1. Epistemological Disjunctivism and the Access Problem

In his Epistemological Disjunctivism (Pritchard Reference Pritchard2012), Duncan Pritchard addresses the so-called Access Problem as representing a prima facie challenge for his view. Before describing the Access Problem in detail, it is worth mentioning that Pritchard takes this problem to mirror the more notorious McKinsey paradox allegedly showing the incompatibility between Content Externalism and Privileged Access. Take Privileged Access to be, roughly put, the thesis that one can know by reflection the content of one's mental states.Footnote ⁴ Take Content Externalism to be, roughly put, the thesis that the content of one's mental states is determined by factors in the external environment. Imagine now that one has a “water thought”, e.g., the thought that the water is wet. Privileged Access entails that one can know by reflection that one has a water thought. Content Externalism entails that if one has a water thought then one has interacted with H₂O. Assuming one can know by reflection that Content Externalism is true, and assuming a closure principle for reflective knowledge, the McKinsey paradox aims to show that one can know by reflection a contingent fact about the environment, e.g., one can know by reflection that one has interacted with H₂O, yet this is absurd.Footnote ⁵, Footnote ⁶

By requiring the rational support to be both factive and reflectively accessible, and assuming the plausible thesis that it is possible to know by reflection alone that seeing that p entails p, Epistemological Disjunctivism (ED) seems to immediately face a problem analogous to the one faced by Content Externalism.Footnote ⁷ Pritchard's formulation of this so-called Access Problem goes as follows:

The first premise AP1 is a direct consequence of Pritchard's ED. In particular, what matters for the foregoing argument to go through is the thesis, entailed by ED, that a subject S can know by reflection alone that R, where S's rational support R is S's seeing that. The second premise AP2 is a thesis about S's knowledge of the truth-entailing nature of R. This second premise is usually taken to be plausible, for it merely assumes that S can know by reflection alone that seeing that p entails p. However, the conclusion, which is the result of S making a competent deduction, looks (allegedly) unacceptable. Before evaluating Pritchard's response to the Access Problem, let me highlight the crucial role that a Closure Principle for Reflective Knowledge plays in the foregoing argument. I take the Closure Principle for Reflective Knowledge to be the following CRK:

The foregoing formulation of closure is what underpins the argument driving the Access Problem. That is, the argument's validity crucially depends on the truth of CRK. We can thus briefly summarise the Access Problem as follows: given that it is possible to know by reflection alone that R (AP1), and that it is possible to know by reflection alone that R entails p (AP2) then, if CRK is true, it follows the disastrous conclusion that it is possible to know by reflection alone that p. If Pritchard wants to save ED he thus needs to block the argument from AP1 to APC.

2. Pritchard's response to the Access Problem

As described above, what distinguishes Epistemological Disjunctivism is the conjunction of the following two constraints:

1. (a.) The possibility of having purely reflective access to one's rational support R.

2. (b.) The truth-entailing nature of one's rational support R.

These two constraints underpin AP1 and AP2 respectively. It is thus the conjunction of (a.) and (b.) that, together with CRK, gives rise to the Access Problem. Fortunately, Pritchard seems to have a solution to this problem. That is, he argues that:

Insofar as Pritchard is not willing to reject either (a.) or (b.), his line of response can be defined as compatibilist. Footnote ¹¹ That is, he maintains that the accessibility thesis (a.) and factivity of one's rational support (b.) are jointly compatible. Instead, given the argument from AP1 to APC, he rejects the conclusion APC: given the possibility of reflective knowledge of one's rational support, and given factivity of one's rational support (as entailed by ED), it does not follow that one can have reflective knowledge of the empirical proposition p, thereby leaving ED untouched. But what is Pritchard's motivation for rejecting the entailment from AP1 and AP2 to APC?

To see why, according to Pritchard, the conclusion APC does not follow from the premises, Pritchard invites us to unpack premise AP1 and to reformulate the Access Problem as follows:

According to Pritchard, once we specify that one's rational support is the empirical reason that one sees that p, we can appreciate that, although one can have purely reflective knowledge of one's rational support R, the conclusion APC’ doesn't seem to follow, for it looks like S does not derive that p by reflection alone.

Perhaps one can reformulate the argument in a way that resist Pritchard's response here. In particular, one could appeal to a feature of Pritchard's own view, namely, the fact that seeing that p is a way of knowing that p only in paradigmatically good cases. This means that, according to Pritchard, one can see that p without knowing or even believing that p (e.g., in cases in which one has a misleading defeater) (Pritchard Reference Pritchard2012: 25–34). The proponent of the Access Problem could thus reformulate the first premise of the argument, such that S has purely reflective access to her empirical rational support, but S does not believe that p:

Reformulating the argument with AP1’’ is supposed to prevent Pritchard from claiming that, given S already knows empirically that p in AP1’’, the conclusion does not follow from the premises. After all, it is explicit in AP1’’ that S doesn't even believe that p! Pritchard anticipates this objection but he does not think it represents a threat to ED. For note that, on his view, it is only in paradigmatic cases of perceptual knowledge that one can have purely reflective access to one's factive rational support. According to Pritchard, one's seeing that p is not reflectively accessible in cases that are epistemologically sub-optimal, such as cases in which one is in epistemically favourable circumstances but one fails to believe (and thus know) that p (e.g., because in possession of misleading defeaters) (Pritchard Reference Pritchard2012: 51). If ED is true, the first premise AP1’’ is just plainly false.

One might wonder why we should not grant some sort of reflective access in sub-optimal cases as well. However, I will not be concerned with this issue here. For the sake of the paper, I will assume, with Pritchard, that we can have purely reflective knowledge of our rational support in paradigmatic cases of reflective knowledge only.Footnote ¹³ Once we accept this restriction, it seems Pritchard can say that the conclusion of the Access Problem does not follow from the premises, and this is because

This, in a nutshell, explains why S's knowledge in APC fails to constitute a genuine instance of ‘exclusively reflective knowledge’. More precisely, what Pritchard is claiming in the foregoing quotation is that, in order for one's instance of propositional knowledge to be in the market for ‘purely reflective knowledge’, it cannot be essentially dependent on one's empirical reasons. But here's the problem: it's far from clear what it means for one's knowledge (and for one's propositional attitude more in general) to essentially depend on one's empirical reasons, and, unfortunately, Pritchard does not explicitly spell out which notion of dependency is in play here. Crucially, in order to assess whether Pritchard's response to the Access Problem is as effective as he hopes it to be, we need to have a better grasp of what it means to have purely reflective access. That is, we need an account of what it takes for an instance of knowledge to be essentially dependent on any empirical reasons one might have. More precisely, for Pritchard's response to be successful, we need a notion of dependency that meets the following two desiderata:

1. (i) it predicts that one can have purely reflective knowledge of one's empirical rational support (R);

2. (ii) it predicts that one cannot have purely reflective knowledge of what the rational support entails (P).

In what follows, I will consider three accounts of ‘essential dependency’, each of which underpins three different ways of understanding “purely reflective access”: a metaphysical understanding, a folk understanding (Section 3), and an epistemic understanding (Section 4). While I will consider and develop all three possible readings, I will devote most of my attention to the epistemic understanding because, as I will point out later, I take this to be the most charitable interpretation of Pritchard's words. However, I will argue that none of these interpretations is able to satisfyingly meet both the above-mentioned desiderata.

Before analysing these different accounts of essential dependency, I want to make two important methodological remarks. First, as mentioned before, part of the aim of this paper is to offer different interpretations of what it means for S to have “purely reflective knowledge”. Therefore, I will take the notion of “purely reflective knowledge” to work as a placeholder standing for something that can be cashed out in different ways.Footnote ¹⁴ Second, in this paper I am going to assume that, if purely reflective knowledge is indeed sometimes possible, then anything that is deduced from this original instance of reflective knowledge (and from premises that are known purely reflectively) will remain reflective knowledge. After all, as I have pointed out before, reflective knowledge is closed under competent deduction. This amounts to saying that deduction does not add any empirical element into one's knowledge. This is highly uncontroversial. However, it means that for any of the interpretations of “purely reflective knowledge” I offer, “purely reflective knowledge” will be preserved by competent deductive inference. A straightforward and important consequence of assuming that deductive inference does not introduce any empirical element to one's original reflective knowledge is that we cannot legitimately accommodate (i) and (ii) by appealing to two different interpretations of “purely reflective knowledge”. For the only way in which one can do so would be by saying that there is at least one interpretation of “purely reflective knowledge” such that deduction-based knowledge (and deductive inference in general) does not classify as purely reflective knowledge (or purely reflective inference) under that interpretation.

3. On the metaphysical and folk interpretation of essential dependency

As mentioned in the previous section, Pritchard's way of dealing with the Access Problem argument is to reject its conclusion. Although we can have reflective knowledge of our rational support R (i.e., our seeing that), we do not have reflective knowledge of what R entails (i.e., p). This is because, Pritchard says, our knowledge of p essentially depends on our having empirical reason R in the first place, and reflective knowledge is knowledge that does not essentially depend on empirical reasons. But what does it mean for a doxastic attitude to essentially depend on empirical reasons? The first interpretation I consider here is what I call the metaphysical notion of dependency:

Assume Metaphysical Dependency is the correct way of understanding the notion of ‘essential dependency’. We can then say that S's knowledge that p is reflective knowledge only if, S's knowledge that p does not metaphysically depend on S's empirical support, e.g., S's seeing that. That is, S's knowledge that p is reflective knowledge only if, it is not the case that: if S hadn't had the “seeing” in the first place, S wouldn't believe that p.

Can we rely on this metaphysical understanding of the notion of dependency in order to satisfyingly resist the Access Problem? If S knows that p in virtue of seeing that p, then it's plausible to say that S's knowledge that p metaphysically depends (in the sense expressed by Metaphysical Dependency) on S's empirical support. For remember that Pritchard is concerned with paradigmatic cases of perceptual knowledge.Footnote ¹⁵ Therefore, if S knows that p by seeing that p, then S wouldn't have believed that p if S hadn't seen that p. Thus, it looks like Metaphysical Dependency rightly predicts that one does not have reflective knowledge of the empirical proposition p. However, if Metaphysical Dependency is the notion of essential dependency that we should consider, then, how can we account for the fact that one has purely reflective knowledge of one's rational support in paradigmatic good cases? If what Pritchard has in mind is a notion of “purely reflective knowledge” that should be understood in the light of this metaphysical interpretation of the notion of essential dependency, then it looks like we can't allow for one to have reflective knowledge of one's rational support either. In fact, one's knowledge of one's seeing that p clearly metaphysically depends (in the sense expressed in Metaphysical Dependency) on S's empirical seeing in the first place. This first metaphysical interpretation of essential dependency is thus not a promising one, for it fails to meet one of the desiderata that a satisfying interpretation of essential dependency should have: although it successfully denies reflective knowledge of the empirical proposition one's rational support entails, it fails to allow for reflective knowledge of one's rational support. Furthermore, note that this would be a particularly disastrous conclusion for Pritchard. For remember that on his definition of Epistemological Disjunctivism, those paradigmatic cases of perceptual knowledge are cases of perceptual knowledge in virtue of the fact that one has reflectively accessible (factive) rational support. Embracing Epistemological Disjunctivism together with a metaphysical account of reflective access would thus entail the sceptical conclusion that paradigmatic cases of perceptual knowledge are not possible. We can thus safely say that this metaphysical interpretation of “essential dependency” surely cannot be the notion Pritchard has in mind when talking about “purely reflective knowledge”.

Let's now move onto the second interpretation of essential dependency, namely, what I call the folk interpretation. Remember, once again, that, on Pritchard's view, one's knowledge can be classified as purely reflective knowledge iff it does not essentially depend on one's empirical reasons. However, a defender of Epistemological Disjunctivism might claim that we should weaken the foregoing notions of “essential dependency” and “reflective access”. That is, one might argue that we should understand “reflective access” as referring to a very cheap and undemanding notion, one that is compatible with the layman's everyday usage of “knowing something by reflection alone and without further empirical enquiry”.Footnote ¹⁶ The idea here is that, even if one had to see that p before coming to know that one sees that p, one can acquire this knowledge merely by ‘reflecting’ on one's situation and without having to carry any further empirical check. Contrary to the metaphysical interpretation of reflective knowledge, appealing to this folk interpretation allows for purely reflective knowledge of one's rational support. At the same time, however, it entails that one can have purely reflective knowledge of the target empirical proposition entailed by one's rational support. After all, one can just “reflect” on what seeing that p entails without having to conduct an empirical check. While this folk interpretation meets the first desideratum, it does not seem to accommodate the second one. This seems problematic for Pritchard, given that he takes one advantage of his view to be that it is a non-revisionary thesis, one that leaves our pre-theoretical intuitions untouched. Accepting the conclusion APC would thus fail to account for the general intuition that it is in fact absurd to have “reflective knowledge” of specific empirical propositions.

Perhaps Pritchard could thus bite the bullet here, thereby granting the possibility of “reflective knowledge” of a specific empirical proposition p. At the same time, he could say that, once we are not taking “reflective knowledge” to be any technical notion, this would not be a disastrous conclusion after all. In fact, Pritchard could rightly point out that, given this weakened notion of essential dependency, accepting that one can have “purely reflective knowledge” that p does not even constitute a real biting the bullet strategy. For note that the Access Problem does not concern the possibility of extrapolating by reflection the consequences of what one already knows empirically, as there is nothing problematic about that. Rather, it concerns the possibility of having purely reflective knowledge of an empirical proposition. And yet, on this folk interpretation, one's “purely reflective knowledge” that p in APC would be “purely reflective knowledge” only in name.Footnote ¹⁷

Despite its prima facie plausibility, I am afraid this strategy is also problematic, for it clashes with Pritchard's overall project of offering Epistemological Disjunctivism as the holy grail of epistemology, namely, as a view that combines the core commitments of both traditional internalism and externalism, thereby shedding light on a new logical possibility. First, note that if we say that, in APC, one has purely reflective knowledge that p in the “folk sense” (i.e., in the sense of having “knowledge without further empirical enquiry”), we will have to insist that one's rational support in AP1 is also reflectively accessible in the same “folk sense”. Crucially, a weak interpretation of “reflective knowledge”, such as the one offered by the folk interpretation, would fail to capture the internalist conception of “access” that, by contrast, has been traditionally understood in relation to the technical notions of cognitive accessibility and doxastic responsibility. More precisely, what lies behind Pritchard's accessibility requirement seems to be an internalist conception of rationality, one on which one's rational support has to be unproblematically and immediately given to one.Footnote ¹⁸ Furthermore, embracing a more conservative account of “purely reflective knowledge” in AP1 (in a way that is in line with access internalism), while saying that we should read “purely reflective knowledge” in APC along this more liberal folk interpretation (in a way that makes the conclusion APC unproblematic) is also not an option available to the defender of Epistemological Disjunctivism. For, as I pointed out in the previous Section, we cannot accommodate desirata (i) and (ii) by appealing to two different notions of “purely reflective access”.Footnote ¹⁹

Finally, if the accessibility requirement underpinning Pritchard's Epistemological Disjunctivism has to be understood along this folk interpretation, then the worry is that Pritchard fails to distinguish his view from more radical and traditional externalist theories of rational support.Footnote ²⁰ That is, if all Epistemological Dijsunctivism entails is that, after seeing that p, we can know our rational support (seeing that) by reflecting on our epistemic situation and without conducting any further empirical inquiry, then this seems to be something that almost everyone would be willing to accept. After all, everyone – externalist included – would grant that in paradigmatic cases of perceptual knowledge we can have knowledge of what our rational support is. For instance, Williamson, who notoriously rejects access internalism, would still acknowledge that in favourable cases we are in the position to know what our evidence is (Williamson Reference Williamson2000: 15). Similarly, as John Greco has pointed out, if this is the conception of “purely reflective knowledge” Pritchard has in mind, then, even paradigmatic externalist theories, such as reliabilism, would grant that in paradigmatic cases of perceptual knowledge we know “purely by reflection” that our (reliable) rational support is our seeing that something is the case (Greco Reference Greco2014: 120).

4. On the epistemic interpretation of essential dependency

In the previous section, I have considered two interpretations of the notion of essential dependency, namely, a metaphysical interpretation and a folk interpretation. I have shown that none of them is actually satisfying. The former is too restrictive and it predicts that one can have reflective knowledge of neither one's rational support R nor what R entails. The latter seems too liberal, for it predicts that one can have reflective knowledge of both one's rational support R as well as the empirical proposition P one's rational support R entails. In the remainder of this paper, I will focus on what I call the epistemic interpretation of the notion of essential dependency. I take this to be the most charitable interpretation of Pritchard's words. Crucially, I will argue that, despite being more promising than the metaphysical and folk interpretations, the epistemic interpretation is nonetheless highly problematic. I will thus conclude that, as things stand, there's no plausible account of essential dependency (and thus of Reflective Access) that would enable the defender of Epistemological Disjunctivism to resist the Access Problem.

I believe that Pritchard's notion of dependency is to be understood in the light of what constitutes his main target of interest, namely, the nature of perceptual knowledge and, most of all, the nature of a (rational) epistemic basis, together with his interest in doxastic justification.Footnote ²¹ Bearing this in mind, I will here understand the notion of epistemic dependency as the following conditional:

If we take Epistemic Dependency to be the notion of dependency Pritchard has in mind, then the fact that S's knowledge that p essentially depends on S's having the empirical reason that she sees that p is to be understood as the fact that the (rational) reason why S is believing that p (and, granted that S is in a paradigmatic good case, S knows that p) is that she sees that p.Footnote ²⁴ In other words, S's knowledge that p epistemically depends on S's seeing that p, insofar as S is basing her belief that p on the fact that she sees that p. We can recapitulate things as follows: Pritchard has to resist the argument from AP1 to APC in order to save his ED. Given he is committed to both premises AP1 and AP2, he rejects the conclusion APC. What is Pritchard's motivation for rejecting APC? The answer seems to be that, while S's knowledge that R is reflective, S's knowledge that p is not reflective after all, rather it is an instance of empirical knowledge, given that S's reason for believing that p is primarily S's seeing that p. That is, S's knowledge that p is mainly based on S's seeing that p.

5. An alternative diagnosis: Tim Kraft's source/content distinction

As I mentioned in the introduction of this paper, not much attention has been devoted to the Access Problem for Pritchard's Epistemological Disjunctivism. In this final section, I consider one of the few papers explicitly addressing this problem, namely, Tim Kraft's recent paper (Kraft Reference Kraft2015), which provides an alternative diagnosis of where Pritchard's response to the Access Problem goes wrong. I argue that the source/content distinction underpinning his diagnosis leads to undesirable results.

Kraft's starting point in critically evaluating Pritchard's response is the same as mine. He quotes the relevant passage where Pritchard explicitly states that, contrary to first appearances, the conclusion of the argument fails to follow from the premises (Kraft Reference Kraft2015: 318 [quoting Pritchard]). He then acknowledges that, by rejecting the entailment from the premises to the conclusion, Pritchard is rejecting a plausible Closure Principle for Reflective Knowledge. In fact, he takes rejection of closure to be the only palatable option available to the disjunctivist. He writes:

However, Kraft points out that Pritchard does not provide an explanation as to why closure for reflective knowledge fails in the Access Problem argument. In his paper, Kraft attempts to fill in this gap. He argues that an explanation can be given by appealing to a distinction between the source and the content of the empirical support. Roughly put, the idea is that there are two ways in which our epistemic support can be empirical:

The advantage of appealing to the source/content distinction is, according to Kraft, twofold. First, the source/content distinction constitutes the conceptual tool the disjunctivist needs in order to account for the possibility of having reflective knowledge of an empirical support. Second, and more importantly, the source/content distinction would offer an explanation of why reflective knowledge fails to be closed under competent deduction. In order to see how it can do so, let us recall the original Access Problem argument:

With respect to the first point, Kraft claims that, once the source/content distinction is available, we can classify S's knowledge that R in the premise AP1 as reflective knowledge because, despite the fact that R is empirical in the content sense, R is nonetheless reflective in the source sense. Hence, what Kraft seems to be suggesting is that the nature of the belief's source is what determines whether the target instance of knowledge should be classified as reflective rather than empirical.

But how is the source/content distinction supposed to explain why reflective knowledge is not closed under competent deduction? Here's what Kraft says with respect to this second point:

As I have just explained, Kraft takes S's knowledge in AP1 to be reflective because R in AP1 is reflective in the source sense, despite being empirical in the content sense. Kraft does not explicitly say anything about S's knowledge in AP2, but in the light of what we have just said, we can plausibly assume he would classify it as an instance of reflective knowledge as well. In fact, the rational support R in AP2 is empirical in the content sense, but reflective in the source sense. Finally – and here is the crucial bit – Kraft takes S's knowledge in APC to be empirical knowledge because, as he says, “reflective access is lost if the first belief is reflective in the source sense but not in the content sense”. Why is this so? Here's Kraft's motivation behind this:

Remember that the source of our knowledge that R is what defines whether this target knowledge is empirical or reflective. To say that “my belief that P is empirical in the source sense” is thus to say, in Kraft's jargon, that my belief (and hence my knowledge) that P is an instance of empirical knowledge. This is supposed to explain why APC fails to follow from AP1 and AP2. Nevertheless, in the second half of his paper, Kraft argues that the Access Problem for Pritchard's disjunctivism cannot be solved anyway. This is because – Kraft says – the source/content distinction can be employed only if a so-called “independent requirement” is met, but, as it turns out, Pritchard's disjunctivism does not meet such a requirement. Here I shall not discuss whether Pritchard's disjunctivism does or does not meet this “independent requirement”. What I will focus on is the plausibility of Kraft's source/content distinction in the first place. In particular, in what follows, I make four brief remarks showing that Kraft's source/content distinction is highly controversial, and that the way this distinction is exploited fails to capture the real reason why the Access Problem originally arises for Pritchard's Epistemological Disjunctivism.

To begin with, note that, while Kraft takes the rejection of closure for reflective knowledge to be the option chosen by Pritchard, this is far from being the case. In fact, as I have stressed in the previous sections, Pritchard never mentions denial of closure as the key to solving the Access Problem. Now, Kraft takes rejection of closure as the only plausible option available to the disjunctivist. Whether this is true or not, it certainly is not something that we should light-heartedly welcome. Surely, it is something Pritchard should not happily welcome. Instead, as I've shown in the first half of this paper, it is the undesirable consequence Pritchard has to pay if he wants to resist the Access Problem for his Epistemological Disjunctivism.

Second, and more importantly, Kraft's source/content distinction, and the way he exploits it in order to explain why APC fails to follow from AP1, is unconvincing. On the one hand, Kraft seems to take the source criterion to be what determines the nature of one's knowledge. In fact, remember that it is exactly by appealing to S's knowledge being reflective in the source sense that Kraft explains out the possibility of having reflective knowledge in AP1. On the other hand, remember that Kraft takes S's knowledge that p in APC to be not reflective, because S's belief in APC is deduced from her belief that R in AP1, which, despite being reflective in the source sense, is empirical in the content sense. As Kraft puts it: “Reflective access is lost if the first belief is reflective in the source sense but not in the content sense”. However, there is no reason why the content of S's knowledge in AP1 should have a role in determining the nature of S's belief in APC, namely, of the belief S infers from S's original instance of knowledge in AP1, together with S's reflective knowledge in AP2. That is, there is no reason why the content of S's belief in AP1 should have a role in determining the source of S's belief in APC. Furthermore, it seems arbitrary to state, as Kraft seems to be doing, that the content of S's belief in AP1 is what determines the source of S's belief in APC, while leaving the source of the belief in AP1 untouched. I take this to be a very puzzling aspect of the way Kraft cashes out the source/content distinction.

Third, and related, if we follow Kraft's suggestion to reject closure, while motivating closure failure by appealing to his source/content distinction, we have a very controversial scenario in which one starts off with reflective knowledge (as in AP1), and ends up with empirical knowledge (as in APC), without doing anything but bringing about a competent deduction. I believe this upshot is even more absurd than the one predicted by the original Access Problem.

Fourth, the source/content distinction, as Kraft uses it, has the undesirable result of classifying paradigmatic instances of reflective knowledge as cases of empirical knowledge. We have seen that, according to Kraft, one can have reflective knowledge of R, insofar as one's knowledge is reflective in the source sense. However, if one's knowledge of R is empirical in the content sense, then any other knowledge one has after the first belief will lose its reflective status. Or at least, this is what Kraft maintains. Let's assume for a moment that he's right in saying that S's knowledge that R in AP1 can be classified as reflective, while S's knowledge that P in APC should be classified as empirical. From S's knowledge that P (in APC), S can easily infer, and thereby know, that (P v ¬P). Crucially, if Kraft's diagnosis is correct, then we are forced to classify S's knowledge that (P v ¬P) as empirical. But this is false: everyone would agree in taking S's knowledge that (P v ¬P) to be non-empirical. However, here Kraft might have an explanation accounting for the puzzling phenomenon I have just described, namely the fact that:

If this were so, perhaps we should just accept this way of exploiting the source/content distinction – pace our pre-theoretical intuitions.Footnote ³² Note, however, that the way he sets up the problem is mistaken in the first place. Kraft is assuming that S's belief that P in the conclusion APC is based on S's seeing that P. Crucially, this is exactly what is under discussion in the Access Problem. As I mentioned in the previous sections, the original problem is one in which S comes to know, by deductive reasoning alone, the conceptual implications of S's previous reflective knowledge. For the same reason, to say that it does not matter how I can know that I see that P (Kraft Reference Kraft2015: 319), is to simply dismiss the Access Problem as it originally arises for Pritchard's view. Once we take seriously the challenge posed by the Access Problem, we are then back to considering a case in which S comes to believe and know that P in APC by deductive reasoning.

In the light of the above considerations, I take my diagnosis of where Pritchard's argument goes wrong to be preferred over Kraft's one. Besides avoiding the unacceptable results that Kraft's source/content distinction faces, my diagnosis better captures what is at the bottom of the Access Problem. That is, the diagnosis I've offered better highlights that what generates the Access Problem can eventually be traced back to a weakness of Pritchard's Epistemological Disjunctivism: a lack of a clear and plausible account of what he calls “reflective access”. Furthermore, while Kraft's diagnosis appeals to a controversial distinction, one that Pritchard does not explicitly embrace, the diagnosis I have provided relies on (and only on) claims that can be derived from Pritchard's own version of Epistemological Disjunctivism.