1. PRITCHARD'S BISCOPIC SOLUTION TO RADICAL SKEPTICISM

According to Pritchard, radical skepticism is best seen as a paradox that results from logical inconsistency between claims that are either intuitive or apparently immediate consequences of intuitive claims. He argues that there are actually two skeptical paradoxes, one based on closure and another based on a principle of rational underdetermination.

2. ASSESSING THE ARGUMENT FOR THE WITTGENSTEINIAN PROPOSAL

Pritchard finds neo-Moorean epistemological disjunctivism epistemically immodest because it violates No Knowledge. As I indicated in the introduction, what proper epistemic modesty amounts to is contestable, and in section 4 I shall present an alternative account of it. For the time being, let me say that I do not share Pritchard's sense that No Knowledge is a deep-seated intuition, and as a result I don't find its denial so problematic. However, Pritchard also gives an argument for the claim that anti-skeptical commitments are not beliefs, and if it is sound then neo-Mooreanism is doomed in any case. My aim in this section is to critically examine this argument.

Before turning to it, I must say something about Pritchard's theory of hinge commitments. On his account, our hinge commitments are those commitments that we are most confident in and are allegedly necessary for any rational evaluations of beliefs. Pritchard distinguishes three types of hinge commitment: personal hinge commitments, anti-skeptical hinge commitments, and the “über hinge commitment.” The über hinge commitment is a commitment to the proposition that one is not “radically and fundamentally mistaken in one's beliefs” (the über hinge proposition) (Pritchard Reference Pritchard2016: 95). Pritchard thinks that everyone is at least implicitly committed to the über hinge proposition, and that this commitment manifests itself in hinge commitments of the other two types. First, personal hinge commitments concretize or “codify” an individual's über hinge commitment in light of his personal circumstances. For instance, for most people the commitment to the proposition I have hands is a personal hinge commitment, since if they were mistaken about this then their beliefs about the world would be radically and fundamentally mistaken. Second, commitments to the denials of radical skeptical hypotheses also codify the über hinge commitment: since radical skeptical hypotheses are designed to call the über hinge proposition into question, anyone who is committed to that proposition and is aware of skeptical hypotheses will have (or perhaps, be aware of) a commitment to their denials.Footnote ³

The central contention of the Wittgensteinian proposal is that no hinge commitments – including commitments to the denials of radical skeptical hypotheses – are beliefs. Pritchard's central argument for this conclusion is as follows:

The constitutive connection mentioned in (P1) is that beliefs are essentially responsive to rational considerations that bear on their truth. Specifically: it is not possible for a subject to believe that p while taking herself to have no reason for thinking that p is true, and it is not possible for a subject to believe that p while taking herself to have overwhelming reason for thinking that p is false (Pritchard Reference Pritchard2016: 90).Footnote ⁴

The more controversial premise is (P2). Pritchard's defense of this premise turns on two claims related to the über hinge proposition: (1) It is not possible for a subject to have a rationally supported belief in the über hinge proposition. (2) From this it follows that a subject also cannot have rationally supported beliefs in personal and anti-skeptical hinge propositions, insofar as commitments to these propositions simply “codify” one's commitment to the über hinge proposition. One might question whether (2) really follows from (1), but I will concentrate on (1). In its support of (1) Pritchard writes, “[W]hat possible reason could we have for holding the über hinge commitment? Whatever grounds one cited would already presuppose the truth of this commitment after all” (Pritchard Reference Pritchard2016: 98). The idea seems to be that any attempt to provide rational support for the über hinge proposition would be viciously circular, for any grounds one cited in its support would already presuppose its truth in a manner that defeats the attempt.

However, a neo-Moorean epistemological disjunctivist can deny that the sense in which the über hinge proposition needs to be presupposed in an attempt to rationally support it is a sense that renders the attempt viciously circular in any obvious way.Footnote ⁵ Consider the following justificatory structure (where → stands for “justifies belief in”):

1. 1. Perception, memory, testimony, etc. →

2. 2. Everyday propositions (I'm standing up, Napoleon won the Battle of Austerlitz in 1805, etc.) →

3. 3. Anti-skeptical propositions (An external world exists, it has existed for more than five minutes, etc.) →

4. 4. The über hinge proposition.Footnote ⁶

Now let us distinguish between three types of circularity this justificatory structure might exhibit:

If the structure exhibited Type 1 circularity then it would be uncontroversially vicious. Fortunately, however, an epistemological disjunctivist can hold that a subject does not need a justified belief in the über hinge proposition in order for the sources mentioned in (1) to justify beliefs in everyday propositions. For example, in a good case a subject can be justified (indeed, can know) an everyday proposition p on the basis of seeing that p. And for this belief to be justified it is not the case that the subject must also have a justified belief in the über hinge proposition. Rather, the fact that she sees that p by itself constitutes a conclusive rational basis for the knowledgeable belief that p.Footnote ⁷ The subject can then deduce from her perceptual knowledge the anti-skeptical proposition that an external world exists, which can in turn constitute part of a rational basis for justifiedly believing the über hinge proposition.

I used perception as an example, but the epistemological disjunctivist can take a similar line about the other sources mentioned in (1). For instance, in a paradigmatic case of testimonial knowledge, a subject's rational basis for believing that Napoleon won the Battle of Austerlitz in 1805 can be that she heard that this is the case, where “heard that,” like “sees that,” is factive.Footnote ⁸ Likewise, in a paradigmatic case of memorial knowledge, a subject's rational basis for believing that she was alive in 1985 can be that she remembers being alive in 1985, where “remembers” is factive. In these cases, if the subject forms the beliefs that Napoleon won the Battle of Austerlitz in 1805 and that she was alive in 1985 on the basis of the appropriate factive states, then she will have rationally grounded knowledge of the propositions in question. Either of these pieces of knowledge can then serve as a rational basis from which she can deduce the proposition that the world has existed for longer than five minutes.

While the justificatory structure above does not exhibit Type 1 circularity, it does exhibit Type 2 circularity. However, it is not obvious that there is anything vicious about this. Of course it's the case that in order for a subject to acquire a quantity of knowledgeable beliefs about everyday propositions on the basis of the sources mentioned in (1) that is sufficient to form a rational basis for acquiring a justified belief in the über hinge proposition he cannot actually be radically and fundamentally deceived about the world. So while the truth of the über hinge proposition is “presupposed” in the sense that it must be true in order for the subject to acquire a justified belief that it is true, this seems unproblematic.

Circularity of Type 3 often goes by the name “epistemic circularity” (Alston Reference Alston1986). Is the justificatory structure above epistemically circular? Arguably, it is. For being committed to the über hinge proposition involves inter alia being committed to the idea that the beliefs one forms about the world on the basis of the epistemic sources mentioned in (1) are generally true. Therefore, the structure embodies an attempt to justify the reliability of certain belief sources on the basis of the deliverances of those very sources.

Epistemic circularity is controversial, and there are two views of the matter. One camp contends that epistemically circular arguments are always vicious, in that the premises of such an argument never provide epistemic support for its conclusion. The other camp holds that epistemically circular arguments are not necessarily vicious.Footnote ⁹ I fall in the latter camp, but since it would take us too far afield to seriously engage the matter I will confine myself to making two observations.

First, it is sometimes objected that employing any epistemically circular argument is akin to the absurd method of using astrology to try to vindicate the reliability of astrology (Fumerton Reference Fumerton1995). But to this it may be replied that what's bad about trying to using astrology to vindicate itself is not that it's epistemically circular but that we possess overwhelming evidence that discredits astrology. Second, there seem to be some instances of epistemically circular arguments that are perfectly cogent. For instance, it seems I can acquire rational support for thinking that my memory is reliable by using my memory: I can remember times I correctly remembered information I was trying to recall, that I don't typically forget important facts, and that I am not habitually subject to memory failures and false memories.

Someone might object that there is a difference between using a single faculty to acquire rational support for the reliability of itself and attempting to use one's entire suite of epistemic faculties to acquire rational support for the über hinge proposition.Footnote ¹⁰ When the reliability of a single faculty is at issue, it might be said, we typically have evidence that bears on its reliability that comes from sources other than the faculty under consideration, and in vindicating the faculty the subject can note that she does not have evidence against its reliability that comes from independent sources. But this sort of independent evidence is not available when we try to acquire a rationally supported belief in the über hinge proposition.

The worry is that epistemic circular arguments for the reliability of a certain source or set of sources S are vicious when all the evidence regarding the reliability of S is supplied by S. However, it does not seem to be the case that this is always objectionable, as an epistemically circular argument for the reliability of memory illustrates.Footnote ¹¹ The main sort of non-memorial evidence one could have for the reliability of one's memory would be testimonial evidence. Since I am not currently speaking with anyone who can attest to the reliability of my memory, if I am going to try and acquire rational support for the reliability of my memory here and now then all of the evidence available to me will have to come from my memory.Footnote ¹² Yet it seems, nonetheless, that I can here and now acquire rational support for the reliability of my memory by using my memory.

Employing our entire suite of epistemic faculties to acquire rational support for belief in the über hinge proposition is in this respect analogous to using one's memory to acquire rational support for the belief that one's memory is reliable. Just as we can imagine one's memory supplying evidence that undermines belief in its reliability, we can imagine the possibility that the output provided by the totality of our epistemic facilities is self-undermining. Given that this possibility is not realized, it's hard to see why the justificatory structure for the über hinge proposition above should be thought to be vicious, especially if we assume that epistemological disjunctivism is true. After all, if epistemological disjunctivism is correct, then the epistemic powers mentioned in (1) provide one with conclusive reasons for believing the sorts of everyday propositions at issue in (2), and (3) and (4) then follow by inferences that seem perfectly cogent. At the very least, these considerations should shift the burden of proof on to the critic.

I acknowledge that there is a sense in which a commitment to the über hinge proposition expresses what Pritchard calls a “visceral” attitude. It does seem that we normally operate on the assumption that our beliefs about the world are not radically mistaken and that this assumption is not typically held on the basis of reasoning. Rather, it forms part of our general and unarticulated background understanding of the world. But it does not follow that it is not possible for someone to hold a rationally grounded belief in the über hinge proposition on the basis I have described.

The conclusion of this section is that a neo-Moorean epistemological disjunctivist may reasonably doubt the second premise of Pritchard's argument for the Wittgensteinian proposal. In the next section I turn to the motivation for that proposal.

3. THE MOTIVATION FOR THE WITTGENSTEINIAN PROPOSAL

Recall that Pritchard motivates the Wittgensteinian proposal through a consideration of the closure-based skeptical paradox in light of epistemological disjunctivism. If we accept epistemological disjunctivism then there appears to be a simple route to rejecting (SC1), i.e., to rejecting the claim that we cannot have rationally grounded knowledge of the denials of radical skeptical hypotheses. For if epistemological disjunctivism is true, I can have perceptual knowledge of everyday propositions, and I can then apparently deduce from the truth of an everyday proposition the conclusion that I am not a BIV. The Wittgensteinian proposal is intended to block this neo-Moorean route to denying (SC1) without having to deny Closure _RK.

The ultimate motivation for the Wittgensteinian proposal therefore lies in Pritchard's commitment to:

Why accept No Knowledge? In his initial presentation of the closure-based skeptical paradox, Pritchard appeals to the subjective indistinguishability of the BIV scenario from ordinary perceptual circumstances:

Call the proposition that radical skeptical scenarios are subjectively indistinguishable from normal perceptual conditions “Rationale.” I now want to argue that given epistemological disjunctivism, there are two problems with attempting to ground No Knowledge on Rationale.

The first problem arises when we turn to certain cases where local skeptical hypotheses are under consideration. Suppose John wakes up in unfamiliar surroundings and looks at what appears to be a red wall. As a matter of fact the wall is red and he is looking at it in paradigmatic conditions for acquiring perceptual knowledge of its color. We can also stipulate that John doesn't have any specific reason for doubting the wall is red, that there is no way for him to check whether the lighting is deceptive, and that he has no views about the likelihood that his captors would deceive him about its color. John then considers whether the wall is actually a white wall that is being undetectably illuminated by red lighting and he thinks to himself, “I see that the wall is red and that means it is red. But if the wall is red then it is not a white wall that is being undetectably illuminated by red lighting.” On this basis he comes to believe the denial of the skeptical hypothesis that the wall is white but undetectably illuminated by red lighting.

If one holds No Knowledge on the basis of Rationale, it seems one should also hold that John cannot have rationally grounded knowledge that it is not the case that the wall is white and undetectably illuminated by red lighting. After all, the skeptical hypothesis that the wall is white and undetectably illuminated by red lighting is subjectively indistinguishable from normal perceptual conditions, there is no possibility of doing checks to rule it out, and John has no relevant background information about the likelihood of deception. The trouble is that there is nothing in neo-Wittgensteinian epistemological disjunctivism that can explain why John's belief that the wall is not white but undetectably illuminated by red lighting is not a bit of rationally grounded knowledge. An epistemological disjunctivist is committed to the claim that John's belief that the wall is red is a case of rationally grounded perceptual knowledge. We may also suppose that John's deduction of the anti-skeptical proposition that the wall is not white but undetectably illuminated by red lighting is a competent one. And in this case the Wittgensteinian proposal cannot be used to block the conclusion that John's resulting endorsement of the anti-skeptical proposition is a case of rationally grounded knowledge. The reason is that John's endorsement of the claim that the wall is not white but undetectably illuminated by red lighting is not plausibly a hinge commitment: even if he were mistaken about this it would not imply that his beliefs about the world were radically and fundamentally mistaken.

A second problem arises when we ask how Rationale is supposed to ground No Knowledge. The typical path from Rationale to No Knowledge seems to be as follows:

1. (1) The fact that radical skeptical scenarios are subjectively indistinguishable from normal perceptual conditions means that we cannot discriminate, on the basis of the senses, between radical skeptical hypotheses and normal perceptual conditions.

2. (2) If we cannot discriminate, on the basis of the senses, between radical skeptical hypotheses and normal perceptual conditions, then the rational support for our perceptual beliefs about our environment provided by our senses is compatible with the truth of radical skeptical hypotheses and the falsity of those beliefs.

3. (3) If the rational support for our perceptual beliefs about our environment provided by our senses is compatible with the truth of radical skeptical hypotheses and the falsity of those beliefs, then we cannot have rationally grounded knowledge of the denials of radical skeptical hypotheses.

4. (4) So, we cannot have rationally grounded knowledge of the denials of radical skeptical hypotheses.

The problem is that there is a step in this reasoning that epistemological disjunctivists should reject, namely, (2). For the point of epistemological disjunctivism is that, despite the indiscriminability between paradigmatic perceptual conditions and radical skeptical hypotheses, in the former case the rational support provided by one's perpetual powers is that one sees that (or otherwise perceives that) p, and this rational support is not compatible with the truth of radical skeptical hypotheses and the falsity of one's everyday perceptual beliefs.Footnote ¹³

This account of the origins of the No Knowledge intuition suggests the two skeptical paradoxes that Pritchard discusses have a common root, namely, a certain conception of the rational support provided by perceptual experience, according to which even in good cases one's experiences never constitute grounds for perceptual beliefs that guarantee their truth. Pritchard explains how this conception underlies (SU1) of the underdetermination-based paradox, and if I am correct then it also underlies (SC1) of the closure-based paradox.

Perhaps Pritchard does not actually think that No Knowledge is to be grounded on Rationale. It might be that he invokes Rationale simply as a presentational device in his initial motivation of the skeptical paradox.Footnote ¹⁴ But then why should we accept it? An alternative would be to claim that the No Knowledge is a basic, pre-theoretical intuition that epistemologists should try to respect in their theory construction. The idea would be that No Knowledge has a status akin to the status that many moral philosophers believe is possessed by common pre-theoretical moral intuitions, such as the intuition that it is wrong for a surgeon to kill one patient in order to make his organs available for transplant into five others. If this is Pritchard's view, he could say that (in the case of local skepticism I raised above) while the notion that John can know that the wall is not white but undetectably illuminated by red lighting might seem surprising to some, this is something we can accept or even embrace, so long as a similar move is not available in the case of radical skepticism. It would also enable him to avoid the problem that the typical route from Rationale to No Knowledge relies on an intuition that epistemological disjunctivists should reject.

Is it plausible to accord No Knowledge the status of a basic intuition? My own sense is that it is not. In my own experience most people are positively incredulous upon first hearing radical skeptical claims such as the claim that we cannot know we are not a BIV. If they come to hold No Knowledge at all, it is only after engaging with skeptical arguments that appeal to the subjective indistinguishability of radical skeptical scenarios from normal perceptual conditions. But this takes us back to the inference from Rationale to No Knowledge, and we have seen that epistemological disjunctivists should reject this inference.

In this section I have raised problems for Pritchard's motivation for the Wittgensteinian proposal. My argument can be summarized as a dilemma. Either Pritchard grounds No Knowledge on its most plausible rationale or he asserts that No Knowledge is a basic intuition. If he takes the former option he should also hold that in certain cases subjects cannot have rationally grounded knowledge of the denials of local skeptical hypotheses, but he lacks the resources to explain why this is so. Moreover, the inference from Rationale to No Knowledge is in tension with his endorsement of epistemological disjunctivism. On the other hand, the claim that No Knowledge is a basic intuition is simply not very plausible.

My view is that epistemological disjunctivists should go neo-Moorean. But I have also acknowledged that radical skepticism forces us to accept a degree of modesty in our epistemic self-assessment. Therefore, in the next section I articulate an alternative conception of epistemic modesty that is compatible with neo-Moorean epistemological disjunctivism.

4. MODERATE NEO-MOOREAN EPISTEMOLOGICAL DISJUNCTIVISM

Radical skepticism is usually taken as posing a challenge to our claim to have environmental knowledge. I have no problem referring to this form of skepticism as “Cartesian,” but it is worth noting that for Descartes himself the skeptical arguments of the First Mediation were not supposed to challenge the claim that we possess knowledge of the external world, but rather the notion that we can know anything about it with absolute certainty.Footnote ¹⁵ While absolute certainty is a higher degree of certainty than is ever required for purposes of everyday life, according to Descartes it is required for scientific knowledge, or scientia. Descartes's rationalist claim is that only reason, and not the senses, is a source of absolute certainty.

Moore also understood radical skepticism as challenging the idea that we possess certainty about the external world. For example, in “Certainty” he claims that he is certain that he is standing up, and he imagines a skeptical philosopher countering, “You are wrong: you do not know you are standing up; it is not absolutely certain that you are; there is some chance, though perhaps only a small one, that you are not” (Moore Reference Moore1959a: 245). And Moore's own anti-skeptical stance includes the claim that he possesses certainty on the basis of sense perception, not just knowledge.Footnote ¹⁶

There is a middle ground between Moore and the skeptic, however. According to what I shall call “moderate neo-Mooreanism” a subject can have rationally grounded environmental knowledge on the basis of perception, and she can use this knowledge as a basis for rationally grounded knowledge of the denials of radical skeptical hypotheses. But, crucially, neither of these sorts of knowledge amount to absolute certainty. Moderate neo-Mooreanism does not concede to skepticism the alarming claim that we cannot have rationally grounded knowledge of the world, but it does incorporate a degree of epistemic modesty insofar as it concedes that the senses are not a source of certain knowledge. In this section I defend a specific version of this view: moderate neo-Moorean epistemological disjunctivism.

In the passage from “Certainty” quoted above, the skeptic alleges that Moore cannot be absolutely certain that he is standing up on the ground that there is a small chance that he is not. This suggests the following account of (absolute) certainty:

This explains why, if a subject is certain that p, then it is not reasonable for her to have any doubts about whether p; for if there is no chance of one's being in error about p then one lacks any rational basis for doubting p. Moreover, if we were to ask someone who isn't sure about whether p why she is uncertain, a natural response would be for her to say, “Because there is some possibility that not-p, even if it is a remote one.”

If Certainty is true, then a full account of certainty requires an account of epistemic chance and epistemic modality. Lacking any such account, I will not attempt to provide one. Instead I will rely on our intuitive grasp of these notions, which I believe is sufficiently robust for my purposes. But note that it's essential to moderate neo-Mooreanism that a subject can know that p and not be certain about whether p, and this is incompatible with a common theory of epistemic modality, according to which p is possible for S if and only if p is consistent with what S knows. As Fantl and McGrath point out, however, we intuitively do take ourselves to know many things that have an epistemic chance less than 1, e.g., that Plato was the teacher of Aristotle or that Napoleon was the emperor of France (Fantl and McGrath Reference Fantl and McGrath2011: 12).

Let us say that a subject has epistemically fallible knowledge that p just in case she knows that p even though there is a non-zero chance for her that not-p.Footnote ¹⁷ Since a subject's being certain that p requires there be no epistemic chance for her that not-p, if her knowledge that p is epistemically fallible, then her knowledge that p will not be a case of certain knowledge. If moderate neo-Mooreanism is correct, then perceptual knowledge is always epistemically fallible knowledge.

Epistemic fallibilism should be distinguished from two other versions of fallibilism, logical fallibilism and rational fallibilism. According to logical fallibilism, it is possible for a subject to know that p even though she knows that p on the basis of evidence that does not entail that p, while according to rational fallibilism, it is possible for a subject to know that p on the basis of a reason r even though r is a reason that she can have if p is false.Footnote ¹⁸ It is important to distinguish these different versions of fallibilism because if epistemological disjunctivism is true then paradigmatic cases of perceptual knowledge are cases of logically and rationally infallible knowledge. For in paradigmatic cases a subject knows that p on the basis of seeing that p, where seeing that p both entails that p and is not a reason that she can have for believing that p if p is false.

Nonetheless, it seems that epistemological disjunctivists should hold that perceptual knowledge is at least sometimes epistemically fallible. Consider the following case:

In this case you are looking at a 1934-S dollar, the conditions are optimal for acquiring perceptual knowledge, and you have no specific reason for believing the seller is trying to deceive you. An epistemological disjunctivist should therefore count this as a good case in which you acquire perceptual knowledge that there is a 1934-S dollar in front of you on the basis of the reflectively accessible ground that you see that there is a 1934-S dollar in front of you.Footnote ¹⁹ Yet intuitively it seems that there is an epistemic chance for you that the coin is not a 1934-S dollar.

It may seem difficult, however, for a disjunctivist to account for the existence of epistemically fallible perceptual knowledge. How is it possible for perceptual knowledge to be fallible if it is based on grounds that are both factive and reflectively accessible? Consider that if epistemological disjunctivism is true then in Peace Dollar you can reason to yourself as follows:

In this way you can come to know, by reflection and a priori reasoning, that your perceptual evidence guarantees that the coin you are looking at is a 1934-S dollar. But if you know this, then how can there be a chance for you that the coin is not a 1934-S dollar?

A tempting reaction is to say that you cannot be certain that there is a 1934-S dollar in front of you because you cannot know by reflection and a priori reasoning that your perceptual evidence guarantees this is the case. And you can't know by reflection and a priori reasoning that your perceptual evidence guarantees there is a 1934-S dollar in front of you because you can't know by reflection that you see that there is one in front of you (P1). Rather, all you can know by reflection is that it seems to you as if there is a 1934-S dollar in front of you, so that either you see that there is a 1934-S dollar in front of you or you merely seem to see that.Footnote ²⁰ On this line of reasoning, the fact that you cannot be certain that the coin is a 1934-S dollar shows that epistemological disjunctivism is false.

But there is room for a more disjunctivist-friendly response. According to this alternative, you know the conclusion of the argument on the basis of the premises, but this knowledge is itself epistemically fallible. The reason is that your reflective knowledge that you see there is a 1934-S dollar in front of you, from which your knowledge of the conclusion is derived, is a piece of epistemically fallible knowledge. In other words, both your perceptual knowledge that there is a 1934-S dollar in front of you and your reflective knowledge that you see that there is a 1934-S dollar in front of you are epistemically fallible, and so uncertain.

This view rejects the “Cartesian” notion that, if we are sufficiently attentive, we can never be in error about what our own experiential states are. John McDowell has presented a view of this sort in a pair of recent essays (Reference McDowell2011, Reference McDowell2013).Footnote ²¹ On his theory, for a range of environmental propositions we have the power to perceive that p in suitable circumstances.Footnote ²² But insofar as we are rational beings, we also have the power to be in a position to know that we perceive that p, where this power is not a distinct power from the power to perceive that p. Rather, our perceptual powers are self-conscious powers. To say that a perceptual power is self-conscious is to say that in a non-defective actualization the subject recognizes her perceptual state as a perception that p, so that when she knows that p through a non-defective actualization of the power she is in a position to explain why she believes that p, namely, on the basis of recognizing her perceptual state as a perception that p. Therefore, in a non-defective actualization of a self-conscious perceptual power, the subject knows that p on the basis of a ground (perceiving that p) that is reflectively accessible and entails the truth of her belief.

As McDowell explains it, fallibility is most properly a property of a power or a subject qua possessor of a power (Reference McDowell2011: 37). To say that a power is fallible is to say that it is liable to defective actualizations, in which it does not do what it is defined as a power to do. Our perceptual powers are fallible because they are liable to defective actualizations in which they do not do what they are specified as powers to do – provide us with factive rational support for perceptual knowledge (such as seeing that p) – but merely seem to do that. Due to the fact that merely seeming to perceive that p is subjectively indistinguishable from perceiving that p, in a case of a defective actualization a subject may mistakenly think she recognizes her perceptual state as a perception that p. In such a situation, the subject does not know that p (even if she thinks she does) and is not in a position to reflectively know that she perceives that p (even if she thinks she is).

Fallibility, then, is in the first instance a property of our perceptual powers rather than the knowledge they provide. But the fact that our perceptual powers are fallible explains why the knowledge they give rise to is epistemically fallible. This can be illustrated by the Peace Dollar case. I concluded above that your belief that there is a 1934-S dollar in front of you is a case of knowledge. Why, then, are you uncertain that the coin is a 1934-S dollar?

The reason is that you know there exist very good counterfeits that you cannot distinguish from genuine peace dollars. This just is the recognition that your power to see that there is a 1934-S peace dollar in front of you is fallible. For what you recognize is that your power is liable to defective actualizations in circumstances in which you are presented with a very good counterfeit. In such a situation you would not see that there is a 1934-S peace dollar in front of you, but your perceptual experience would be subjectively indistinguishable from seeing that there is a 1934-S dollar in front of you. Since you recognize that you cannot distinguish between genuine peace dollars and very good counterfeits, and therefore between defective and non-defective actualizations of the relevant perceptual power, it seems you should admit there is a non-zero epistemic chance for you that you're currently in a situation in which you are looking at a very good counterfeit and your power is undergoing a defective actualization. The upshot is that your knowledge, springing as it does from a fallible perceptual power, is epistemically fallible.

But if epistemological disjunctivism is true, can't you legitimately think that there is no chance that the coin is a very good counterfeit? After all, insofar as your belief that the coin is a 1934-S dollar is based on your recognition of your perceptual state as a seeing that there is a 1934-S dollar in front of you, you can reflectively know that you see that there is a 1934-S dollar in front of you, which excludes the possibility that it's a very good counterfeit. What this reasoning overlooks is that your power to know by reflection that you see that there is a 1934-S dollar in front of you is a fallible power – indeed, the very same power as the power to see that there is a 1934-S dollar in front of you. So just as there is a chance for you that you are looking at a very good counterfeit and your power to see that there is a 1934-S dollar is undergoing a defective but subjectively indistinguishable actualization, there is an equal chance that you are looking at a very good counterfeit and your power to recognize your perceptual state as a seeing that there is a 1934-S peace dollar in front of you is undergoing a defective but subjectively indistinguishable actualization.

What I have done so far is to make the case that epistemological disjunctivists have the resources to make sense of epistemically fallible perceptual knowledge. But moderate neo-Moorean epistemological disjunctivism goes further, for it claims that perceptual knowledge is always epistemically fallible, and so never certain. This is where radical skepticism comes into play, for the moderate neo-Moorean thesis can be established through the use of radical skeptical hypotheses and a version of closure.

In Peace Dollar the fact that you cannot be certain that the coin you are looking at is a genuine peace dollar can be explained by the following principle:

Closure _C is eminently plausible. Since when q is deduced from p, q is logically entailed by p, competent deductions ought to preserve certainty. You cannot be certain that the coin you are looking at is a genuine peace dollar, then, because if you could then you could also be certain that it is not a very good counterfeit by undertaking the appropriate competent deduction. But you cannot be certain that the coin is not a very good counterfeit, for as we have seen, there is some epistemic chance for you that you are in a situation in which your power to see that there is a 1934-S dollar in front of you is undergoing a subjectively indistinguishable defective actualization.

The lesson of radical skeptical skepticism, though, is that there are never circumstances in which there is no epistemic chance for one that one's perceptual powers are undergoing defective actualizations that are subjectively indistinguishable from non-defective ones. Radical skeptical hypotheses are designed to show that even in apparently ideal conditions there is still some epistemic chance that our perceptual powers are undergoing a defective actualization in a circumstance in which we are massively deceived about the world. For instance, none of us can distinguish between a situation in which we have hands and a situation in which we are a handless BIV. And since we cannot distinguish between these situations, there exists some chance for us, however miniscule,Footnote ²³ that we do not have hands but our power to see that we have hands is undergoing a defective actualization whereby we merely seem to see we have hands. If that is right, then it follows that no one can be absolutely certain that he is not a handless BIV. Finally, by Closure _C it follows that if one cannot be certain that one is not a handless BIV, then one cannot be certain that one has hands.

Moderate neo-Mooreanism therefore specifies what proper epistemic modesty consists in as follows:

However, No Certainty does not undermine the claim that we can have rationally grounded knowledge both of everyday propositions and of the denials of radical skeptical hypotheses. Even if the knowledge that I have hands that I acquire on the basis of seeing them is not certain knowledge, it is rationally grounded knowledge nonetheless. And if I know I have hands then I can deduce the anti-skeptical proposition that I am not a handless BIV, thereby acquiring rationally grounded, epistemically fallible knowledge I am not a handless BIV.