Epistemological Disjunctivism and the Conditionality Problem for Externalism

Abstract

Epistemological disjunctivism (ED) has been thought to solve the conditionality problem for epistemic externalism. This problem arises from externalists’ characterization of our epistemic standings as conditional on the obtaining of worldly facts which we lack any reflective access to. ED is meant to avoid the conditionality problem by explicating subjects’ perceptual knowledge in paradigmatic cases of perceptual knowledge via their possession of perceptual reasons that are both factive and reflectively accessible. I argue that ED's account of reflectively accessible factive perceptual reasons is also prey to the conditionality problem. After that, I submit that Stroud's influential formulation of the conditionality problem relies on a methodological requirement that is potentially dissociable from ED. So, it is possible to put ED in the service of a more modest anti-skeptical strategy. I close by suggesting that this modest anti-skeptical strategy successfully undercuts the underdetermination-based skeptical paradox. Nevertheless, it does not have substantial advantages over some forms of epistemic externalism.

1. Introduction

While Hannah is strolling through the market, she sees some asparagus exhibited in a stand and forms the belief that there are asparagus in the stand. The asparagus are in plain view, the illumination conditions are normal, Hannah's visual system is in good order, and she lacks any reasons for doubt. So, Hannah knows that there are asparagus in the stand.

A central philosophical task is to elucidate the contribution of perception to Hannah's perceptual knowledge. Epistemological disjunctivism (ED) offers an influential proposal:

ED: epistemological disjunctivism. In paradigmatic cases of perceptual knowledge, a subject, S, has perceptual knowledge that p in virtue of being in possession of reasons for her belief that p which are both factive and reflectively accessible to S (Pritchard 2012: 13).¹

A familiar view holds that, if a subject, S, has reflective access to p, then S reflectively knows p. There has been little work on the nature of reflective knowledge of factive perceptual reasons (but see Greco 2014; Fratantonio Forthcoming; Giananti Forthcoming). Nevertheless, the following paraphrase will work for our current purposes: ‘reflection’ characterizes the ability that we deploy when we search our memories to answer some questions (e.g., ‘Who was the first president of the United States?’), or when we make the contents of our experiences explicit (e.g., ‘I have an experience as of a green apple’). Reflective access denotes knowledge gained by deploying that ability (Cunningham 2016). In our original example, Hannah sees that there are asparagus in the stand, her reason is factive because it entails that there are asparagus in the stand, and Hannah can know that she has this factive reason at her disposal by deploying the ability mentioned in the previous examples.

It has been argued that ED can improve upon internalist and externalist accounts of perceptual knowledge (McDowell 1995, 2002, 2008, 2010, 2011, 2013, 2019; Rödl 2007; Pritchard 2012, 2016; Kern 2017).² This article focuses on ED's attempt at improving upon externalist treatments of radical skepticism.³ Externalist treatments of skepticism have been thought to face the ‘conditionality problem’. The conditionality problem arises when a theory of knowledge makes our epistemic standings conditional on the obtaining of worldly facts which we lack any reflective access to (Stroud 1989, 1994, 2004; Fumerton 1990, 1995, 2006; Pritchard 2005, 2012; Ahlstrom-Vij 2011; Fernández-Vargas 2020; Echeverri Ms a). An alleged advantage of ED over externalism is that it offers a way out of the conditionality problem.

I will argue that ED's account of reflectively accessible factive perceptual reasons is also prey to the conditionality problem. After that, I will submit that the conditionality problem is parasitic on a methodological requirement that is potentially dissociable from ED. So, it is possible to put ED in the service of a more modest anti-skeptical strategy. I will close by showing that this modest anti-skeptical strategy successfully undercuts the underdetermination-based skeptical paradox. Nevertheless, it does not have substantial advantages over some forms of externalism.⁴

The article has six sections. I start with a reconstruction of two alleged advantages of ED over externalist theories of perceptual knowledge (section 2). I then use that reconstruction to present the conditionality problem (section 3) and explain how it arises for ED (section 4). Next, I respond to some pressing objections (section 5). Subsequently, I identify a methodological requirement that generates the conditionality problem and suggest that it is potentially dissociable from ED (section 6). Finally, I use the results of the discussion to sketch a modest solution to the underdetermination-based skeptical paradox (section 7).

2. Epistemic responsibility and radical skepticism

Duncan Pritchard has advertised ED as having two advantages over externalism. The first one concerns epistemic responsibility:

From epistemic internalism [ED] takes the idea that knowledge demands epistemic support that is reflectively available to the subject. In doing so it is in a position to capture, in line with standard forms of epistemic internalism, the role that epistemic responsibility plays in our acquisition of (perceptual) knowledge. (Pritchard 2012: 3)

The second putative advantage of ED over externalism concerns skepticism. For Pritchard, epistemic externalism ‘side-steps’ the skeptical problem instead of facing it ‘head on’ (Pritchard 2012: 120, 134, 137):

[O]pting for epistemic externalism as a route out of the skeptical predicament … appears to side-step the issue altogether by making our epistemic standings conditional on the obtaining of worldly facts which one lacks any reflective access to and which are in any case in dispute in a debate with the skeptic. In contrast, if epistemological disjunctivism were a viable theory then a potential route out of this problem becomes available, since we can now appeal to reflectively accessible elements of our epistemic standings which entail facts about the world. Thus, the problem is not side-stepped by appealing to facts beyond our reflective ken. (Pritchard 2012: 4)

The concept of epistemic responsibility that features in the first passage is key to ED's attempt at facing the skeptical challenge ‘head on’. We will need to elucidate this link before we can understand the conditionality problem mentioned in the second passage.

Let us introduce the concept of epistemic responsibility with a familiar, internalist conception of knowledge. We will generalize this account in due time.

internalist knowledge. If a subject, S, has internalist knowledge of p, then S is in a position to rationally claim knowledge of p in response to challenges.

Knowledge claims can be explicit, as in ‘I know p’. But they can also be implicit, as when a subject asserts ‘p’, presenting herself as someone who knows p. We can use this internalist conception of knowledge to formulate a necessary condition for epistemic responsibility:

epistemic responsibility. If S's knowledge of p is epistemically responsible, then S is in a position to rationally claim ‘(I know) p’.

Internalist knowledge requires epistemic responsibility. Under what conditions is it rational for a subject to make a knowledge claim? Let us abstract from pragmatic considerations and focus on the epistemic conditions to rationally make a knowledge claim. If we bracket knowledge of obvious propositions like [I am thinking] and [A=A], two epistemic conditions seem necessary: If it is rational for a subject, S, to claim knowledge of p, then:

1. (1) S has reasons that favor p over some salient, incompatible alternative not-p, and

2. (2) S could potentially cite those reasons to back up her knowledge claim.

Conditions 1–2 partly explain why epistemic responsibility has led some philosophers to defend a form of internalism known as ‘accessibilism’. accessibilism is at the heart of ED as well:

accessibilism. S's internalist epistemic support for believing that p is constituted solely by facts that S can know by reflection alone (Pritchard 2012: 36).⁵

Suppose that a subject, S, is in a paradigmatic case of perceptual knowledge and perceptually knows p. If S cannot know her perceptual reasons by reflection, it is hard to see how S could potentially cite those reasons to back up her claim to know p. Suppose now that our subject can know her perceptual reasons by reflection, but her reasons do not favor p over some incompatible alternative q that is salient in the relevant context. Once again, it is hard to see how S could potentially cite those reasons to back up her claim to know p. The situation seems to change dramatically with ED. Given ED, S knows p in virtue of seeing that p. Seeing that p is factive, so it favors p over any incompatible proposition. This factive reason is reflectively accessible to S in paradigmatic cases of perceptual knowledge. So, S could potentially cite this reason to back up her claim to know p.

Many internalists think of epistemic responsibility and accessibilism as necessary for knowledge. Nevertheless, weaker views are available. A subject who can take epistemic responsibility for her knowledge of p is surely in a better epistemic position than another subject who knows p but who cannot take epistemic responsibility for her knowledge of p. As an illustration, think of the proverbial chicken sexers, a group of subjects who reliably form beliefs about the sex of chickens but who also lack reflectively accessible reasons in favor of their beliefs (Foley 1987; Brandom 1998). Compare these subjects with some ‘enlightened chicken sexers’ who have reflectively accessible reasons in favor of their beliefs. One can recognize that the enlightened chicken sexers are in a better epistemic position than the proverbial chicken sexers without thereby construing epistemic responsibility and accessibilism as necessary for knowledge. In addition, one might hold that satisfying epistemic responsibility is necessary to attain some epistemic aims. It is this weaker view that we need to understand why externalism has been thought to provide an unsatisfactory treatment of skepticism. For Pritchard, the externalist ‘side-steps’ the skeptical problem “by making our epistemic standings conditional on the obtaining of worldly facts which one lacks any reflective access to and which are in any case in dispute in a debate with the skeptic”. In doing so, the externalist cannot face the skeptical problem ‘head on’. To face the skeptical problem ‘head on’, we need to provide an account of perceptual knowledge that puts us in a position to make perceptual knowledge claims in the context of a skeptical challenge.

Interestingly, ED seems to offer what we are looking for. Suppose that a subject, S, is in a paradigmatic case of perceptual knowledge. If S perceptually knows p, S has reflectively accessible reasons that conclusively favor p over any incompatible proposition. So, ED gives S all it takes – from an epistemic point of view – to rationally claim knowledge of p in response to a skeptical challenge. Abstracting from pragmatic considerations, S could back up her claims to know p by citing her seeing that p.

I will argue that this view is still prey to the conditionality problem mentioned in the second quote from Pritchard. I will first present the conditionality problem in a general form (section 3) and then show how it arises for ED (section 4).

3. The conditionality problem

Several authors think that externalism provides an unsatisfactory treatment of skepticism because it makes our epistemic standings conditional on the obtaining of worldly facts which we lack any reflective access to (Stroud 1989, 1994, 2004; Fumerton 1990, 1995, 2006; Pritchard 2005, 2012). Why is this a problem? Different authors have provided different answers to this question. Rather than offering an exhaustive review of these answers, I will make a relatively uncontroversial proposal: the conditionality problem is a problem because the conditional character of externalist accounts of knowledge prevents subjects from rationally making any knowledge claims when the conditional character of knowledge becomes salient. Crucially, the conditional character of knowledge becomes salient when a subject recognizes radical skeptical hypotheses as incompatible with her everyday beliefs.⁶

Let us focus on those ‘reflective subjects’ who have examined the problem of skepticism (Stroud 1984: 80–1; Pritchard 2005: 210, 246). Reflective subjects want to determine whether they have any perceptual knowledge of the world. It will be useful to think of those subjects as trying to determine whether a given perceptual belief is a case of knowledge. Simplifying a bit, we could think of reflective subjects as using their favored theory of knowledge to perform ‘self-ascription transitions’ of the following form:

self-ascription transition (I)

Premise 1. I believe p.

Premise 2. A given theory of knowledge is true.

Conclusion. So, I know p.

Starting from a (perceptual) belief, the reflective subject wonders whether that belief is a case of knowledge. Our subject tries to make use of her favored theory of knowledge to move from the self-ascribed (perceptual) belief to a first-person knowledge claim. There is a problem, though. Any theory of knowledge has a conditional structure:

If S's belief in p satisfies conditions C₁, C₂…, C_n, then S knows p.

Given the conditional structure of theories of knowledge, a subject cannot rationally move from her two premises to a first-person knowledge claim. To do so, the subject should be able to ‘detach’ the antecedent of her favored theory of knowledge and apply it to her herself (Stroud 1994: 152). So, our reflective subject should be able to reason as follows:

self-ascription transition (II)

Premise 1. I believe p.

Premise 2. If S's belief in p satisfies conditions C₁, C₂…, C_n, then S knows p.

Premise 3. My belief in p satisfies conditions C₁, C₂…, C_n.

Conclusion. So, I know p.

To make a first-person knowledge claim, the reflective subject needs to determine whether her belief in p satisfies conditions C₁, C₂…, C_n. Recall now the example of the proverbial chicken sexers. Externalists hold that the chicken sexers can have knowledge of the sex of chickens despite their lack of reflective access to the conditions that explain that knowledge. So, the chicken sexers do not know whether the conditions for their own knowledge are satisfied. Perhaps we could grant that the chicken sexers have knowledge. However, this verdict becomes problematic when we generalize the epistemic standing of the chicken sexers to reflective subjects. If reflective subjects only have ‘crude’ externalist knowledge at their disposal, they won't be able to determine whether a given perceptual belief is a case of knowledge. That is, in a nutshell, the conditionality problem.⁷

4. The conditionality problem for epistemological disjunctivism

A solution to the conditionality problem must provide reflective subjects with the resources to ‘detach’ the antecedent of their favored theory of knowledge. One might try to achieve this goal by incorporating an accessibilist component into one's favored theory of knowledge. If the conditions imposed by the theory of knowledge are accessible by reflection, reflective subjects could manage to self-ascribe knowledge (Stroud 2004: 170–1). We have seen that ED incorporates reflective access to factive perceptual reasons. So, one might think that ED provides all we need to detach the antecedent. However, when we consider ED in a broader context, we can see that the conditionality problem has not been solved; it has been pushed one level up (see also Stroud 1994: 152; Fumerton 1995: 178–9).

Here is my argument in a nutshell. ED's account of reflective knowledge of factive perceptual reasons is also conditional on the obtaining of worldly facts which one lacks any reflective access to. The conditional character of reflective knowledge of factive perceptual reasons becomes salient when a reflective subject recognizes a radical skeptical hypothesis as incompatible with her everyday beliefs. Alas, there is nothing in ED that would enable a reflective subject to ‘detach’ the antecedent of ED's conditional characterization of reflective knowledge. This problem immediately spreads to the subject's attempt at self-ascribing first-order knowledge of everyday propositions. Therefore, ED's appeal to reflective knowledge of factive perceptual reasons does not genuinely put a reflective subject in a position to rationally make perceptual knowledge claims when the conditional character of her own reflective knowledge becomes salient in the light of a radical skeptical hypothesis.

I will develop this argument by spelling out the epistemic consequences of a reflective subject's consideration of a radical skeptical hypothesis.

Step 1. While Hannah is strolling through the market, she sees that there are asparagus in the stand. So, she decides to call her husband to report on her finding: ‘I see that there are asparagus in the stand. Would you like to have asparagus for lunch?’ Being a sincere speaker, Hannah has expressed the belief that she sees that there are asparagus in the stand. After Hannah hangs up, she suddenly remembers the brain in a vat (BIV) hypothesis. This irruptive thought is not as surprising as it might seem. Hannah has been working on a paper on radical skepticism. The BIV hypothesis says that Hannah's brain has been recently envatted and her experiences are now produced by a supercomputer. Upon realizing that this error possibility is incompatible with her current belief that she sees that there are asparagus in the stand, Hannah wonders whether she knows that she sees that there are asparagus in the stand.

Step 2. It might be tempting for Hannah to appeal to her reflectively accessible factive perceptual reason to settle the question. But this does not seem to be a viable option. The BIV hypothesis casts doubt on her seeing that there are asparagus in the stand. Dismissing a hypothesis that casts doubt on Hannah's possession of a perceptual reason by citing the same perceptual reason would be like trying to justify the reliability of the testimony of an eyewitness by asking the eyewitness whether she is telling the truth. These considerations lead Hannah to conclude that she may not avail herself of a factive perceptual reason to rationally dismiss the BIV hypothesis.⁸

Step 3. Hannah has some familiarity with externalist theories of knowledge. However, those theories of knowledge do not enable Hannah to answer her original question. After all, no matter what externalist view she favors, any externalist theory will make her epistemic standing “conditional on the obtaining of worldly facts which one lacks any reflective access to and which are in any case in dispute in a debate with the skeptic” (Pritchard 2012: 4). Perhaps ED offers her a way out of her predicament. She recalls the following definition: “In paradigmatic cases of perceptual knowledge, a subject, S, has perceptual knowledge that p in virtue of being in possession of reasons for her belief that p which are both factive and reflectively accessible to S” (Pritchard 2012: 13). Unfortunately, this formulation has a hidden, conditional structure. It tells her that, if she is in a paradigmatic case of perceptual knowledge, she has perceptual knowledge in virtue of being in possession of perceptual reasons that are both factive and reflectively accessible to her. So, the key to answering her original question lies in the concept of a paradigmatic case of perceptual knowledge.

Defenders of ED have developed a complex analysis of paradigmatic cases of perceptual knowledge. For our current purposes, we can focus on one key feature. Paradigmatic cases of perceptual knowledge are ‘objectively epistemically good’. If Hannah is in one of those cases, she is in an environment that allows for the reliable formation of perceptual beliefs and her cognitive and perceptual capacities are functioning properly (Pritchard 2012: 29).⁹

With these materials in hand, Hannah tries to self-ascribe knowledge that she sees that there are asparagus in the stand:

self-ascription transition (III)

Premise 1. I believe that I see that there are asparagus in the stand.

Premise 2. If I am in a paradigmatic case of perceptual knowledge (which entails that I am in an environment that allows for the reliable formation of perceptual beliefs and my cognitive and perceptual capacities are functioning properly), then I can know by reflection that I see that there are asparagus in the stand.

Upon reaching Premise 2, Hannah is stuck. ED tells her that, if she is in a paradigmatic case of perceptual knowledge, she is in possession of perceptual reasons which are both factive and reflectively accessible to her. However, nothing in ED tells her that the condition of being in a paradigmatic case of perceptual knowledge is indeed satisfied. So, Hannah may not self-ascribe knowledge that she sees that there are asparagus in the stand.

To be sure, there is a key difference between ED and (some forms of) externalism: there is less epistemic opacity in ED than in the case of the proverbial chicken sexers.¹⁰ ED explains the possession of perceptual knowledge in paradigmatic cases of perceptual knowledge via reasons that are both factive and reflectively accessible to the subject. However, this difference is of little significance when it comes to Hannah's attempt at self-ascribing second-order knowledge that she sees that p after having recognized a radical skeptical hypothesis as incompatible with her belief that she sees that p. For ED, the possession of this reflective knowledge is relative to the obtaining of worldly facts that the subject “lacks any reflective access to and which are in any case in dispute in a debate with the skeptic” (Pritchard 2012: 4). Unfortunately, this conditional structure is sufficient to generate the conditionality problem.

Some readers might wonder why this version of the conditionality problem matters. After all, it might seem that ED fares better than (some forms of) externalism. While (some) externalists treat all subjects as epistemically on a par with the proverbial chicken sexers, ED enables us to accommodate epistemic responsibility.

The answer is that the conditionality problem for reflective knowledge of factive perceptual reasons will soon contaminate Hannah's ability to rationally claim first-order perceptual knowledge. Suppose that Hannah tries to self-ascribe knowledge that there are asparagus in the stand. Trying to employ ED, Hannah will get stuck at the same point as before:

self-ascription transition (IV)

Premise 1. I believe that there are asparagus in the stand.

Premise 2. If I am in a paradigmatic case of perceptual knowledge (which entails that I am in an environment that allows for the reliable formation of perceptual beliefs and my cognitive and perceptual capacities are functioning properly), then I know that there are asparagus in the stand in virtue of being in possession of reasons for my belief that there are asparagus in the stand which are both factive and reflectively accessible to me.

Unfortunately, it is not reflectively accessible to Hannah whether she is in a paradigmatic case of perceptual knowledge. So, Hannah cannot rationally claim to know [I see that there are asparagus in the stand]. If she cannot rationally claim to know [I see that there are asparagus in the stand], she has no way of backing up her claim to know the first-order proposition [There are asparagus in the stand]. Once Hannah realizes that she cannot take epistemic responsibility for the truth of the second-order belief, and she realizes that this is her way of backing up her first-order belief in paradigmatic cases of perceptual knowledge, the rational support for the first-order belief starts to look baseless.

The conditionality problem relies on a simple logical observation. To make a first-person knowledge claim, the reflective subject should be able to determine whether her perceptual belief satisfies all the conditions for knowledge. Otherwise, the reflective subject won't be able to rationally detach the antecedent of her favored theory of knowledge and apply it to her herself (Stroud 1994: 152). ED tries to improve upon externalist theories of knowledge by reducing the opacity of the subject's epistemic standing. However, this move is insufficient to overcome the logical obstacle that generates the conditionality problem. It suffices that ED posits one opaque worldly condition to prevent the reflective subject from rationally detaching the antecedent. This problem soon contaminates the subject's ability to rationally claim first-order perceptual knowledge.

5. Objections and replies

In this section, I consider some ways in which defenders of ED might try to block the conditionality problem for ED.

Objection 1. Paradigmatic cases of perceptual knowledge are cases where there are no defeaters. In the presence of defeaters, however, subjects lack reflective access to factive perceptual reasons. When a subject engages in self-ascription transitions (III) and (IV), she becomes aware of a defeater. Therefore, ED's account of reflective access is not undermined.

Reply. Talk of defeaters is a red herring. It obscures the fact that the conditionality problem relies on a logical observation. If you are a reflective subject who is in the business of self-ascribing knowledge, you must be able to detach the antecedent of your favored theory of knowledge and apply it to you yourself. One can make this point even if one remains neutral on whether the mere recognition of an error possibility as incompatible with a given belief is sufficient to introduce a defeater. Suppose now that the mere recognition of an error possibility as incompatible with a given belief is sufficient to introduce a defeater. In this case, ED should avail itself of the means of rationally dismissing the defeater and show that those means will enable the subject to detach the antecedent of ED. I shall examine three salient options.

Objection 2. We all tacitly know what it takes to be in a paradigmatic case of perceptual knowledge. Defenders of ED rely on this tacit knowledge when they formulate their theory. This tacit knowledge is sufficient to know that the antecedent of ED is satisfied. Thus, it is a mistake to hold that ED does not provide the reflective subject with sufficient conditions to detach the antecedent of ED.

Reply. Let us grant that we all have tacit knowledge of what it takes to be in a paradigmatic case of perceptual knowledge. We can also concede that some of the conditions to be in a paradigmatic case of perceptual knowledge are reflectively accessible. For example, we are in a position to know whether we have grounds for doubt about a proposition. Thus, we are in a position to tell whether our case is ‘subjectively epistemically good’—to use Pritchard's (2012: 30) apt phrase. However, nothing in ED entails that we can know, by reflection, that our current environment allows for the reliable formation of perceptual beliefs and that our cognitive and perceptual capacities are functioning properly.

Perhaps one might insist that subjects normally have background empirical evidence that speaks to the ‘benevolent’ character of their environment and the proper functioning of their cognitive and perceptual capacities (Vogel 1990; Williams 2001; Pritchard 2012, 2016). Nevertheless, this background empirical evidence is not available in the current context. Radical skeptical error possibilities cast doubt on all one's empirical beliefs. Thus, no empirical reasons are available to ground one's knowledge that one's current environment allows for the reliable formation of perceptual beliefs and that one's cognitive and perceptual capacities are functioning properly (McDowell 1982: 273–4, 391, 1995: 398; Pritchard 2012: 85, 123, 126, 2016: 137–8). Crucially, the satisfaction of these objective conditions is necessary to be in a paradigmatic case of perceptual knowledge.

It might be argued that Hannah has a priori knowledge that she is in an objectively good case (Cohen 1999). Unfortunately, defenders of ED have expressed skepticism about the possession of this type of a priori knowledge (Pritchard 2012: 19, 2016: 129–30). To my mind, they are antecedently committed to making this move. If Hannah had a priori knowledge that she is in an objectively good case, ED's claim that internalism provides an inadequate treatment of skepticism would be unwarranted. If Hannah had a priori knowledge that she is in an objectively good case, it would be hard to understand why ED insists on the possession of perceptual reasons that are both factive and reflectively accessible. Hence, ED is best understood as arising from the conviction that those a priori reasons are unavailable.

Having exhausted all her options, Hannah recalls the ‘indistinguishability intuition’:

indistinguishability intuition. There can be pairs of veridical and non-veridical experiences that cannot be discriminated by introspection alone.

If Hannah grants the indistinguishability intuition, she should also grant that the veridicality of her experience lies beyond the way things seem to her. From a phenomenological perspective, it is an incidental fact about her experience that it is veridical.

This point can be generalized to cases in which a subject's perceptual and cognitive capacities are not working properly.

indistinguishability intuition*. There can be cases in which a subject's perceptual and cognitive capacities are not working properly and yet they cannot be discriminated by introspection alone from cases in which a subject's perceptual and cognitive capacities are working properly.

If Hannah grants the indistinguishability intuition*, she should also grant that the proper functioning of her perceptual and cognitive capacities lies beyond the way things seem to her. From a phenomenological perspective, it is an incidental fact about her own case that her perceptual and cognitive capacities are working properly.

Given the two indistinguishability intuitions, there is nothing in Hannah's tacit knowledge of what it takes to be in a paradigmatic case of perceptual knowledge that would enable her to know, by reflection, that she is in an objectively good case. Without that additional piece of knowledge, Hannah cannot solve the conditionality problem.

Objection 3. It might be objected that Hannah's reasoning has been too hasty. As McDowell (1995, 2002, 2008, 2011, 2013, 2019) has forcefully argued, indistinguishability intuitions register the fact that we are fallible. In other words, they register the fact that it is possible to be misled into thinking that one sees that p when one does not genuinely see that p. However, it does not follow from this familiar fact that, in cases where Hannah has no reason for thinking that she is being misled, she cannot have reflective knowledge that she sees that p. Thus, it is simply not true that, given the prior indistinguishability intuitions, Hannah's being in a paradigmatic case of perceptual knowledge lies beyond the way things seem to her.¹¹

Reply. There is an important insight in this remark: indistinguishability intuitions are consistent with a subject's possession of reflectively accessible factive perceptual reasons in paradigmatic cases of perceptual knowledge. However, we would need to go beyond this plausible observation to solve the conditionality problem. We would need to assume that the consistency between the two claims (fallibility and the possession of reflectively accessible factive perceptual reasons in paradigmatic cases of perceptual knowledge) is sufficient for a reflective subject like Hannah to rationally claim that she sees that p. This is not obviously true, given Step 2 of the argument: “Dismissing a hypothesis that casts doubt on Hannah's possession of a perceptual reason by citing the same perceptual reason would be like trying to justify the reliability of the testimony of an eyewitness by asking the eyewitness whether she is telling the truth. These considerations lead Hannah to conclude that she may not avail herself of a factive perceptual reason to rationally dismiss the BIV hypothesis.” If we factor in this observation, we have a situation in which two incompatible propositions are on an equal footing: [I see that there are asparagus in the stand] and [I merely seem to see that there are asparagus in the stand]. Of course, it does not follow from Hannah's fallibility that she merely seems to see that p. However, it does not follow that she genuinely sees that p either. At this stage of the game, the rational attitude would be agnosticism.

Objection 4. The previous reply underestimates the resources available to ED. Defenders of ED have not merely identified an invalid argument from fallibility to the claim that subjects lack perceptual reasons that are both factive and reflectively accessible. They have also argued that, in the absence of reasons to think the contrary, it is rational for Hannah to believe that she sees that there are asparagus in the stand. Pritchard (2012, 2016) has developed this line of argument on behalf of the Austinian distinction between ‘merely raised’ and ‘rationally motivated error possibilities’. On Pritchard's view, a subject can rationally dismiss an error possibility in two different ways. If the error possibility is rationally motivated, the subject should provide independent reasons that speak to the reasons offered in support of the error possibility. If the error possibility is not rationally motivated (but it is ‘merely raised’), it is rational for the subject to ignore the error possibility. Think of hypotheses formulated ‘on a whim’. It seems wrong to expect someone to provide independent reasons to dismiss whimsical hypotheses.

Pritchard thinks that this distinction can be used to rationally dismiss radical skeptical hypotheses (see also McDowell 2014: 319–20). On Pritchard's view, radical skeptical hypotheses are never rationally motivated. He reasons as follows. Radical skeptical hypotheses call into question all our empirical grounds. So, there can be no empirical considerations that speak in favor of radical skeptical hypotheses. If one assumes that the only way of rationally motivating a radical skeptical hypothesis is to cite empirical grounds, it follows that radical skeptical hypotheses are never rationally motivated.¹² So, we may ignore radical skeptical hypotheses without being required to introduce independent considerations that speak against them:

[I]f this just means the mere presentation of a not-p possibility, then there is nothing inherently suspect about the idea that our agent can continue to cite the factive rational support she has for the target proposition, and hence regard this error possibility as excluded. Effectively, she is treating this particular evidence for not-p as misleading (though ‘evidence’ is not quite the right word, given that the error possibility isn't rationally motivated). (Pritchard 2016: 211, footnote 19; see also his 2012: 125–9)

Reply. Let us concede that there is a sense in which it is rational to set aside merely raised error possibilities. Let us also grant that radical skeptical hypotheses are never rationally motivated. However, the key question is whether this move is sufficient to solve the conditionality problem. In my view, the distinction between ‘merely raised’ and ‘rationally motivated error possibilities’ cannot solve the conditionality problem. To see why, let us consider the following self-ascription transition:

self-ascription transition (V)

Premise 1. If I believe p and I recognize a hypothesis h both as incompatible with p and as a hypothesis that is never rationally motivated, then I may believe p.

Premise 2. I believe that I see that there are asparagus in the stand.

Premise 3. I recognize the BIV hypothesis both as incompatible with my seeing that there are asparagus in the stand and as a hypothesis that is never rationally motivated.

Conclusion. So, I may believe that I see that there are asparagus in the stand.

This self-ascription bypasses the relativity of ED to paradigmatic cases of perceptual knowledge by relying on the rationality of ignoring hypotheses that are not rationally motivated and the claim that the BIV is never rationally motivated. Unfortunately, Premise 1 is too weak to solve the conditionality problem.

Let us grant that, in everyday conversations, it is legitimate to set aside error possibilities that are not rationally motivated. Nevertheless, this concession is insufficient to solve the conditionality problem. To solve this problem, Hannah needs an epistemic reason to believe that she is in an objectively good case. Roughly, an epistemic reason in favor of p is a reason that speaks to the truth of p. Nevertheless, the fact that the BIV hypothesis is never rationally motivated neither speaks to the truth of her being in an objectively good case, nor to the falsity of the BIV hypothesis. In objectively good cases, the subject can reliably form perceptual beliefs and her cognitive and perceptual capacities are functioning properly. The fact that the BIV hypothesis is never rationally motivated speaks neither to the reliability of Hannah's perceptual belief-forming mechanisms nor to the proper functioning of Hannah's cognitive and perceptual capacities. To paraphrase Pritchard (2016: 40), the fact that a hypothesis has been merely raised seems to be “completely indifferent to whether or not we are the victims of a radical skeptical scenario”. Imagine that Hannah is a BIV. At some point, she has an experience as of someone who tells her, ‘You are a BIV’. Suppose now that Hannah realizes that there is no way in which this hypothesis could be rationally motivated. So, it might be rational for Hannah to set aside the BIV hypothesis. Nevertheless, Hannah still lacks epistemic reasons that speak to the truth of her being in an objectively good case and the falsity of the BIV hypothesis. For all Hannah knows, she might be living an envatted existence.¹³

6. A diagnosis

I have argued that ED cannot solve the conditionality problem. In this section, I submit that the conditionality problem is parasitic on a controversial methodological requirement. After that, I suggest that ED need not meet that requirement. Therefore, it is possible to put ED in the service of a more modest anti-skeptical strategy.¹⁴

Stroud thinks that the conditionality problem originates from a requirement of the traditional epistemological project:

generality requirement. The theory of knowledge should explain all our knowledge of the world at once.¹⁵

In the project of ‘pure inquiry’, the inquirer sets aside all her interests and attachments to examine all her knowledge at once (Williams 1978, 1991: 6, 26, 193; Stroud 1984). The generality requirement entails that the pure inquirer should set aside all her knowledge of the world; she should see herself at the outset as “not knowing anything about the external world” (Stroud 1989: 120; see also Stroud 1984: 209, 223; Williams 1991: 193). This creates a problem, though. If the inquirer sets aside all her knowledge of the world, a successful explanation of knowledge of the world should restitute the subject's ability to self-ascribe knowledge of the world. To perform this self-ascription, however, the subject must determine that all the conditions for knowledge are satisfied. Alas, some of these conditions involve worldly facts. Thus, the satisfaction of all the conditions for knowledge can only be ascertained by relying on some knowledge of the world. This leads to a dilemma: either we must embrace skepticism, or we must give up on the project of offering a fully general explanation of our knowledge.

Many philosophers have noted that the conditionality problem imposes very demanding epistemological requirements (Sosa 1994, 2016; Greco 2000; Bergmann 2008; Echeverri Ms a). In my view, they are right to think so. However, very few philosophers have realized that, when a reflective subject tries to rationally dismiss a radical skeptical hypothesis, she is tacitly subject to the generality requirement.¹⁶ A radical skeptical hypothesis casts doubt on all our empirical grounds. So, when the reflective subject tries to rationally dismiss a radical skeptical hypothesis, she finds herself in the same situation as someone who must restitute her knowledge of the world without illicitly relying on any knowledge of the world. This raises the question of whether there is any way of rejecting the generality requirement without giving up on the project of providing an adequate treatment of skepticism.

Since I have discussed this issue elsewhere (Echeverri Ms a), I will only draw a moral for our current discussion. The way ED entered the philosophical scene seems to be at odds with the project of pure inquiry and its correlative idea of facing skepticism ‘head on’. Rather than embracing the generality requirement, ED is best seen as starting from the commonsense assumption that subjects often are, or can be, in paradigmatic cases of perceptual knowledge. From this perspective, the philosophical task is not to provide an account of perceptual knowledge that confers on us the ability to rationally claim perceptual knowledge in response to skeptical challenges, but to preserve the central role of reflectively accessible factive perceptual reasons in our everyday practices. In this vein, ED can be put in the service of undercutting putative skeptical paradoxes. Given that putative skeptical paradoxes are not challenges one must meet by rationally claiming perceptual knowledge, the two tasks are potentially dissociable.

7. Underdetermination-based skepticism

Pritchard has offered an illuminating contrast between two ways of understanding the skeptical problem. On one interpretation, radical skepticism is a position. On this view, there is a (possible or actual) skeptic who takes on several commitments such as ‘knowledge is impossible’. On another interpretation, radical skepticism is a putative paradox that arises from a set of claims that strike us as individually true, but which (seem to) entail that knowledge of the external world is not possible (Pritchard 2012, 2016; see also Stroud 1984; Wright 1985; Williams 1991, 2001; Byrne 2004). This second view depicts “[t]he dispute with the ‘skeptic’ [as] a quarrel that is completely internal to our conceptual realm” (Pritchard 2016: 161).

If skepticism is construed as a position, it may seem reasonable to expect a theory of perceptual knowledge to secure our ability to rationally make perceptual knowledge claims in response to skeptical challenges. After all, knowledge claims are intended for an audience, even if that audience turns out to be the skeptic in you. If skepticism is construed as a putative paradox, however, our ability to rationally make perceptual knowledge claims seems irrelevant.

There are two main ways of solving a putative paradox. An overriding strategy grants that radical skepticism is a bona fide paradox, for “it arises out of our most fundamental epistemological commitments” (Pritchard 2016: 161). Therefore, an anti-skeptic project should provide independent arguments to revise (some of) the pre-theoretical commitments that generate the paradox (see also Wright 1985; Greco 2000: 3). By contrast, the undercutting strategy denies that the putative skeptical paradox is a genuine paradox. It holds that radical skepticism “smuggles contentious theoretical claims into the set-up of the skeptical argument, disguised as common sense” (Pritchard 2016: 172; see also Clarke 1972; Williams 1991, 2001).

Imagine that you opt for an overriding strategy. It suffices to rationally motivate a revision of (some of) the pre-theoretical commitments that generate the paradox. These commitments can have the form of general principles that we tacitly rely on when we assess knowledge claims in everyday practices. Alternatively, suppose that you choose the undercutting strategy. In this case, it seems sufficient to display the theoretical character of the putative paradox. In doing so, you do not need to take the extra step of securing the capacity to rationally make perceptual knowledge claims in response to skeptical challenges.

Of course, one's verdict on a putative skeptical paradox will depend on how exactly one understands it. ED has been used to block the closure-based skeptical paradox (Pritchard 2012) and the underdetermination-based skeptical paradox (Pritchard 2016). Pritchard's considered view is that ED is best suited to block the underdetermination-based skeptical paradox. Therefore, I will focus on this putative paradox. To this end, I will present Pritchard's account of the underdetermination-based skeptical paradox (section 7.1), sketch a modest solution to it (section 7.2), show that this modest solution is available to some forms of externalism (section 7.3), and suggest that this modest solution undercuts the putative paradox (section 7.4).¹⁷

7.1. The underdetermination-based skeptical paradox

The underdetermination-based skeptical paradox is the conjunction of three claims:

underdetermination-based skeptical paradox

U1. One cannot have reflectively accessible reasons that favor one's belief in an everyday proposition over an incompatible radical skeptical hypothesis.

U2. If one cannot have reflectively accessible reasons that favor one's belief in an everyday proposition over an incompatible radical skeptical hypothesis, then one does not have widespread everyday knowledge based on reflectively accessible reasons.

U3. One has widespread everyday knowledge based on reflectively accessible reasons.

U1–U3 form an inconsistent triad. An undercutting solution must reject at least one of the three claims without revising any of our pre-theoretical commitments. Otherwise, we are led to advocate an overriding solution. Pritchard thinks that externalists are committed to offering an overriding solution.

Consider a radical externalist. Our radical externalist holds that the satisfaction of an external condition on knowledge (reliability, safety, sensitivity, and so on) is necessary and sufficient for knowledge. Our radical externalist is committed to the truth of U1, for there is nothing in her theory that licenses the claim that one can have reflectively accessible reasons that favor one's belief in an everyday proposition over an incompatible radical skeptical hypothesis.¹⁸ This creates a problem. If one cannot have reflectively accessible reasons that favor one's belief in an everyday proposition over an incompatible radical skeptical hypothesis, and U2 is true, then U3 is false: It is not the case that one has widespread everyday knowledge based on reflectively accessible reasons. But this leads us to a troubling form of skepticism. The conjunction of U1 and U2 leads us to assimilate all our perceptual knowledge to the knowledge possessed by the chicken sexers (Pritchard 2016: 35–6; see also Pritchard 2005: 115, 206). This result conflicts with our pre-theoretical epistemic commitments. As we have seen, it is very natural to hold that there is an epistemic asymmetry between the proverbial chicken sexers and their enlightened counterparts (section 2).

Another option would be to reject U2. However, Pritchard thinks that U2 relies on a very intuitive principle:

underdetermination. If S knows that p and q are incompatible propositions, and yet S lacks reflectively accessible reasons that favor p over q, then S lacks knowledge of p based on reflectively accessible reasons.¹⁹

Suppose that Hannah knows that the proposition [There are asparagus in the stand] is incompatible with the proposition [There are leeks in the stand] (Let us assume that there is a single type of vegetable in the stand). Suppose that Hannah lacks reflectively accessible reasons that favor [There are asparagus in the stand] over [There are leeks in the stand]. Therefore, it seems highly intuitive that Hannah lacks knowledge of [There are asparagus in the stand] based on reflectively accessible reasons. It is hard to find anything objectionable in this inferential pattern.

Pritchard thinks that internalist accessibilism faces similar problems. Assuming that he is right to think so, the only way of undercutting the putative paradox is to reject U1: It is not the case that one cannot have reflectively accessible reasons that favor one's belief in an everyday proposition over an incompatible radical skeptical hypothesis. A good way of rejecting U1 is to identify a principle underlying it. Pritchard suggests that U1 relies on the insularity of reasons thesis:

insularity of reasons. The reasons we have reflective access to in paradigmatic cases of perceptual knowledge are compatible with their widespread falsity (Pritchard 2016: 3, 55–6, 172).

If we reject insularity of reasons, we can reject U1 while endorsing U2 and U3. Happily, ED offers us the materials to reject insularity of reasons. Assume that Hannah can be in paradigmatic cases of perceptual knowledge. Given ED, Hannah can have reflectively accessible reasons that favor the everyday proposition [There are asparagus in the stand] over an incompatible radical skeptical hypothesis like [I am a BIV]. Having rejected U1, we can keep U2 and U3 without contradiction. Thus, we can avoid the counterintuitive result of assimilating all our perceptual knowledge to the knowledge possessed by the proverbial chicken sexers.

7.2. Modest ED

ED's denial of U1 relies on a key assumption: Hannah can be in paradigmatic cases of perceptual knowledge. To solve the conditionality problem, a reflective subject should be able to discharge this assumption. Let us define ‘ambitious ED’ as the view that uses ED to discharge this assumption. Interestingly, we do not need to discharge this assumption if we are pursuing the more modest goal of solving a putative paradox. This follows from our definition of a putative paradox. One can successfully solve a putative paradox if one manages to sketch a theory of knowledge that eliminates the contradiction generated by a set of claims. One can successfully undercut a putative paradox if one can show that the resulting view coheres with one's pre-theoretical commitments. Crucially, one can sketch an account of perceptual knowledge that coheres with one's pre-theoretical epistemic commitments even though one cannot vindicate those commitments in response to a skeptical challenge. What did the work in the prior solution was the denial of insularity of reasons. But what justifies this denial? McDowell (2002: 98) and Pritchard (2012: 17, 2016: 134) have argued that insularity of reasons is at odds with our pre-theoretical epistemic commitments. They have pointed out that ED captures the practice of citing factive perceptual reasons in everyday life. They have also insisted that it is very unusual to cite non-factive perceptual reasons in everyday practices.²⁰

Here is another way of looking at the dialectic. Ambitious ED holds that a reflective subject can solve the conditionality problem when she recognizes a radical skeptical hypothesis as incompatible with her everyday beliefs. Suppose now that the conditionality problem has no solution. Thus, ambitious ED is not available. Modest ED holds, by contrast, that a reflective subject cannot solve the conditionality problem when she recognizes a radical skeptical hypothesis as incompatible with her everyday beliefs. After all, radical skeptical hypotheses reveal the conditional character of S's reflectively accessible reasons and there is nothing in ED that would enable a reflective subject to detach the antecedent of ED. Given that the theoretical work was done by ED's denial of insularity of reasons because it deviates from our pre-theoretical epistemic commitments, the proposed solution is orthogonal to the debate between ambitious ED and modest ED. insularity of reasons is a synchronic claim about the limited strength of the perceptual reasons that are reflectively accessible in paradigmatic cases of perceptual knowledge. All we can get from ED is the denial of this synchronic claim. However, the contrast between ambitious and modest ED concerns the diachronic question of whether a reflective subject can solve the conditionality problem after she recognizes a radical skeptical hypothesis as incompatible with her everyday beliefs.

Does modest ED lead to first-order skepticism? The answer to this question will depend on one's further commitments about knowledge. If there can be non-luminous knowledge (Williamson 2000; Lasonen-Aarnio 2010), a subject who recognizes a radical skeptical hypothesis as incompatible with her everyday belief might retain a more primitive, non-luminous perceptual knowledge of the world. If one denies the existence of non-luminous knowledge, the proposed analysis will lead to a diachronic form of first-order skepticism.²¹

7.3. Modest ED and externalism

Radical externalists are compelled to grant U1. Nevertheless, not all externalists are radical. We have already introduced a weaker form of externalism according to which externalist conditions on knowledge might be necessary and sufficient for knowledge when skeptical hypotheses are ‘live’. Other externalists could try to offer externalist accounts of perceptual reasons that are both factive and reflectively accessible (Greco 2014). For example, a subject can know p in virtue of seeing that p, which is both factive and reflectively accessible. Crucially, her reflective knowledge that she sees that p could be explained in virtue of the obtaining of some external condition (reliability, safety, sensitivity, and so on). This type of externalism could mimic modest ED's solution to the underdetermination-based skeptical paradox. Our discussion suggests, however, a stronger, externalist claim. The reason why the conditionality problem has no solution is that all our knowledge is ultimately externalist knowledge. When we go all the way down to vindicate all our perceptual knowledge at once, we reach a type of explanation that traces knowing to the obtaining of some external conditions. When the reflective subject faces the skeptical problem ‘head on’, she lacks epistemic reasons that speak to the truth of her seeing that p because her seeing that p is ultimately grounded in the obtaining of some external condition. Yet, this is consistent with a subject's possession of reflectively accessible factive perceptual reasons in paradigmatic cases of perceptual knowledge and her retention of a more primitive first-order perceptual knowledge when skeptical hypotheses are ‘live’.

7.4. Overriding vs. undercutting

Recall Pritchard's distinction between overriding and undercutting strategies. He seems to think that ambitious ED undercuts the underdetermination-based skeptical paradox. I beg to disagree. Ambitious ED conceals an overriding component. Indeed, modest ED is in a better position than ambitious ED to genuinely undercut the underdetermination-based skeptical paradox.

A way of determining whether a theory of knowledge reflects our pre-theoretical epistemic commitments is to ask whether it articulates the principles at work in our everyday epistemic practices. ED tells us that, in paradigmatic cases of perceptual knowledge, a subject, S, has perceptual knowledge that p in virtue of being in possession of reasons for her belief that p which are both factive and reflectively accessible to S (Pritchard 2012: 13). This claim gains support from the role of factive perceptual reasons in everyday exchanges of reasons. However, ambitious ED goes beyond this simple observation. Ambitious ED requires that a subject can avail herself of her reflectively accessible factive perceptual reasons when she becomes aware of a radical skeptical hypothesis as incompatible with her everyday beliefs. Everyday practices do not support this stronger claim. While it may seem natural to cite a factive perceptual reason when radical skeptical hypotheses have been bracketed, it is extremely unnatural to do so when radical skeptical hypotheses are ‘live’. Indeed, it requires a lot of reasoning to make the view palatable. So, ambitious ED is the result of theoretical reasoning that is presented as common sense. Pace Pritchard, ambitious ED is an overriding strategy.²²

If modest ED can capture our pre-theoretical epistemic commitments better than ambitious ED, why is it so tempting to prefer ambitious ED over modest ED? Defenders of ED are reluctant to assimilate all our perceptual knowledge to the knowledge possessed by the proverbial chicken sexers. There is, however, an ambiguity in U3: One has widespread everyday knowledge based on reflectively accessible reasons. On a modest reading, the epistemic position of the enlightened chicken sexers is relatively better than the epistemic position of the proverbial chicken sexers. Any epistemological theory that blurs the relative epistemic asymmetry between the two groups of subjects would be inadequate. On an immodest reading, preserving this relative epistemic asymmetry is not enough. It is also necessary to show that the reasons available to the enlightened chicken sexers are so sound that they remain available when they recognize a radical skeptical hypothesis as incompatible with their everyday beliefs. Pritchard seems to flirt with this immodest reading when he describes the (radical) externalist as embracing the counterintuitive “claim that our (perceptual) knowledge is in its nature lacking in rational support” (2016: 35–6). This claim misleadingly prejudges the case against any externalist outlook. The insolvability of the conditionality problem suggests that, even if there is a relative epistemic asymmetry between the proverbial chicken sexers and their enlightened counterparts, our knowledge is ultimately externalist knowledge. So, in a sense, all our knowledge is in its nature ultimately lacking in rational support.

Some might find this conclusion disquieting. However, our unease with a solution need not reveal its failure to capture our pre-theoretical epistemic commitments. Some disquieting conclusions could reveal overlooked aspects of our natural condition. A young child is oblivious to the unavoidability of death. Growing up will inevitably face her with this fundamental aspect of life. This realization might be a cause of long-lasting anxiety, even if our child had no prior epistemic commitment to being immortal. Something similar might be at work in radical skepticism. Not all disquieting revelations contradict our pre-theoretical epistemic commitments. Some disquieting revelations can help us sharpen the relatively vague contours of common sense. If skepticism is a putative paradox, one can eliminate the tension in our conceptual framework and remain anxious with the result. The remaining anxiety can arise from the realization that we cannot occupy an unconditioned epistemic position. Learning to live with this realization is part and parcel of growing up.

8. Conclusion

Epistemological disjunctivism (ED) has been thought to offer a way out of the conditionality problem for epistemic externalism. I have shown that ED's account of reflectively accessible factive perceptual reasons is key to ED's attempt at solving that problem. I have also argued that ED's conception of reflective access faces a version of the conditionality problem. Given that this problem prevents a reflective subject from self-ascribing first-order knowledge of external world propositions, ED's attempt at solving the conditionality problem fails. Fortunately, the conditionality problem arises from a generality requirement that is potentially dissociable from ED. This makes room for a modest understanding of ED's anti-skeptical project that can survive the realization that the conditionality problem has no solution. ED can help us undercut the underdetermination-based skeptical paradox by enabling us to reject the insularity of reasons thesis on behalf of our pre-theoretical commitment to our possession, in paradigmatic cases of perceptual knowledge, of perceptual reasons that are both factive and reflectively accessible. Unfortunately, this modest ED offers no substantial advantages over all forms of externalism. Indeed, a moderate externalism provides a good framework to explain the anti-skeptical limitations of our reliance on reflectively accessible factive perceptual reasons.^{23, 24}