianism is the only theory strict enough to accept Reactionary Foundationalism, because it will only accept a priori reasoning, deduction and introspection as true principles of justification. Intuitionism tends more towards Moderate Foundationalism. (Graham 8-9)

So the question hinted at by Stroud was whether or not epistemological realism is necessary for scepticism. How does epistemological realism fit into the schema Graham has illustrated for us? Graham explains that Williams' description is vague enough to allow for weak and strong epistemological realism (Graham 18-19). Graham interprets Williams' epistemological realism as the view that certain processes of belief formation are necessarily justification conferring, which is just the Fundamentalism side of the Fundamentalist/Non-Fundamentalist distinction about the status of epistemic principles (Graham 8). As such, epistemological realism may still either be Actual-Result Fundamentalism, or Proper-Aim Fundamentalism. We have identified Actual-Result Fundamentalism with the Cartesian theory, which is sympathetic to Reactionary Foundationalism and its strict acceptance of epistemic principles, and Proper-Aim Fundamentalism with the Intuitionist theory, which is sympathetic to Moderate Foundationalism and its more lenient acceptance of epistemic principles. Thus Williams' epistemological realism may either be strong realism, which is the Cartesian, Reactionary Foundationalist theory, or weak realism, which is the Intuitionist, Moderate Foundationalist theory. But Williams' depiction of realism restricts epistemological realism to strong realism; Williams' "epistemological realism" is just Cartesianism (Graham 19).

Graham's contention is that strong epistemological realism leads to scepticism, but weak epistemological realism does not, so that epistemological realism as such does not necessarily lead to scepticism (Graham 19-20). Graham characterizes radical scepticism, what he calls academic scepticism, in the following way: radical scepticism contends that it is a priori known that no beliefs about the external world are justified, so that they are all epistemically worthless (Graham 10-11). The academic sceptic requires two premises, first, that there is the possibility for massive error in our beliefs about the external world, such as brain-in-a-vat scenarios, and second, that the possibility for massive error means that no beliefs about the external world are justified;12 radical scepticism requires an epistemological theory that holds that the possibility for massive error rules out justification.

What sort of theory holds this? If a belief is unjustified when there is the possibility for massive error, then for a belief to be justified it must be based on a process in which massive error is not possible; for a belief to be justified it must be based on necessarily reliable processes, so that any possibility for massive error is ruled out. The only theory that holds this is Cartesianism: all beliefs must be formed by a priori necessarily true principles that make a belief more objectively likely to be true (Actual-Realist Fundamentalism,) so that only the Reactionary Foundationalist principles of a priori insight, introspection and deduction are acceptable. Cartesian, Reactionary Foundationalism is of course Williams' strong epistemological realism; strong realism implies radical scepticism (Graham 10-11).

Weak realism, or Intuitionist, Moderate Foundationalism, does not imply scepticism. We may grant the possibility of massive error, but Moderate Foundationlism's principles do not require that a belief be based on a necessarily reliable process for it to be justified; justification properly aims at truth insofar as truth is a norm, so that perception, for example, is a justificatory principle. If a belief is formed through perception, which Intuitionist, Moderate Foundationalism allows, then it is justified, so that radical scepticism is not a threat. Further, this is weak realism, which falls under the umbrella of Williams' epistemological realism. Realism does allow for non-sceptical interpretations, so that the contextualist and the Intuitionist, Moderate Foundationalist are in the same anti-sceptical boat. Williams' epistemological realism is too narrow, and does not allow for non-sceptical realisms. Either that or it is too broad, in which case it is wrong that epistemological realism necessarily leads to scepticism; if Williams means by Epistemological realism just the Cartesian interpretation, then he has failed to account for non-sceptical forms of realism, but if he means the broader interpretation of realism, then he is wrong that realism leads to scepticism, because Intuitionism is both anti-sceptical and realist. Epistemological realism as such is not sufficient for scepticism, but a brand of it, specifically the Cartesian, Reactionary Foundationalist brand, does lead to it.

So a certain type of epistemological realism is sufficient to reach scepticism, but it is not necessary for epistemologically realist positions to be sceptical. For full blown scepticism we must take the Reactionary Foundationalist route of epistemological realism, or perhaps the Conservative route, but there are other anti-sceptical approaches besides Williams'. This is similar to a point made by Rorty; he explains that Davidson's approach to anti-scepticism works just as well as Williams', despite Williams' contention that Davidson's approach is coherentist. Rorty explains that Davidson bears little resemblance to the coherence theorists that Williams argues against, and is instead giving a theoretical diagnosis of his own: just as Graham has shown that Williams' approach is not the only anti-sceptical path, because weak realism is also anti-sceptical, Rorty too has shown that Williams is not the only successful anti-scepticism, because Davidson's approach works just as well (Rorty 156-163).

Nonetheless, I think we have made a case for showing that foundationalism is required to make radically sceptical arguments, so that we have shown that sceptical arguments tacitly presuppose foundationalism. Foundationalism is presupposed by scepticism, so that a certain type of realism must also be presupposed, but other types of realism can be anti-sceptical as well.