The Hard Problem of Access for Epistemological Disjunctivism

Abstract

In this paper, I identify the hard problem of access for epistemological disjunctivism (ED): given that perceptual experience E is opaque with respect to its own epistemic properties, subject S is not in a position to know epistemic proposition (i) (that E is factive with respect to empirical proposition p) just by having E and/or reflecting on E. This is the case even if (i) is true. I first motivate the hard problem of access (Section 2) and then reconstruct and analyze three of the ways in which EDists have argued for the internal accessibility of the factive character of perceptual experience. These arguments explain internal access in terms of the unity of perceptual and rational capacities (Section 3), favoring support (Section 4), and the outward-looking model of self-knowledge (Section 5). My conclusion (Section 6) is that none of these responses works. I then suggest how ED might be modified to succeed as an access internalist epistemology.

1. Introduction

In this paper, I identify the hard problem of access for epistemological disjunctivism (ED). I understand ED as follows:

ED: In the good case, subject S is in better epistemic standing with respect to empirical proposition p than in the corresponding bad case in virtue of having perceptual experience E such that:

(i) E is factive with respect to p. That is, E guarantees the truth of p by presenting S with worldly fact p ¹

and

(ii) the factive character of E is internally accessible to S. That is, S can know epistemic proposition (i) (that E is factive with respect to p) just by having E and/or reflecting on E.

In the bad case, experience does not meet condition (i) and/or condition (ii).

“Epistemological disjunctivism” is sometimes understood in a broader sense. It is understood as the view that a subject having perceptual experience in the good case has better epistemic standing with respect to the relevant proposition than a subject in the bad case simpliciter. There is no commitment to internal accessibility. This means that the view is compatible with access externalist theories of justification. However, the version of ED I am concerned with is the distinctively access internalist version. In this version, the overall epistemic standing of the subject is determined not only by the epistemic power of her experience (its factivity), but also by her ability to know that epistemic power.

The access internalist commitments of ED have been defended by McDowell (1982, 1995, 2008, 2010, 2011, 2013, 2019), Pritchard (2012, 2016), and Logue (2018). These views differ from externalist views on the epistemic role of perceptual experience (which can also be called “disjunctivist” in the broader sense (e.g., Millar 2019; Williamson 2011)).

The problem I raise in this paper is specifically related to the access internalist aspect of ED ((ii) above).

The hard problem of access for ED: Given that perceptual experience E is opaque with respect to its own epistemic properties, subject S is not in a position to know epistemic proposition (i) (that E is factive with respect to empirical proposition p) just by having E and/or reflecting on E. This is the case even if (i) is true.

I shall unpack the idea of epistemic opaqueness in Section 2. Let me first explain why this is the hard problem of access rather than Pritchard's (2012) “access problem” for ED. The access problem concerns the possibility of purely reflective knowledge of empirical proposition p. In contrast, the hard problem concerns reflective knowledge of the epistemic proposition that E is factive. Pritchard's access problem might be called the easy problem of access. It can be neutralized just by showing that ED is not committed to the claim that, in the good case, S has purely reflective access to empirical fact p. That is Pritchard's reply to the access problem (2012: 46–52). In contrast, the internal access condition (ii) (in the good case, S can know the factive character of experience by reflection alone) is among ED's commitments. This is why proponents of ED cannot easily neutralize the hard problem of access.

As stated, the hard problem of access is little more than the denial of one of ED's claims. Just raising the problem in this way would be dialectically inert against ED. Moreover, proponents of ED offer arguments for (ii) that are eo ipso answers to the problem. However, I shall devote the main body of this paper to analyzing these answers.

In the remainder of this paper, I first motivate the hard problem of access (Section 2) and then reconstruct and analyze three of the ways in which EDists have argued for the internal accessibility of the factive character of perceptual experience. These arguments explain internal access in terms of the unity of perceptual and rational capacities (Section 3), favoring support (Section 4), and the outward-looking model of self-knowledge (Section 5). My conclusion (Section 6) is that none of these responses works. I then suggest how ED might be modified to succeed as an access internalist epistemology.

2. Motivating the hard problem of access for ED

ED is a self-presentational variety of access internalism. The property of experience that provides S with access to her surroundings and the property that makes the factive nature of perceptual experience internally accessible is the same property. It is the phenomenal character of experience.² In other words, in the good case, perceptual experience presents S with her surroundings. Perceptual experience also presents S with its own factive nature. From the perspective of access internalism, perceptual experience is then epistemically self-sufficient in the sense that, in the good case, having perceptual experience that presents subject with worldly facts at the same time satisfies possession-condition (ii) on having that experience as perceptual evidence.

Like some other self-presentational variants of access internalism, ED takes the subject's self-awareness of possessed evidence to be grounded in phenomenal character. But, unlike more classic forms of access internalism, ED is far more ambitious when it comes to how much of the possessed evidence's epistemic power phenomenal character reveals. On ED, the self-presentational character of perceptual experience enables S to know, through reflection alone, not only that she has perceptual experience, or that there is some relation of correspondence or probabilistic support between what is presented in experience and empirical proposition, but that perceptual experience supports empirical proposition in the best, i.e., truth-entailing or factive way. It follows that, on ED, the epistemic value of perceptual evidence in the good case is the joint upshot of the fact that perceptual experience E is factive and the fact that having E, in the good case, puts a subject in a position to know that the experience is factive through reflection alone.

Meeting access condition (ii) turns factive perceptual experience into factive evidence that the perceiver self-consciously possesses. If so, ED can promise more than weaker forms of internalism, namely that perceptual evidence possessed by the subject in the good case is better than defeasible prima facie evidence.³

ED predicts that, in the good case, my perceptual experience of my two hands (1) guarantees the truth of the belief that I have two hands and (2) puts me in a position to know this fact about the factive character of experience. As a result, it puts me in a position to know that I have two hands. This is the “holy grail” of (perceptual) epistemology promised by ED (see Pritchard 2012: 1): a successful combination of epistemic externalism and epistemic internalism.

My claim is that ED's quest for the epistemic holy grail faces an uphill struggle. It must explain how just having a factive perceptual experience is sufficient for possessing the kind of internal access that allows S to know that the experience is factive. In a nutshell, the problem is that, even if (i) is true and phenomenal character provides excellent access to the empirical fact p, it does not provide a similar kind of access to the fact that perceptual experience is factive.

As mentioned, ED maintains that the epistemic value of perceptual experience is partly grounded in the phenomenal character presenting S with worldly facts. This presentational view is on par with phenomenological considerations concerning the transparency of perceptual experience. Phenomenal character is transparent in the sense that the most natural characterization of the phenomenal character of perceptual experience is in terms of perceived objects and their properties. Whether transparency should be explained in representationalist, relationalist, or other terms is a subject of ongoing debate (see Crane 2006; Pautz 2021). Nevertheless, transparency gives some plausibility for claim (i) that perceptual experience is factive. That is, by making worldly fact p manifest to S, perceptual experience guarantees the truth of belief that p.

I contend that there is no corresponding phenomenological consideration in favor of the self-presentationalist view explaining the internal accessibility of factiveness. No feature of perceptual experience is sufficient for access to the factive character of E required to know epistemic proposition (i). Even if phenomenal character is transparent with respect to the object of experience, it is opaque with respect to its own factivity. The epistemic opaqueness thesis is that perceptual experience does not provide access to its own epistemic features in the way that it provides access to the subject's environment. It can be supported by considering the following conversation:

Q: How do you know that you have two hands?

A: I see that I have two hands.

Q*: How do you know that your experience of two hands guarantees that you have them?

A: I see that I have two hands.

If ED is correct, then A would be as good an answer to Q* as it is to Q. But, A is a good and proper answer to Q, but not to Q* (despite the fact that Q* is a strange question to ask in most everyday contexts). The opaqueness thesis might be motivated by the inference to the best explanation of the inappropriateness of A as an answer to Q*. If perceptual experience is opaque rather than transparent with respect to its own epistemic properties (in a way that it is not when it comes to the properties of the presented scene), then it explains why simply citing one's experience is not an informative answer to the question of the grounds of self-knowledge (Q*). It is, however, informative as an answer to the question of the grounds of perceptual knowledge (Q).

A natural interpretation of A as an answer to Q* is that the interrogated person is simply ignoring Q*. Rather than providing grounds for self-knowledge, she again provides grounds for the empirical proposition she was asked for in Q. On ED, grounding perceptual knowledge does not have the form of (even internal) conversation or reasoning. Instead, it is immediate. Nevertheless, it is part of ED's distinctively access-internalist view that grounding perceptual knowledge has a rational structure, a structure that is reflected in the practice of giving and asking for reasons (see McDowell 2011: 25–27). This is why proponents of ED cannot neutralize the inappropriateness in the above conversation simply by saying that having self-knowledge of the factive character of experience is one thing while telling that one is having self-knowledge of the factive character of experience is another (see Pritchard 2012: 141–50).

In the epistemological literature, one finds explanations of the epistemic opaqueness of perceptual experience that relate to the subjective indistinguishability of good and bad cases of perceptual experience (Dennis 2014), the fallibility of perceptual capacities (Burge 2005), the fallibility of introspection (Dretske 2012), and the non-luminous character of factive mental states (Williamson 2000). I shall reference some of these explanations during my analysis. Nonetheless, I prefer to remain neutral on which explanation is the best.⁴ Instead, based on the above considerations of the epistemic opaqueness of perceptual experience, I consider there to be an explanatory gap between the psychological relation of having factive experience and the epistemic relation of having factive evidence (in the access internalist sense). EDists attempt to bridge that explanatory gap by filling in the blank in the following sentence:

In the good case, S can know epistemic proposition (i) (that E is factive with respect to empirical proposition p) just by having perceptual experience E and/or reflecting on E because _____________.

If the attempts are plausible, then the epistemic self-sufficiency of perceptual experience in the good case would be unproblematic. Quite the contrary; it would be a source of the highest epistemic value of perceptual experience. However, if the attempts are implausible, then ED will be in serious trouble. This will mean that ED is prone to the hard problem of access.

The hard problem of access is similar to a worry that Crispin Wright has raised. ED is unable to “transmit warrant” from disjunctive premises about the normative force of experience to an anti-skeptical conclusion about the actual rational standing of the perceiver (Wright 2002: 331; see also 2008). The conclusion remains as disjunctive as the premise. However, I agree with disjunctivists that Wright failed to take ED's arguments for internal access to the factive nature of experience sufficiently seriously (Neta and Pritchard 2007: 387). We must therefore carefully examine the ED's arguments to determine how the structure of perceptual capacities secures the perceivers internal access to the factive nature of experience in the good case, regardless of whether we take Wright's objection to be valid.

Many externalists have emphasized that ED struggles to combine access internalism and content externalism (Greco 2014; Kelp and Ghijsen 2016). However, they were not enough interested in focusing specifically on the internalist arguments for ED. Unfortunately, despite its name, more detailed discussions on the so-called “access problem” for ED (Pritchard 2012: 20) concern the different issue of purely reflective access to empirical facts about the environment (Fratantonio 2019; Kraft 2015; Schroeder 2016). These discussions leave aside the question of internal access to epistemic facts about perceptual experiences as factive evidences. The “indistinguishability problem” (Pritchard 2012: 21) has more to do with the issue that concerns me in this paper, namely with the consequences of the limitations of self-knowledge about the perceptual experience E for the epistemic value of E. However, debates on this topic have largely focused on the possibility of using the concept of epistemic blamelessness to explain the effects of indistinguishability on perceivers' rational standings (Boult 2017; Madison 2013; Shaw 2016; cf. Ranalli 2019 for a different take on indistinguishability problem). Meanwhile, the key issue for access internalism is distinct from the question of why epistemic blamelessness is insufficient for justification.

The problem with ED views on internal accessibility of factive character of perceptual experience has been nicely tracked in discussions on the understanding of fallibility that is entailed by ED (Burge 2011; Dennis 2014; Leddington 2018). However, arguments against ED that are discussed during that discussion are typically based on a specific understanding of fallibility that epistemological disjunctivists do not in fact share. My aim here is thus to show why ED's conception of internal access to perceptual reasons is problematic from ED's own, internalist perspective. Taking all these factors into account, I contend that the hard problem of access for ED still requires investigation.⁵

In the next three sections, I reconstruct and analyze three of the ways in which EDists have responded to the hard problem of access.

3. Anti-hybrid strategy

Anti-hybrid strategy: In the good case, S can know the epistemic proposition (i) (that E is factive with respect to empirical proposition p) just by having perceptual experience E and/or reflecting on E because E is an exercise of rational capacity entailing self-knowledge of the epistemic power of E.

In hybrid epistemologies, the best available rational support for the belief that p is compatible with the falsity of p. This is because the truth of the belief is guaranteed by a factor external to rational justification, or, in the Sellarsian jargon that McDowell adopts, “standing in the space of reasons.”⁶ Sometimes the best available evidence is not sufficient for being right, even if it is sufficient for being rational.⁷

McDowell (1995) is unsatisfied with the above line of reasoning. Factive experience that p entails that the belief that p is true. If the subject can base her belief that p on that experience, then what more do we need for her to have factive evidence for p? Indeed, when all goes well, the exercise of perceptual capacity is just the exercise of rational capacity for knowledge.

In response, we can argue that this claim follows from methodological principles assumed by anti-hybrid strategy rather than from any independent argument. According to the anti-hybrid strategy, we should seek the epistemology that explains how standing in the space of perceptual reasons itself guarantees the truth of justified beliefs. This may be true, but what we still need here is precisely such an explanation. Claiming simply that there is a unity of perceptual and rational capacities explains a little, if anything.

Proponents of the anti-hybrid strategy sometimes appeal to the specific character of rational beings' consciousness to motivate their view. The argument is that internal access to the factive character of experience is provided by the fact that the requisite self-knowledge of the factive character of experience is internal to the conscious state presenting S with worldly objects in the good case (McDowell 1982; see also de Bruijn 2022). But, this sounds like begging the question against the hard problem of access. Perceptual presentation is of worldly objects. This explains why the subject presented with mind-independent objects has access to external facts that ground knowledge of the relevant empirical proposition in perceptual experience. If subject is perceptually related to her hands by having perceptual experience of her hands, then, at least prima facie, the worldly fact that grounds her knowledge that she has two hands is accessible for her. That said, perceptual presentation is not so straightforwardly of its own epistemic properties. The claim that it is requires extensive argumentative and explanatory support.

In general, it seems as if anti-hybrid strategy is based on confusing objective and subjective reasons for belief, i.e., evidences (see Schroeder 2021 for discussion of the distinction and its implications). Being factive experience is probably sufficient to be a reason in the objective sense. Meanwhile, ED is committed to the claim that factive experiences are the best subjective reasons for belief.⁸ The mere fact that the subject is self-aware of having perceptual experience that happens to be factive does not mean that the subject is aware of its factiveness, and in consequence has that experience as the best subjective reason.

The above difficulties can be illustrated by problems of anti-hybrid strategy with predicting perceivers' epistemic standing in specific situations. Let us consider the following modification of McDowell's (2009) example:

Ornithology. You see a cardinal in your garden. You have a fallible but reliable capacity to recognize cardinals by sight. The bird is an actual cardinal. According to ED, your actual experience of the bird serves as your factive evidence for believing that the bird is a cardinal, even if the bird is not represented by the experience's content under the concept “cardinal” (McDowell 2009: 269). While you are still looking at the cardinal, you receive false information from a reliable ornithologist source that the bird is a very rare species that is indistinguishable from a cardinal. As a result, you rationally suspend your judgement and lose your knowledge, but your experience remains the same.

Now, a challenge for an anti-hybrid strategy is this: if you have a factive (“indefeasible” or “conclusive” as McDowell puts it) evidence for the belief that the bird is a cardinal merely by having a perceptual experience which brings the bird into your view, why does this internal access not rule out the defeater? Before and after being (mis)informed, you have access to the same, best evidence for your belief – the same factive perceptual experience. If internally accessible factive experience cannot rule out the different species scenario from your perspective, then what can? The subject's need for additional evidence to defeat the defeater and regain knowledge suggests that the fact that she has factive experience does not provide her with internal access to the factivity of experience.

In response, McDowell distinguishes between two kinds of ornithology-like cases:

In one kind, someone who has a belief based on an experience of perceiving could have been in circumstances in which there is good reason to believe her apparent perceptions are untrustworthy. In the other, someone who has a belief based on an experience of perceiving could have been persuaded, by misleading evidence, that her apparent perceptions are untrustworthy in the circumstances (McDowell 2013: 269).

McDowell's response to the challenge raised by the first kind of case is that it is bad case: there is simply no perception in the relevant sense, so, there is no question of an indefeasible evidence either. In response to the second kind, he argues that the subject lacks awareness that she is in possession of a conclusive evidence, and, thus, such a warrant is not available for use in justifying her corresponding belief: “But that leaves unchallenged the claim that the warrant her experience gives her – which in such a situation she would not have known she has – is conclusive” (McDowell 2013: 270).

The first problem with McDowell's strategy is that the ornithology scenario simultaneously fits both kinds of cases. On the one hand, there are good subjective reasons to believe that apparent perceptions of cardinals are untrustworthy because the ornithological source is reliable (however, the ornithological information is false in this case). On the other hand, the ornithologist's counterevidence is misleading or downright false when viewed from an objective perspective. In fact, reasons to doubt your experience are not as good as they might seem. This problem undermines McDowell's response. That said, let us assume, for the time being, that the ornithology case fits nicely into one of the kinds presented by McDowell.

If the ornithology scenario is the first case, then, under the subjective understanding of reasons, when you gain good reasons by being informed by the ornithological source, your experience changes in nature. This response seems to be an ad hoc attempt to make the entire anti-hybrid framework coherent at the expense of making implausible metaphysical claims. Only as a claim about a change in propositional seeing,⁹ namely perceptual judgment, does it have some plausibility. Trivially, if the subject withholds her perceptual judgment that p, she stops “propositionally” seeing that p. However, this is a derivative sense of “seeing.” To be empirical for perceptual belief, a veridical experience must be understood as a pre-judgmental perception. There is simply no reason to believe that such a change in the space of reasons affects the perceptual experience.

What if McDowell's response is about objective rather than subjective reasons? In the ornithology case, there were objective normative reasons to not believe your eyes even before you gained ornithological information, i.e., that a reliable ornithological source did (or even just would) recognize the bird not to be a cardinal. So, if the presence of objective reasons is what determines the nature of your experiential state, then in the ornithology scenario you would never have the perceptual experience of a cardinal. In response, a purely objective understanding of reasons is not sufficient for the access internalism entailed by ED.¹⁰ In addition, again, the very idea that the normative status of experience determines its metaphysical nature seems to be highly controversial.

Suppose the ornithology case belongs to the second kind, which I think is the more plausible scenario. In that case McDowell's response explains why having this experience is still incompatible with the falsity of the corresponding belief. At best, this explanation explains only why the experience is a conclusive objective reason. But that says nothing about how the subject gains access to this fact, such that they have it as an evidence. Moreover, McDowell's response to the second kind suggests that by being misinformed you lose access to the perceptual experience qua evidence, which then means that just having an experience is not sufficient for having it as an indefeasible reason (even if the experience is still factive).

Duncan Pritchard modified McDowell's anti-hybrid strategy to explain access conditions in terms of favoring support. Let us turn to his proposal now.

4. Favoring support

Favoring support: In the good case, S can know the epistemic proposition (i) (that the E is factive with respect to empirical proposition p) just by having perceptual experience E and/or reflecting on E because S has favoring support for the belief that her perceptual experience is factive.

Pritchard introduced “favoring support” in his response to the indistinguishability problem for ED. According to Pritchard, the indistinguishability problem arises when a perceiver cannot tell good cases of perceptual experience from bad cases just by introspecting on phenomenal character. For example, the phenomenal character of the perceptual experience of a yellow banana in the good case and the phenomenal character of the illusion of a yellow banana in the bad case are ex hypothesi the same (or at least similar enough). A subject cannot tell them apart. She cannot know whether she is in a good case or a bad case just through introspection. This is clearly an instance of epistemic opaqueness, one that seems to contradict the internal access requirement (ii) (which ED imposes on the epistemic standing of the subject in the good case). For now, I shall assume that the opaqueness of perceptual experience reduces to the above kind of phenomenal indistinguishability.

Pritchard's response is to allow that good and bad cases are subjectively indistinguishable in the sense of being indiscriminable through introspecting phenomenal character. But, he thinks that they are distinguishable on the basis of favoring support. The subject can know the factive nature of an experience by reflecting on evidence that favors the belief that she is in the good case. This is so even if she cannot discriminate between factive and non-factive experiences “just by looking” or through directly introspecting on the phenomenology of experience (Pritchard 2012: 91–100).

To see how this is supposed to work, consider Pritchard's treatment of the zebra case. In the zebra case, Zula has the factive perceptual experience of a zebra in the zoo. We ask Zula for support for her belief that there is a zebra in the zoo rather than a cleverly disguised mule. Our doubt is restricted to this specific belief about the zebra. We are not questioning the justification of all Zula's beliefs. Rather, we are questioning only the epistemic value of a specific experience supporting the relevant belief, and, eventually, belief itself (Pritchard 2012: 77–81). Pritchard calls zebra-like cases local error-possibilities.

Pritchard's proposal is that, in local error-possibilities, favoring support consists of a mental state that is independent of the questioned perceptual experience supporting the belief and belief itself (Pritchard 2012: 77–81). Thus, e.g., Zula's favoring support may consist of her background belief that it is unlikely for zookeepers to disguise mules as zebras to make them indistinguishable from her perspective. That background belief supports the belief that what she sees is in fact zebra and, eventually, favors the belief that she has factive perceptual experience of zebra. Zula, then, has internal access to the factive nature of her experience in virtue of being capable of deducing from that kind of background belief that she is in the good case. In general, internal access to factive perceptual experience in the good case is not secured by the subject's capacity to discriminate between phenomenal properties of experience. Rather, it is secured by the subject's capacity to distinguish good cases from bad cases via reflection and deduction from favoring support.¹¹

Note that, on this view, epistemic labor is divided between the factive perceptual experience (i) and an independent mental state (e.g., background beliefs that render the factive nature of experience internally accessible to the perceiver upon reflection). This is a significant modification of ED's original anti-hybrid program. The overall epistemic standing of the subject in the good case does not only depend on exercising perceptual capacities as capacities for knowledge. It also depends on possessing beliefs that are independent of exercising the perceptual capacity in question. This response resolves the hard problem of access for ED at the price of abandoning the most valuable element in epistemology's holy grail: the epistemic self-sufficiency of perceptual experience. In the good case, the subject does not only reflect on perceptual experience to meet the internal access requirement. She also reflects on independent empirical beliefs. If the epistemic value of perceptual experience as evidence depends on the epistemic status of the subject's other contingent and empirical beliefs, then this response falls short of the promised holy grail (even if it may be valuable in other ways). Background empirical beliefs also need justification, and this threatens to invoke a regress of justification.

Pritchard has fortunately put forward another take on favoring support. He does so while discussing radical skepticism. This take is more in line with the epistemic self-sufficiency of perceptual experience. According to Pritchard, radical skepticism is global (unlike local error-possibilities). It undermines all beliefs in some category rather than a specific empirical belief or the epistemic value of a given experience. In radical skeptical scenarios (e.g., brain-in-a-vat scenarios), we are, then, not allowed to use background empirical beliefs as favoring support that excludes skeptical hypothesis. If all of Zula's empirical beliefs are in question, then she cannot use her belief about the likeliness of disguised mules to support her belief that her experience is factive. This dramatically constrains the possible candidates for favoring support. Pritchard, nonetheless, thinks that the same radical skepticism constrains (if not undermines) the whole skeptical enterprise. The radical skeptic cannot invoke any specific empirical grounds in support of her own doubt. Radical skeptical doubts are not local to a given scenario and not motivated by any specific consideration. Such doubts are, instead, based on the mere possibility of error.

If one asks you to imagine that you are the victim of a mad scientist who is stimulating your brain in a vat to elicit the massive illusion of an external world, then that skeptic simply confronts you (in a specific way) with the possibility of massive error rather than actually raises and motivates any specific doubt about your epistemic standing. It follows that radical skeptical doubt is, as Pritchard puts it, “merely raised” rather than “epistemically motivated” (2012: 125–30).

What favoring support is available in the radically skeptical scenario? One answer Pritchard offers is that the subject's experience itself is her favoring support for the relevant epistemic proposition (2016: 138–40). He states that the “unmotivated” nature of radical skeptical doubts allows one to use perceptual experience in the good case as favoring support for the epistemic proposition that the relevant experience is factive.

The problem is that, if factive experience provides the subject with reflective access to its factive nature by speaking in favor of epistemic proposition (i), then the distinction between “favoring” and “discriminating” types of support collapses. The hard problem of access will then kick in. The plausibility of Pritchard's notion of favoring support lies in the difference between challenged experience and independent, unchallenged background belief. Favoring support (qua favoring support) allows the subject to independently confirm the accuracy of the content of her perceptual experience. It is an independent source of evidence for the epistemic proposition and, eventually, questioned empirical belief. If there is a question related to epistemic power of perceptual experience, then that perceptual experience will lose its evidential status until the verification is complete. In such a scenario, one cannot use evidence to verify its epistemic power against itself: if the epistemic power of experience is questioned, then the experience itself cannot be a source of favoring evidence for its own epistemic power.

Note that, even if ED's core idea is correct – even if perceptual experience in the good case lends factive support to the relevant empirical proposition (despite being indistinguishable from illusion) – this does not explain how perceptual experience can be a favoring support for the epistemic proposition concerning its own factivity. A subject having factive experience of her two hands might be in excellent epistemic standing with respect to the proposition that she has two hands, but the very fact that she has that kind of experience does not explain how she can know that the experience is factive.

Another option for Pritchard is to say that the “unmotivated” nature of radical skepticism allows the subject to simply dismiss the skeptical worry out of hand. Pritchard sometimes suggests as much (e.g., 2012: 125–30, 135). When the doubt is unmotivated, it can be ignored without giving support (favoring or otherwise) to the epistemic proposition. Here, perceptual experience maintains its evidential status all the time. Note that, in this case, the distinction between discriminating and favoring support does no epistemological work. A subject having a perceptual experience in the good case “just knows” that the experience is factive. Unmotivated doubts have no force against that kind of self-knowledge.

Pritchard has admitted that ED puts us in a position to argue that radical skeptical hypotheses (e.g., the brain-in-the-vat hypothesis) “should be rejected out of hand” (2012: 135). This appears dialectically problematic given that Pritchard also states that “rejecting the intuitions which drive the new evil genius thesis out of hand” would generate “a problem for epistemological disjunctivism, since the view could then be charged with being unable to accommodate the core guiding intuitions of epistemic internalism (the new evil genius thesis, after all, is typically endorsed by both accessibilists and mentalists)” (2012: 42).

More importantly, doubt related to indistinguishability or epistemic opaqueness cannot be so easily dismissed as “unmotivated.” Epistemic opaqueness is neither local (in the way that doubt in the zebra case is local) nor radical (in the way that doubt in the brain-in-a-vat scenario is radical). The hard problem of access is not limited to specific cases, nor does it only arise in radically skeptical scenarios. Epistemic opaqueness is not motivated by any scenario-specific doubt about a subject's environment or by defects in her cognitive or perceptual abilities. It is also not motivated by the mere possibility of being a mad scientist's victim. Instead, it is motivated by phenomenological considerations concerning every normal perceptual experience. Perceptual experience does not present the subject with its own epistemic features in the way that it presents (or seems to present) her with properties of worldly objects. This doubt cannot be rejected out of hand. Doing so simply begs the explanatory question raised by the hard problem of access.

Having said that, I think that Pritchard prematurely retreated from seeking favoring support that is different from experience itself. Even if independent empirical beliefs about the environment are unsuitable candidates for favoring support, other non-empirical beliefs can favor the relevant epistemic proposition. I return to this point in the conclusion. For now, let us consider the last version of ED, which is significantly different from the first two.

5. Outward-looking procedure

Outward-looking procedure: In the good case, S can know the epistemic proposition (i) (that the E is factive with respect to empirical proposition p) just by having perceptual experience E and/or reflecting on E because S gains knowledge of the factive nature of experience by attending to mind-independent objects constituting the phenomenal character of experience.

Heather Logue (2018) has proposed this solution. It differs significantly from the anti-hybrid strategy and favoring support. Both McDowell and Pritchard's versions of ED are compatible with the phenomenal reading of the indistinguishability of experiences (according to which good and bad cases are subjectively indistinguishable due to the sameness of phenomenal character).¹²

However, we observed a tension between the phenomenal conception of indistinguishability and the epistemological goals of McDowell's and Pritchard's versions of ED. Logue resolves this tension by claiming that good and bad cases of experience are epistemically different because they are phenomenally different.¹³ In other words, she combines ED with metaphysical disjunctivism proposed by naive realists (Brewer 2011; Campbell 2002; Fish 2009; Martin 2004). Logue calls the metaphysical disjunctivism the extended view of phenomenal character.

In Logue's words, perceptual experience in the good case puts the subject in an “infallible epistemic position” (it guarantees the truth of the empirical proposition it supports) because the very phenomenal character of the experience in the good case is constituted by the mind-independent object having properties ascribed to it by the belief based on that experience. The extended view is controversial on its own and deserves a separate discussion (a discussion of the sort Matthew Soteriou 2016 and Adam Pautz 2021 provide). Below, I restrict myself to problems related to combining the extended view with the “outward-looking” model of internal accessibility.

The idea behind the outward-looking model of internal accessibility is that the factive nature of perceptual experience is available to the subject not because she can attend it by directing her reflection inward (into the subject's mind), but rather outward, into an outer world constituting perceptual experience. Logue cites Gareth Evans to motivate her view. She states that the natural way to figure out whether “I believe that there will be a Third World War” is to direct my attention out to the world; I must assess “the evidence for and against the proposition at issue” (Logue 2018: 219; see also Evans 1982: 225–26). Evans puts forward a similar account of self-knowledge of perceptual experience (1982: 227–28). He did not endorse the extended view. That said, one could (prima facie) argue that the outward-looking model of internal accessibility sits well with the extended view: if the phenomenal character of experience is literally constituted by mind-independent objects presented in perception, then it seems trivial that attending to perceived objects equates to attending to the phenomenal character constituted by those objects.

Contrary to these first impressions, I contend that combing the extended view and the Evan's outward-looking model of self-knowledge is highly problematic. Evans' idea of an outward-looking model is plausible because of the transparency of experience with respect to properties of perceived objects. It is not due to the properties of experience itself. The transparency claim is that experience does not reveal its own properties (epistemic or otherwise). I can know that I have an experience of a yellow banana by just attending to a yellow banana in the relevant perceived scene. But knowing (by the same means) that my experience is factive or constituted by mind-independent objects would require experience to present itself as factive much the same way as it presents banana as yellow. However, Evans states explicitly that, unlike perceived mind-independent particulars (e.g., bananas), experience itself cannot be the object of experience. He writes: “What it means is that there is no informational state which stands to the internal state as the internal state stands to the state of world” (Evans 1982: 228).

Note that Evans was a representationalist and not a naive realist. But even on naive realism, factiveness is not among the properties presented in perceptual experience (in the way that properties of objects are) (see Fish 2009: 93–94; Martin 2009).¹⁴ It follows that just supplementing Evans' outward-looking model with naive realism will not yield a viable account of the self-knowledge of factive character. Even if attending to perceived objects equates to attending to the phenomenal character they constitute, we still do not have an explanation for how the subject can know that attended objects constitute phenomenal character just by attending to them. Even when factive, experience provides no grist to the mill of the outward-looking procedure for gaining self-knowledge of the factive character of experience.

We can approach this problem in greater detail by focusing on how Logue treats illusion cases (2018: 227–29). She maintains that there is a mind-independent object constituting the phenomenal character of perceptual experience in illusion cases. This is just like good cases of perceptual experience but not cases of hallucination. The phenomenal character of illusory experience is metaphysically different from the phenomenal character of perceptual experience for the following reason. In the illusory case, the subject's contribution (e.g., the subject's cognitive processes) renders phenomenal character such that it appears to be as of a different object from the one actually constituting it.¹⁵

Compare a perceptual experience of a gray banana, a perceptual experience of a yellow banana, and an illusory experience of a yellow banana. In the first two cases, phenomenal character is constituted by the gray and the yellow bananas, respectively. Phenomenal character appears to be as of the gray and the yellow bananas; it is transparent with respect to the worldly properties that constitute it. These are good cases. In the case of the illusory experience of a yellow banana, phenomenal character is constituted by a gray banana, which is real (as in the first case). But, the subject's contribution renders it in such a way that it appears to be yellow (as in the second case) (Logue 2018: 220–22). The rendered phenomenal character that results is distinct from yellowness or grayness. Logue calls it “grellowness” (2018: 228). Logue thinks that grellowness is constituted by the banana's grayness, but neither grayness constituting grellow phenomenal character nor the fact that the phenomenal character is badly rendered are accessible for the subject (Logue 2018: 221). The illusory experience is opaque with respect to these aspects of phenomenal character. This implies that the phenomenal character of experience constituted by mind-independent objects is not sufficient for the subject to know factive nature of experience. A suitable appearance is also required. In other words, phenomenal character must be transparent with respect to the properties that constitute it.

If the subject in the illusory case is not in a position to know that phenomenal character was rendered by a contribution of her own cognitive capacities, then how she can be in a position to know that phenomenal character is not rendered this way in the subjectively indiscriminable second case? Logue answers that the subject can discriminate the phenomenal character of experience in the good case from the phenomenal character in the bad case when she is in the good case, but not when she is in the bad case. Logue writes:

In general, sometimes indiscriminability is due to sameness or similarity between the entities that can't be discriminated. But it could also be down to the fact that one's mode of access to just one of the entities is severely compromised. In cases with this latter type of structure, indiscriminability is asymmetrical. The subject might not be able to tell the entity to which she has compromised access apart from the other (perhaps radically different) entity. But this certainly doesn't mean that she can't tell the entity to which she has uncompromised access apart from the other entity (especially if the latter is radically different). (2018: 226)

Applied to good and bad cases of perceptual experience, Logue's reasoning proceeds as follows. In good cases, the subject's access to worldly properties is uncompromised, while it is compromised in the bad cases. The good cases are a suitable context for applying the outward-looking procedure, while the bad cases are unsuitable. By directing her attention to the worldly objects presented in perception, a subject in the good case can know the factive nature of the experience. In the bad case, the subject is not in a position to know the factive nature of the experience, even if the experience is partly constituted by worldly objects (as in illusory scenarios). In an illusory scenario, the phenomenal character of an illusory experience (grellow phenomenal character) is indiscriminable from the phenomenal character in the good case (yellow phenomenal character). This is because the subject cannot attend to the grayness of a banana that appears to be yellow. However, in the good case, they are discriminable because the subject can then attend to yellowness (Logue 2018: 228).

We can now see the crux of the problem. My perceptual experience (with requisite recognitional capacities) allows me to discriminate between yellowness and grayness when I perceptually attend to the yellowness of the banana in the good case. But, this is only if I have some access to grayness (e.g., via actual experience or in memory). I obviously cannot discriminate between two different things if one of them is inaccessible. This is precisely how Logue explains the indiscriminability of grellow phenomenal character and yellow phenomenal character. A subject cannot acquire access to grayness (nor all aspects of phenomenal grellowness) by having the illusion of a yellow banana. Ex hypothesi, this is because both the gray banana and the yellow banana appear to be yellow in the illusory case.¹⁶ How things appear to the subject constrains what she has epistemic access to. The problem is that the same principle makes Logue's conception problematic.

Now, the question is whether Logue's conception of asymmetric indiscriminability entails that the grellow phenomenal character of the illusion appears to be yellow when the subject is in the good case (whatever provides subject with epistemic access to phenomenal character of the illusion in the good case).

If the answer is “yes,” then, contra what Logue suggests, phenomenal grellowness is symmetrically indistinguishable from phenomenal yellowness. How can perceptual experience of the yellow banana put the subject in a position to know that the banana in the illuded case is actually gray if the phenomenal character appears to be the same, i.e., yellow? Even if we agree that the phenomenal characters of these two experiences are different, the perceiver is (at least, partially) unable to access certain aspects of the phenomenal character of the illusion. Everything that is epistemically accessible does not allow for discriminating between phenomenal grayness and phenomenal yellowness (even when accessed from the perspective of the good case).

If the answer is “no” and phenomenal grellowness appears different from the perspective of the good case, then an implausible psychological prediction follows. Just veridically perceiving the yellow banana will (perhaps retroactively) change how things appear in the illusion of the actually gray yellow banana. Consider a situation where the subject undergoes a visual experience in which the object she visually tracks is temporarily replaced by a phantom lacking the properties of the tracked object. The phantom only appears to have those properties (as they are perceived when the object is present). This is how the so-called miniature effect is used to give the illusion of depth and size in cinema.¹⁷ It is both possible and psychologically plausible that the change in the object (which Logue maintains is a change in phenomenal character) will be inaccessible to the subject. From the subjective perspective, the whole experience will be uniform with respect to how the relevant aspect of the object appears.

The only time when moving between the illusory case and the good case changes how things appear to the subject is when an object in an illusory situation appears different from how it (the very same object) appears when perceived in the good case. For example, take the subject undergoing illusion constituted by a yellow banana appearing to be gray. At some point the subject starts to see gray banana accurately. There is a clear change in how things appear when the illusion fades and the yellow banana reveals its yellowness. This is, presumably, the paradigmatic way in which we learn the difference between appearance and reality. We learn it by reidentifying the same object appearing to us differently (Moll and Tomasello 2012; Peters 2022). But, this is not a case of an illusion of a gray banana appearing to be yellow and the veridical perception of a yellow banana. These two different things – gray banana and yellow banana – appear to be the same (have the same color). Even if the subject in the good case of perceptually experiencing a yellow banana can discriminate the phenomenal character of this experience from the phenomenal character of the illusory experience constituted by yellow banana that appears to be gray, she cannot do the same for the illusory experience constituted by gray banana that appears to be yellow. This very fact compromises internal access to the factive nature of perceptual experience in the good case.

My objection is based on the observation that knowledge by experiential discrimination is not only determined by the mode of access to objects (compromised versus uncompromised). It is also determined by the content of experiential appearances.¹⁸ If two objects appear identical (e.g., when access to one of the objects is compromised), then the subject cannot know the difference between them through experiential discrimination. One cannot discriminate things in a given context if they appear the same in that context. The only way to attain such knowledge is to obtain it from a reliable source other than an experience where they appear the same. I can gain knowledge of the illusory character of my experience of the yellow banana through reflecting on that experience and my veridical experience of the gray banana (which I recognize as the same banana that was involved in my illusory experience). Note that these two experiences are trivially and symmetrically discriminable. I can also learn something about the phenomenological effects of cognitive penetration and know that there is a cognitive stereotype that bananas are yellow. I can be less confident the next time I attend to a yellow-looking banana and eventually come to know the difference between perceiving a yellow banana and having the illusion of a yellow banana. Regardless, what matters here is that this kind of self-knowledge requires more than applying Logue's outward-looking method.

Logue does, in fact, suggest something similar at the end of her (2018) paper. She admits that self-knowledge of the factive character of experience presupposes not only factive perceptual experience, but also knowledge of the relevant empirical proposition. On this proposal, a subject gains knowledge of the empirical proposition just by basing it on the factive perceptual experience, without having self-knowledge of factive character of experience. Based on this empirical knowledge, the subject then acquires self-knowledge about the factive nature of phenomenal character (Logue 2018: 229). Logue puts this forward as an untraditional version of access internalism.

I believe that the view that perceptual experience alone cannot ground self-knowledge of its own factive nature is correct. That said, this proposal seems problematic for two reasons.

First, the process of gaining self-knowledge about the nature of perceptual experience based on empirical knowledge (which is, in turn, based on the perceptual experience in question) seems to face a problem that bears an uncanny resemblance to reliabilism's bootstrapping problem (Cohen 2002; Vogel 2008; see also Weisberg 2010, 2012). The worry is that the above view would allow a subject to acquire self-knowledge too easily. She can acquire it just by reflecting on her epistemic justification for perceptual knowledge.

Second, it is difficult to see how this view is a version of access internalism (traditional or not). On the current proposal, neither discriminating between good and bad cases nor knowledge of factive character plays a role in gaining knowledge of the relevant empirical proposition. Once that knowledge is acquired, it constitutes an independent ground for self-knowledge, self-knowledge acquired from the ground provided by perceptual experience. Posterior self-knowledge of factive phenomenal character is a kind of epistemic bonus. It plays no explanatory role in the basic epistemological story Logue tells about perceptual knowledge. As such, the resulting epistemological theory of perceptual knowledge turns out to be access externalist rather than access internalist. It is not an access internalist theory if the subject can gain knowledge of the empirical proposition without internal access to the grounds of that knowledge.

Note that I am not objecting to access externalism as such. As with Pritchard's treatment of the zebra case, I would rather suggest that the above-proposed solution to the hard problem of access transforms ED into a very different epistemological theory.

6. Conclusion

In this paper, I have argued that all three versions of ED face the same hard problem of access, namely, the epistemic opaqueness of experience in both the good and the bad cases precludes internal access to the factive character of veridical experience. The problem, as I have explained it, is distinct from the distinguishability, access and basis problems for ED, which has been widely discussed in the literature. None of the analyzed versions of ED can explain how the perceptual experience alone can be enough to rationally support the perceptual belief. This suggests that having factive perceptual experience is not enough to have it as a factive evidence.

The structural character of the hard problem of access suggests that it cannot be fixed just by minor modifications of the ED. What seems to be the ultimate source of the problem is the idea that just having factive perceptual experience alone is sufficient to have it as factive perceptual evidence. One way to solve the problem is just to abandon the internalist ambitions of ED and the whole talk about internal access. But, the worry is, that the anti-hybrid, internalist motivations are crucial for the whole ED's enterprise.

This leads me to the suggestion of how we might modify ED without abandoning its core internalist commitments. In addition to conscious experience itself, what a subject requires to have it as perceptual reason is at least tacit knowledge of simple theory of perception¹⁹ or folk perceptual epistemology.²⁰ Such a theory would explain, along rationalist and a priori lines (e.g., by metaphysical primacy of good over bad cases),²¹ why experience in good case offers adequate epistemic support for perceptual belief and when the subject is allowed to take themselves as having this kind of experience. Folk epistemology would provide background knowledge analogous to the favoring support described by Pritchard, albeit a very specific one, consisting not of background empirical knowledge, but rather of metaphysical knowledge on the nature of perception and normative a priori knowledge about perceptual entitlements. As a result, it cannot be so easily questioned by just questioning any particular belief based on perceptual experience.

This is merely a hypothesis that emerged from the dialectics of our discussion of the hard access problem for ED, and it undoubtedly requires a further in-depth investigation.²²