Keywords

8.1 Introduction

An idea that has enjoyed influential support in epistemology holds that a belief is known only if it is suitably backed up by justified beliefs that can be adduced as premises. Externalists in epistemology deny this, and doing so brings with it a straightforward way out of the ancient Pyrrhonian Problematic—albeit, one that requires an explanation: how, exactly, does a given belief rise to the level of knowledge in circumstances in which it is not backed by justified beliefs?

A different way to put this question to the externalist is: what kind of thing—if not justified beliefs—can function as a regress stopper? While it is available to both internalists and externalists in epistemologyFootnote 1 to answer this question with ‘experience’, the externalist should hope to be able to account for how experience is suited to play such a role while at the same time steering clear of any commitment to the Myth of the Given. Those who buy into the Myth hold that experience can play such a ‘regress-stopping’ role only by presenting itself to a thinker in a special way, one whereby it is directly (non-inferentially) believed (on Classical Foundationalism, with certainty) to be present; and then the thinker can reason from this non-inferential belief to “conclusions about the world beyond experience” (Sosa 2009, 89). Wilfrid Sellars (1956) and othersFootnote 2 have registered a slew of problems for proposals that rely on such a story, and I won’t attempt to rehearse them here. Instead, I want to simply highlight that if the externalist is going to avoid the Myth of the Given, then a different kind of story altogether is needed for how experience might stop a regress. In the simple case of perceptual beliefs, this story must tell us something about how experience could bear epistemically on our justified (and known) perceptual beliefs without simply ‘being directly apprehended with certainty’.

A straightforward externalist stance that Ernest Sosa (2009) embraces at this juncture, drawing some inspiration from Thomas Reid, holds that “[e]xperience can bear epistemically on the justification of a foundational perceptual belief by appropriately causing that belief” (2009, 89). This very idea might strike some philosophers with internalist leanings as deeply wrongheaded. One way to put the worry is that this stance confuses (as Richard Rorty put it) ‘causation with justification’.Footnote 3 But there is a less contentious way to state the concern: the thought that a belief might be justifiably believed or known simply by being suitably caused seems to concede nothing to the internalist, to any extent. What the externalist should ultimately hope to do is to avoid the Myth of the Given (as well, of course, as scepticism) while doing at least some justice to internalist intuitions. Sosa (2009) has a strategy for doing all of this, and it’s one that relies indespensibly on the idea of an endorsing perspective:

The best way, I contend, is to distinguish between the animal knowledge that makes only minimal concessions to internalism, and reflective knowledge, with its much more substantial internalist component constituted by its requirement of an endorsing perspective. This perspective requires the endorsement of one’s epistemic competences and with this comes an evident threat of vicious circularity (2009, 44).

On this picture, a suitable externalist epistemology must be a bi-level epistemology, one on which the kind of high-grade knowledge that humans ultimately aspire to is perspectival knowledge (even if animal knowledge is non-perspectival). In particular, such high-grade knowledge requires one to take an endorsing perspective on one’s own competences. And here Sosa is right to register that the threat of vicious circularity looms.Footnote 4 Epistemic competences, for Sosa, are reliable, truth-conducive dispositions such as perception, memory, reason, etc. But, as the worry goes, how could we ever arrive at the conclusion that such competences are reliable without taking for granted that these faculties are reliable in the course of our own reasoning?Footnote 5

Before considering the seriousness of this kind of problem for Sosa’s virtue perspectivist epistemology, and whether the view can emerge unscathed, it is worth highlighting (beyond what Sosa himself has) what is at stake here. Sosa has argued in various places throughout his career that a non-sceptical epistemology will have to be an externalist epistemology; only with externalism do we have the resources to respond in an adequate way to sceptical arguments from dreaming and radical deceptions.Footnote 6 That said, a plausible externalist epistemology will need to do (at least some) justice to internalist intuitions while avoiding the Myth of the Given, and here is where it is said that externalism must take the form of virtue perspecitivism. Now, if virtue perspectivism succumbs to vicious circularity, then this will lead to one of two paths: scepticism (which externalism is meant to help us avoid) or relativism, where the latter draws attention to the fact that viciously circular justification is a status that can be attained in principle through any kind of endorsing perspective, including those that are in conflict with one another.Footnote 7 It looks, then, like quite a lot may depend on whether virtue perspectivism succumbs to vicious circularity. In this paper, I am going to focus on circularity-based criticisms of virtue perspectivism raised in various places by Barry Stroud (2004), Baron Reed (2012), and Richard Fumerton (2004), and I will argue that virtue perspectivism can withstand each challenge.

8.2 Virtue Perspectivism and Circularity

On Sosa’s virtue perspectivism, there are two kinds of knowledge: animal and reflective.

[…] animal knowledge does not require that the knower have an epistemic perspective on his belief, a perspective from which he endorses the source of that belief, from which he can see that source as reliably truth-conducive. Reflective knowledge does by contrast require such a perspective (2009, 135).

What animal knowledge positively requires is just that one’s beliefs are reliably truth-conductive, in that their correctness must manifest a reliable disposition of the believer. In the perceptual case, the knower’s perceptual beliefs must be prompted truth-reliably by perceptual experiences. And in this way, one can come to have basic perceptual knowledge without relying on the kind of ‘givenness’ of classical foundationalism.Footnote 8 Furthermore, in securing such knowledge one needn’t reason inductively from beliefs about the qualitative character of perceptual experiences.Footnote 9

That said, let’s now consider the intellectual gain that, on Sosa’s view, a thinker is supposed to have secured when one transitions in the perceptual case (via satisfying the perspectival condition) from animal to reflective knowledge. A full description of this story will surely include the following admission: that in transitioning from the animal to the reflective, one would be arriving at a positive view of one’s faculties that relied on, and so implicitly trusted, the use of those very faculties.

But isn’t this circular? Or, what is more relevant: is what is described here a kind of vicious circularity? At this point, there is an interesting kind of reductio ad absurdum that Sosa first canvasses himself, and which has since become a point of focus in Sosa’s dispute with Stroud. The thought is brought out when we compare an ordinary perceiver with “a crystal ball gazer who thinks that what he can see in the ball enables him to tell about matters beyond” (2009, 135). If we suppose the crystal ball indicated to the gazer a favourable view of the epistemic credentials of the crystal ball, we would not think much of the epistemic status of that favourable perspective (or what it has to offer). But—and this is the worry Sosa initially envisages for virtue perspectivism—isn’t the ordinary perceiver in just the analogously same situation as the crystal ball gazer who comes to trust crystal ball gazing in that each relied on, and so implicitly trusted, the use of their respective faculties (or ways of forming beliefs) in coming to have a positive view of those faculties?

Sosa grants that the kind of epistemic circularity that is implied by the move from animal to reflective knowledge would be viciously circular if it made the gazer equally as justified as the ordinary perceiver. But he denies that it does, and so insists it is not. Here some care is needed because the reasoning Sosa offers for why the two are not on a par is not ultimately convincing to Stroud. What Sosa is willing to grant is the following: what the gazer (all going well) can attain is a kind of internal coherence that is every bit as equal to the internal coherence that the ordinary perceiver attains. Despite this concession, he maintains that:

There are faculties other than reason whose apt functioning is also crucial to the subject’s epistemic welfare. In light of that result, why not distinguish between the gazers and the perceivers in that, although both reason properly and attain thereby coherence and justification, only the perceivers are more fully epistemically competent and attain knowledge? On this view, the crystal-gazers differ from the perceivers in that gazing is not reliable while perceiving is. So the theory of knowledge of the perceivers is right, that of the gazers wrong. Moreover, the perceivers can know their theory to be right when they know it in large part through perception, since their theory is right and perception can thus serve as a source of knowledge. The gazers are by hypothesis in a very different position. Gazing, being unreliable, cannot serve as a source of knowledge. So the perceivers have a good source or basis for their knowledge, but the gazers, lacking any such source or basis, lack knowledge (2009, 200–201).

If the foregoing is right, then Sosa can successfully sidestep the reductio he canvasses. For it’s not true that the kind of circularity implied by one’s move from animal to reflective knowledge would leave one epistemically on an equal footing with the crystal ball gazer. Even more, as he notes, the idea that one would need to use (and in doing so take for granted the reliability of) a given faculty at some point to come to have an adequate view of that faculty’s epistemic status is implied by the very possibility of having ‘an adequate theory of our knowledge and its general sources’ (Sosa 2009, 196). As Sosa submits, no one could provide adequate support of any of our sources of knowing, including perception, memory, deduction, abduction, and testimony, without employing those faculties (see Sosa ibid., 201.)

Sosa’s response to the question of vicious circularity can accordingly be summarised as follows: the kind of circularity his reflective knower succumbs to in the case of perception is vicious only if either (i) an adequate theory of our knowledge and its general sources is impossible, or (ii) the perceiver is on an equal epistemic footing as the gazer. We have no good reason yet to think (i) is true and (ii) is disputed in the passage quoted above. And so the circularity is not vicious, but benign.

Sosa’s reasoning, and in particular his way of distinguishing the epistemic plights of the perceiver as opposed to the gazer, will not seem very amenable to an internalist. Before considering Stroud’s assessment of this situation, it’s worth pointing out one aspect of the dialectic here that might be easily overlooked: while Sosa’s distinguishing the perceiver and the gazer looks like hardened externalism, the way Sosa characterises the epistemic position of the reflective knower is not itself unconcessionary to the internalist. The reason that this point can be easily elided is that the thought experiment simply assumes that the gazer enjoys the same broad coherence in her beliefs as the perceiver. On Sosa’s view—and this is the nod to the internalist—this broad coherence adds to the epistemic value of the perceiver’s belief, though the value it adds is greater than the value that the gazer’s coherence adds to the gazer’s belief. And this is because such broad coherence (in the case of the perceiver) is truth conducive, and yields integrated understanding (even if it would not for the gazer, or for the perceiver were she situated in a demon world).Footnote 10 I mention this point here simply to register that despite the hardened externalist feel of Sosa’s response to the perceiver/gazer reductio, the virtue perspectivist view advanced countenances the epistemic value of coherence (in the right circumstances) in a way that sets it apart from genuinely hardened externalist views (such as Goldman’s (1999) process reliabilism or Armstrong’s (1973) causal account) that make no such internalist concessions at all.

8.3 Stroud on Sosa and Circularity

According to Barry Stroud (2004), Sosa’s virtue perspectivism on closer inspection lacks the resources to suitably account for the difference between the epistemic plights of an ordinary perceiver and a crystal ball gazer. This is quite a charge. Note that the perceiver/gazer analogy was initially raised by Sosa himself in the context of discussing epistemic circularity. To be clear, the worry Sosa initially envisaged for his view (which the preceding section was a response to) went as follows: the thinker who transitions from animal to reflective knowledge (on Sosa’s view) looks to be (prima facie) in an epistemically analogous position as a crystal ball gazer (who relies on the ball to positively endorse its usage) in that each relied on, and so implicitly trusted, the use of their respective faculties (or ways of forming beliefs) in coming to have a positive view of those same faculties.

But Stroud seems to be registering a sense in which, on Sosa’s view of knowledge, the ordinary perceiver turns out to be epistemically impoverished—in an analogous way as the gazer is—regardless of whether either approach succumbs to vicious circularity, per se. If this charge can be made to stick, then it’s obviously a real problem for virtue perspectivism. Let’s now look at the charge more carefully. It is levelled at Sosa against the background of a specific desideratum on an account of knowledge in mind. This desideratum that holds that a philosophically satisfying account of our knowledge of the external world must not merely be true, but it also must be something that we know to be true. Sosa himself does not dispute the desideratum. But he, unlike Stroud, thinks that virtue perspectivism can respect it. So why, then, does Stroud think Sosa’s view cannot respect it? It will be helpful to consider two key passages from Stroud:

The question is whether holding such a theory leaves anyone in a position to gain a satisfactory understanding of knowledge of the world, even if he fulfills the conditions Sosa’s theory says are sufficient for knowledge. Could someone in such a position come to recognize himself as knowing, and not merely confidently believing, perhaps even truly, that sense perception is a way of getting knowledge of the world and crystal ball gazing is not? I think that, on the understanding of perception that appears to be involved in Sosa’s question […], the answer is ‘No’. On that view, what we are aware of in perception is restricted to features of our perceptual experiences. The external facts we know as a result of those experiences are nothing we ever perceive to be so. What we get in sense perception therefore bears the same relation to the world we think we know by that means as what is seen in crystal ball gazing bears to the world the gazers think it gives them knowledge of (2004, 171–2).

In response to the above, Sosa briefly makes explicit that we do believe that our perceptual experiences are reliably connected with what we think we know on their basis—a point he registers before engaging with the following key piece of Stroud’s reasoning:

But anyone who thinks that all it takes to have satisfactory understanding of perceptual knowledge is to conclude by modus ponens that we know by perception that there are external things would have to concede that the crystal ball gazers have a satisfactory understanding of crystal ball gazing knowledge. They could draw the corresponding conclusion equally confidently from what they believe about themselves (2004, 172).

The dialectic at this stage between Sosa and Stroud becomes especially thorny. Whereas Stroud is insisting that Sosa’s account of knowledge is no more philosophically satisfying than the corresponding account of knowledge that might be advanced by a crystal ball gazer, Sosa for his part, finds Stroud’s reasoning “hard to follow” (2009, 207). As Sosa sees things, Stroud was purporting to have granted for the sake of argument Sosa’s view that perceptual knowledge is a matter of perceptual beliefs being prompted truth-reliably by perceptual experiences. That said, Sosa questions whether Stroud can coherently suppose that “if we conclude by modus ponens that we know about the world around us through perception, given that our perceptual faculties are reliable, then we are in the predicament he alleges?” (ibid. emphasis added). Specifically, Sosa thinks such a supposition is refuted by:

[…] a crucial difference that Stroud and I both recognize: namely, that we know our perceptual faculties to be reliable whereas the gazers believe but do not know their gazing to be reliable. So, how can we be in an equally good epistemic position to understand how we know, if we do know but they do not know about the reliability of the faculties involved? (2009, 207)Footnote 11

It’s hard not to think that Sosa and Stroud are in some way talking past one another.

8.4 The Dialectic Between Sosa and Stroud Revisited

I want to now suggest what I think the crux of the impasse is and how, once suitably appreciated, the situation should not be especially problematic for Sosa’s bi-level picture. If my interpretation of this debate is correct, we can trace the misunderstanding between Sosa and Stroud, which was reflected in the previous section, to the fact that they are thinking in very different ways about the relationship between indirect realism and externalism in the epistemology of perception. In the paper that Stroud principally takes issue with—viz., “Reflective Knowledge in the Best Circles” (reprinted in Sosa 2009, 178–210)—Sosa opens with an attack of the kind of indirect realist strategy in the epistemology of perception that he attributes to G.E. Moore, and which aims to vindicate external world perceptual knowledge as based on inference from information about our experiences. Sosa’s argument against this kind of strategy takes the form of a dilemma: the relevant inference can’t be deductive because “experience prompts but doesn’t entail the truth of its corresponding beliefs” (ibid, 179). But an abductive strategy does no better. This is because on such a strategy we (i) “restrict ourselves to data about qualitative character of our own sensory experience” (ibid, 179) (ii) “view belief in a commonsensical external world as a theory best postulated to explain course of our experience” (ibid, 179). The problem is that if we really do restrict ourselves to just data about the qualitative character of our experiences and nothing else (without presupposing the external reality to be inferred), then it’s unclear how the external world hypothesis will beat competitor explanations (as it must do for such an abductive strategy to succeed). It is against the background of rejecting on the basis of the above dilemma the indirect realist approach that Sosa presents his own externalist strategy for vindicating external world perceptual knowledge as a favourable alternative, one on which there is claimed to be a reliable sensory basis for understanding how our perceptual beliefs can constitute knowledge.

Now—and here is where I think the crux of the dispute between Sosa and Stroud lies—notice that Sosa’s own position does have in common one thing with the indirect realist approach he criticises and which he maintains falls prey to the dilemma he raises. And that is that on Sosa’s view, no less than on an indirect realist view, there’s a sense in which deliverances of perception are limited in a certain very general way. Let P be the proposition that the wall is red and let ‘E’ be what is available for conscious inspection whenever one has the experience of seeing a red wall. Both the indirect realist as well as Sosa allow that the “deliverances of perception even at its best are limited to the character of one’s perceptual experiences” (Stroud 2004, 172); and so both in this respect accept that when we know a proposition like P, the deliverances of perception are limited to E.

Sosa, as we’ve seen, takes this commitment to be problematic for the indirect realist specifically because the indirect realist’s strategy for vindicating external world knowledge (e.g., of a proposition like P) involves reasoning from E. Sosa’s strategy for vindicating external world knowledge does not invoke any such reasoning from E. However, Stroud insists that, even so, Sosa’s strategy

still leaves us with something that is epistemically prior to any knowledge of an independent world. If there are no reliable connections between the perceptual experiences we receive and the world we believe in as a result of them, we know nothing of the wider world even though we know what experiences we are having (2004, 172).

I noted previously that I think the impasse between Sosa and Stroud is due principally to the fact that they are thinking in very different ways about the relationship between indirect realism and externalism in the epistemology of perception. We’re now in a position, I think, to see why this is.

Just consider that, according to Stroud, it’s a commitment that Sosa shares with the indirect realist that is Sosa’s undoing—a commitment to accepting that what perception furnishes us with is (as Stroud puts it) something that’s “epistemically prior to knowledge of an independent world” (2004, 172). Now we get to the central point. For an externalist like Sosa, there is no meaningful sense in which whatever perception furnishes us with is epistemically prior to knowledge of an external world, even if it is obviously in some way temporally or metaphysically prior.

To appreciate this point, it’s helpful to consider how Sosa has put things in more recent work, where he emphasises that when one comes to possess animal knowledge, as one does in the case of simple perception, one exercises a competence. Crucially for Sosa, though, it is exactly the manifestation of that competence in the correctness of a belief that “thereby constitutes a bit of knowledge” (2017, 141). One’s experiences might seem a certain way while one is exercising that competence, but it’s the exercising of the competence, not the seeming, that explains the knowledge:

what happens when we manage to open the lid and look inside. Now we may immediately know the answer to our question, with a perceptual belief—say, that there is a necklace in the box—which manifests certain cognitive competences for gaining visual experience and belief. Perhaps this complex, knowledge-constitutive competence first leads to things seeming perceptually a certain way, and eventually to the belief that things are indeed that way, absent contrary indications. A belief manifesting such a competence and, crucially, one whose correctness manifests such a competence, does constitute knowledge, at a minimum animal knowledge, perhaps even full-fledged knowledge (including a reflective component) (2017, 141, my italics).Footnote 12

If perceptual seemings really were epistemically prior to knowledge, on Sosa’s view, the manifestation of a competence in a correct belief could hardly constitute knowledge. And this is so even though the view that the manifestation of a competence in a correct belief constitutes knowledge is compatible with the view that we perceive at best only the character of our perceptual experiences, where the latter is irrelevant to the epistemological story Sosa’s externalism offers for how one comes to know. And the same is the case for Sosa’s story about how one knows that one knows, viz., when one attains reflective knowledge. Though reflective knowledge involves the exercise of a different kind of competence, it is constituted by the manifestation of (reflective) competence in a correct belief that one’s first-order belief would be apt.

With this diagnosis in hand, we can now better see why Sosa does not belabour discussion of whether on his externalism we perceive at best only the character of our perceptual experiences. It’s because this is a fact that is epistemically not significant on the kind of externalism he embraces. Now, as Stroud sees it, attributing to Sosa the view that on his externalism we perceive at best only the character of our perceptual experiences “seemed necessary to make sense of him as trying to answer the kind of question his ‘externalist’ theory is meant to answer” (2004, 172–3). My assessment is that this is because Stroud is reading Sosa as needing the character of perceptual experiences to be doing a certain bit of epistemic work in Sosa’s epistemology that Sosa (in short) does not view as work that needs done.

In sum, I’ve suggested in this section that a kind of circularity worry that Sosa envisaged might be raised against his view has been taken by Stroud and converted into what initially looked like an even more serious epistemological defect with virtue perspectivism. Closer inspection, I think, reveals that Sosa has a satisfactory response to this worry to make from within his own externalist epistemology. While the adequacy of Sosa’s response has been a source of dispute between Sosa and Stroud, this dispute itself rested on some miscommunication, where I’ve suggested that each has taken the other to have in mind a different conception between the (put broadly) the relationship between indirect realism and externalism in the epistemology of perception.

8.5 Reed on Rationalist Perspectivism and Virtue Perspectivism

According to Baron Reed (2012), Sosa’s response to Stroud is stronger than it needs to be, and appreciating why this is so reveals a hitherto unnoticed weak spot in virtue perspectivism, one that comes to the fore when different perspectives come into contact with one another. In such circumstances, Reed thinks, there are important limits to how far virtue perspectivism can go toward giving us what Reed calls a ‘resting place’ for our intellectual pursuits. Or, put another way, virtue perspectivism “does give us an answer, but it is to a question that we will continue to feel the need to ask” (2012, 286). In this respect, Reed hints to an aspect of Stroud’s objection to Sosa that is getting at something right, even if strictly speaking, Sosa is within his rights (as Reed thinks he is) to press back against Stroud’s claim that Sosa’s externalist can’t know that he knows his faculties are reliable when they are.

If Reed is right, then this has important implications for virtue perspectivism as an anti-sceptical strategy. As Sosa sees it, a boon to his two-tiered account of knowledge is that it enables him to respond to the sceptic in a way that is broadly analogous, structurally, to the way Descartes thinks we should respond to the sceptic. Descartes, like Sosa, adverts to a two-tiered, perspectival structure (for Descartes, the lower level is cognitio and the higher-level scientia), where the ascent from lower to higher knowledge marks an intellectually valuable transition attained through reflection on one’s epistemic position.

 

Rationalist Perspectivism

Virtue Perspectivism

1

Begin with foundational knowledge, via rational intuition—Viz., cognitio; does not require knowing first that rational intuition is reliable.

Begin with foundational knowledge, via perception—Viz., animal knowledge; does not require knowing first that perception is reliable.

2

Acquire more knowledge of the same kind, directed toward one’s epistemic position.

Acquire more knowledge of the same kind, directed toward one’s epistemic position.

3

Cognitio now becomes something more valuable—Viz., scientia—By being the object of a higher order perspective (one that features broad coherence).

Animal knowledge now becomes more valuable—Viz., reflective knowledge—by being the object of a higher order perspective (one that features broad coherence).

Descartes’ rationalist perspectivism, as an anti-sceptical project, is one that, if it succeeds (which of course is a big ‘if’) would secure a kind of (as Reed puts it) ‘intellectual stability’ that a thinker aspires to in the face of sceptical doubts. Now, as Sosa sees it, “In structure virtue perspectivism is […] Cartesian, though in content it is not” (2009, p. 194). If Reed is right, though, this structural parallel only goes so far, and this is because crucial to Descartes’ rationalist perspectivism as an anti-sceptical strategy is its rationalist content, in virtue of which a thinker not only can know one’s faculties (for Descartes, the operative faculty being clear and distinct perception) are reliable, but attains a perspective from which she can see this with certainty, beyond doubt.

8.6 Reply to Reed

The starting place for Reed’s critique of Sosa involves drawing attention to the fact that we should classify virtue perspectivism as a fallibilist rather than as an infallibilist theory of knowledge. It’s fallibilist because according to virtue perspectivism, a subject can know a proposition (e.g., that the wall is red) on the basis of some justificatory source (e.g., reliable perception) even though the subject could have had the very same justificatory source and yet fail to know that proposition (e.g., had a jokester tinkered with the lights so that the subject was looking at a white wall bathed in red light, rather than a red wall).Footnote 13

A key observation of Reed’s is that, on a fallibilist theory, knowledge does not suffice to put to rest doubts we might have, in the sense that it is possible to know something (fallibly) while maintaining the epistemic possibility of the denial of what we (fallibly) know. With this point in hand, Reed observes that there is (and has been) in fact a surprising level of disagreement amongst epistemologists about what, exactly, our basic intellectual faculties really are. For example:

Wilfrid Sellars includes introspection, perception, and memory. But surely that list is incomplete. Sosa would presumably add testimony and reason. Others—e.g., John Locke and Alvin Plantinga—would add a faculty of divine revelation. Yet others—the Logical Positivists, say—might want to subtract both revelation and reason. And Thomas Reid thought we have a faculty of ‘common sense,’ which gives us knowledge of his preferred philosophical principles. Which of these conflicting views is correct? (2012, 285)

Reed is of course right that not all of these views can be correct; in fact, at most one could be. Now—the point concerning doubt is this: if one (engaged in such a debate with others about which intellectual faculties are the basic ones and which are not) is such that her view happens to be the correct one, then Sosa’s view permits one to claim that one knows it is correct. And even so, she may legitimately wonder whether it is. Such knowledge, Reed thinks, fails to provide the “healthy sort of stability that Descartes was seeking” (2012, 286), and which one would attain only if one could see with certainty that one’s faculties gave one knowledge.

And so Reed’s position is that even when one makes the ascent from animal to reflective knowledge, one may still have legitimate doubts and thus a lack of intellectual stability that one would not have if making the analogous ascent from cognitio to scientia within the rationalist perspectivism model. Accordingly, then, there is a sense in which virtue perspectivism’s shared structure with rationalist perspectivism affords it with what are ultimately illusory anti-sceptical epistemic credentials. In response to Reed’s line of critique, I’d like to make one observation and then raise two criticisms. The observation concerns the shared spirit of Reed’s critique with Stroud’s. Both locate what they take to be an inadequacy (at least, by the lights of broadly internalist thinking) with the epistemic position of the reflective knower insofar as she is said to know her own faculties are reliable. Stroud takes the inadequacy he locates to call in to doubt whether the reflective knower has genuine knowledge that her faculties are reliable. Reed, by contrast, is not contesting this point but rather calling into doubt whether a theory of knowledge that permits one to count as knowing her faculties are reliable while being in the position of Sosa’s reflective knower is a theory that goes far enough toward meeting our intellectual needs. Given these differences, any response to Reed on behalf of the virtue perspectivist will have to take a very different shape, given that the critique itself targets a different kind of desiderata on a theory of knowledge. Having addressed this point, I’d like to now suggest how I think the virtue perspectivist might be in better shape than Reed has led us to believe.

Firstly, let us grant for the sake of argument Reed’s claims about the importance of quelling doubt as a key desideratum within the project of giving a theory of knowledge. Accordingly, let’s take for granted that if a subject is in the best epistemic position that a given theory of knowledge licenses, and yet legitimate doubt still persists, then the theory of knowledge has failed this desideratum. Now, even on such an assumption, Reed admits that not all disagreements Sosa’s reflective knower might find herself in, about what our basic faculties are, would be likely to incite any such doubt. As he puts it:

If the perceiver found herself only in this one disagreement, with the crystal ball gazer, she perhaps wouldn’t be too badly off. Crystal ball use doesn’t really have much to recommend it (2012, 285).

And it is in the context of this point that Reed draws our attention to more serious disputes in epistemology about what the basic faculties are, those that he thinks can legitimately leave Sosa’s reflective knower with doubts. This move, though, seems to be a double-edged sword, given that it invites the counterreply that ordinary perception is common to all of the lists of basic faculties embraced by, e.g., Sellars, Sosa, Locke, Plantinga, Reid, and the Logical Positivists. It’s accordingly not clear that the ordinary perceiver who attains reflective perceptual knowledge is going to be prompted to doubt in the face of disagreement.

A second point worth noting in response to Reed’s critique of virtue perspectivism concerns his view of the importance of quelling doubt within the project of giving a theory of knowledge. On one way of reading things, this desideratum should be interpreted as a kind of necessary requirement such that it will be failed on Sosa’s theory if the following situation is a metaphysical possibility: a thinker both (i) attains highest-grade knowledge; and (ii) doubts to any degree the reliability of her intellectual faculties. If this is indeed the way to think about the requirement on an account of knowledge, then virtue perspectivism fails it. But then, so arguably does rationalist perspectivism!

Here is the idea. Cartesian scientia can persist in two modes. In the ‘active mode’ one is engaging clear and distinct perception in taking the perspective one does on one’s intellectual faculties. Such engaged clear and distinct perception is plausibly incompatible with doubt in a way that reflective knowledge is not. But what happens to scientia when one’s mind is not so engaged? As Reed puts it, on the Cartesian picture:

The certainty he possesses while he is entertaining his clear and distinct perceptions remains even when he is no longer entertaining them but merely remembering that they were clearly and distinctly perceived (2012, 280).

Of course, if scientia were available only in the active mode, then the consequence would be a radical kind of ‘epistemic presentism’Footnote 14 that cedes all to the sceptic except during rare moments of engaged and self-directed clear and distinct perception. And so it’s natural that a plausible articulation of rationalist perspectivism will allow scientia to be sustained outwith the active mode, and so through the memory of specific clear and distinct perceptions. But whereas scientia in the active mode is plausibly incompatible with doubt, it seems that scientia when sustained merely though the memory of clear and distinct perceptions is not. At least, it is plausibly metaphysically possible that scientia sustained through memory of the relevant kind of clear and distinct perceptions be compresent with some degree of doubt.

What this all means, is that if the relevant kind of quelling of doubt Reed takes to be a desideratum on the project of giving a satisfactory theory of knowledge is failed on Sosa’s theory. This is because his theory allows for the metaphysical possibility that a thinker both attains highest-grade knowledge and doubts to any degree the reliability of her (relevant) intellectual faculties, then the same charge applies to rationalist perspectivism, and so there would be no basis for favouring the latter to the former.

A natural response to the above point would be for Reed to articulate the general doubt-related desideratum on an account of knowledge he is appealing to differently. Perhaps, rather than to say that an account of knowledge fails the doubt-related desideratum outright if it is a metaphysical possibility, on the account of knowledge, that a thinker both attains highest-grade knowledge and doubts to any degree the reliability of her intellectual faculties,Footnote 15 we might instead opt for something different. Perhaps better is the following: that for two accounts of knowledge, A and B, ceteris paribus, A is to be preferred to B if doubt is compresent with high-grade knowledge on A to a lesser extent than on B. With this kind of requirement, it might then be claimed that Sosa’s reflective knower will more often find herself in a position of doubt than will someone with Cartesian scientia. And therefore, as this line of thought goes, Sosa’s theory of knowledge lacks the resources that rationalist perspectivism does to quell sceptical doubts—and so (to conclude the argument) virtue perspectivism does not inherit the anti-sceptical import of rationalist perspectivism despite sharing its perspectival structure.

But retreating to this line is, I think, problematic for two reasons. For one thing, it’s not obvious that the reflective knower is comparatively more inclined to doubt than is one with scientia sustained through memory of past clear and distinct perception. For another—and this is a point I don’t think Reed has really addressed—it’s unclear just what the reflective knower who is lacking any such doubts is lacking, epistemically, in comparison with the thinker with scientia. And given that the circumstances of disagreement Reed points to about basic faculties are not contexts where perception itself has been or is inclined to be seriously called into doubt, it would seem as though perceptual knowledge that rises to the reflective level will de facto be knowledge that lacks any such doubts.

For these reasons, I think that even if we grant Reed that quelling doubts is an important aspect of an epistemological project, it’s not clear on closer inspection that virtue perspectivism is disadvantaged in comparison with rationalist perspectivism, or for that matter that virtue perspectivism falls short of giving us anything we should rightly expect an account of knowledge to provide.

8.7 Fumerton on Virtue Perspectivism and Coherence

The suggestion that a philosophically satisfying account of our knowledge of the external world should at least do some justice to internalist intuitions is one Sosa makes explicitly. And it’s evident that this concession is one he views himself as making not at the animal level, but at the second-order, reflective level. But, how exactly are we best to understand this concession?

As Sosa tells us, ‘broad coherence’ is a feature of the second-order perspective characteristic of a reflective knower. But the coherence at the second-order is not ‘untethered’ coherence, but coherence that arises from a suitable provision of first-order animal knowledge. But animal knowledge itself is knowledge the attainment of which is accounted for by Sosa on externalist lines that are not concessionary to the internalist.

The matter of how to understand the sense in which Sosa’s virtue perspectivism succeeds in doing justice to internalist intuitions seems to turn on how to think about the second-order perspective, with careful attention to how it incorporates the products of the first-order perspective.

To that end, let’s take as a starting point two concise statements Sosa offers for how the first- and second-order perspectives interact. In a very recent statement, Sosa has said of his bi-level picture that it

[…] allows the use of our basic foundational faculties in attaining a second-order assuring perspective. So we can use the animal knowledge that we attain through the exercise of such faculties; we can use such animal knowledge in the (proper, coherence-aimed) elaboration of the endorsing perspective. This endorsing perspective would be a proper awareness of our competences through whose exercise we can gain our first-order knowledge (2017, 45–46).

And previously, in Reflective Knowledge (1997b) he wrote

[…] reflective knowledge, while building on animal knowledge, goes beyond it precisely through integration in a more coherent framework. This is it achieved via an epistemic perspective within which the object-level animal beliefs may be seen as reliably based, and thus transmuted into reflective knowledge (2009, 75).

If Richard Fumerton’s (2004) read of things is right, the appearance of internalist concessions at the second-order of Sosa’s picture is specious. Fumerton’s thinking here—drawing originally from a well-known point due to Laurence BonJour (1985)—is that there are two fundamentally different ways one might think about how it is that coherence is justification-conferring. On one way of thinking about things, the fact that a given belief coheres with other beliefs in one’s doxastic system suffices to raise the epistemic status of the belief in question. Alternatively, and more demandingly, one might hold that the mere fact of a belief’s cohering with other beliefs in one’s system of beliefs does not alone confer justification upon (or otherwise raise the epistemic status of) the target belief unless the subject is aware of the fact that the belief coheres in this way. With this distinction in mind, Fumerton maintains that whatever boost to the epistemic status of a belief derives from the mere fact of its cohering with other beliefs would be one that is ‘intellectually unsatisfying’. Though it’s not clear from Sosa’s proposal that he opts for any sort of further awareness requirement, or indeed, how such a requirement is something he could meet in a principled way.

Fumerton accordingly sees for Sosa a kind of dilemma, according to which:

[…] coherence without access to coherence doesn’t do the job of giving us the sort of justification that would satisfy an internalist. Without access requirements to coherence, however, it’s not clear that we have given the internalist anything that would allow the internalist to view the internalism/externalism debate as a false dichotomy (2004, 81).

Fumerton is right that without access requirements to the kind of coherence that features for Sosa at the second-order, we likely won’t satisfy an internalist, or at least, an accessibilist internalist. However, let’s bear in mind the context of this criticism: virtue perspectivism does not aim at internalism, but at preserving some of the elements of a philosophical account of knowledge that the internalist values—something that pure ‘thermometer’ model reliablists (e.g., Armstrong 1973) are unable to do. A criticism according to which virtue perspectivism would not satisfy an internalist then misses the mark.

Secondly, there is an important sense in which a kind of access requirement really is satisfied in connection with the coherence one attains on virtue perspectivism at the second order. Indeed, when one transitions from animal knowledge that p to reflective knowledge that p, the kind of broad coherence that features at the second order furnishes the thinker with “a proper awareness of our competences through whose exercise we can gain our first-order knowledge” (Sosa 2017, 46)—and this is so even when, in transitioning from animal to reflective knowledge that p, one needn’t be aware that p coheres in the relevant way when it does.

A principal value of internalism is that good epistemic standing involves not only the obtaining of certain epistemically good-making properties of our beliefs, but that we should be aware of their obtaining. Broad coherence at the second-order helps to provide a thinker with such awareness of one’s good standing at the first order. This is accordingly a feature of virtue perspectivism that does justice to a key value of internalism in epistemology. And it needn’t require one to have access to facts about a given belief’s cohering when it does in order to do justice in this way.

A third line of response to Fumerton’s critique requires us to return with a more critical eye to his distinction (originally due to Bonjour) between two kinds of coherence, one that is especially internalist friendly in that it involves not only coherence but awareness of coherence, and the other which is not and does not. Might we have positive reason, when giving an account of knowledge, to part ways with the kind of specifically internalist thinking about coherence that Fumerton rightly suggests virtue perspectivism isn’t in a position to countenance? I think there is. Even more, one such argument to this effect can be extracted from one of Sosa’s (1985) early papers on the value of coherence, one that pre-dates his bi-level epistemology.Footnote 16 One of Sosa’s key insights in this early paper is that epistemic value of coherence is itself plausibly explained in terms of the value of reliability understood along externalist lines. In order to illuminate this idea, a thought experiment is presented, one involving a kind of ‘random’ world:

Let us suppose […] that beyond the causal regularity required at the mind/world interface there is no systematic and orderly depth either in the mind or in the world. Beyond a certain elementary level, in such a world coherent unity adds nothing to the likelihood of having the truth, and random scatter in our body of beliefs seems no less likely to get it right. How plausible is it to insist even for such a world that knowledge even of the mind/world interface is aided by the most elaborate possible webs both worldward and mindward? Surely it is very little plausible to suppose that such artificial and wholly false webs add anything at all to one’s knowledge of what is there knowable. This suggests that coherence has derivative and not fundamental status as a source of cognitive justification. It justifies in our world, or so we believe, in virtue of its reliability as a source of truth (1985, 20).

The above thought experiment is meant to function as a reductio against the thought that coherence confers justification upon a given belief in a way that is not derivative upon its reliability. And this reductio, in our present context, motivates a kind of dilemma for Fumerton. The first horn of the dilemma is that, if the epistemic value of coherence is itself plausibly explained in terms of the value of reliability understood along externalist lines, then it’s no problem for virtue perspectivism that “coherence without access to coherence doesn’t do the job of giving us the sort of justification that would satisfy an internalist” (Fumerton 2004, 81). And this is because, put simply, the value of reliable coherence is more fundamental than the value of coherence with access to coherence. The other horn of the dilemma for the critic of virtue perspectivism is to deny that epistemic value of coherence is itself plausibly explained in terms of the value of reliability understood along externalist lines, but then to account for how the kind of coherence one might attain in Sosa’s random world adds value to the knowledge one is able to acquire at the mind/world interface.

8.8 Concluding Remarks

Virtue perspectivism, much like rationalist perspectivism, is an account of the nature of knowledge that emerges in direct response to questions about the very possibility of knowledge. Sosa’s virtue perspectivism has changed in some of its peripheral details over the years, but the basic structure of the proposal, and in particular its anti-sceptical strategy, has remained the same: virtue perspectivism allows the use of our basic foundational faculties in attaining not only first-order animal knowledge without first knowing these faculties to be reliable, but also what Sosa calls a second-order assuring perspective, one whereby we can appreciate those first-order faculties as reliable and in doing so place our first-order knowledge in a competent second-order perspective.

Is this problematically circular? And, relatedly, is it even possible to vindicate the circularity that does seem to feature in the proposal as benign while at the same time doing justice to internalist intuitions in any meaningful sense? Can this even be done by a foundationalist proposal that builds its entire edifice not on the direct apprehension of anything that is ‘given’ in experience but rather by beliefs appropriately caused? Engaging seriously with such questions leads us, almost immediately, to the most fundamental questions in epistemological theory involving circularity, scepticism, doubt, and assurance. This essay has attempted to navigate at least some of these issues, with a focus on what I think are three especially rich criticisms of virtue perspectivism raised by Stroud, Reed, and Fumerton respectively. The conclusion reached is that virtue perspectivism can ultimately withstand these criticisms. Showing why this is the case involves (among other things) a careful engagement not only with the general dispute between internalists and externalists, but also with questions about what a philosophical theory of knowledge should be expected to do.