FELIX CULPA: LUCK IN ETHICS AND EPISTEMOLOGY

FELIX CULPA: LUCK IN ETHICS AND EPISTEMOLOGY

GUY AXTELL

Abstract. Luck threatens in similar ways our conceptions of both moral and epistemic evaluation. This paper examines the problem of luck as a metaphilosophical problem spanning the division between subfields in philosophy. I first explore the analogies between ethical and epistemic luck by comparing influential attempts to expunge luck from our conceptions of agency in these two sub-fields. I then focus upon Duncan Pritchard’s challenge to the motivations underlying virtue epistemology, based specifically on its handling of the problem of epistemic luck. To respond to Pritchard’s challenge, I argue that (1) Consideration of the multi-fold nature of the problem of epistemic luck to an adequate account of human knowledge drives us to a mixed externalist epistemology; and (2) The virtue-theoretical approach presents a particularly advantageous way of framing and developing a mixed externalist epistemology.

Keywords: virtue epistemology; virtue theory; moral and epistemic luck; internalism and externalism; metaphilosophy.

Felix Culpa! (‘Oh, most fortunate fault!’)—St. Augustine

Luck has been a topic of philosophical interest dating back to the ancients. Aristotle’s Nichomachean Ethics gave qualified acceptance to the role of luck (tuchē) and “external goods” in attaining complete happiness, thereby retaining connections with the archaic roots of the term “eudaimonia,” which means literally having a good-spirit watching over oneself from birth. But following the Socratic tendencies in Aristotle’s own thought, the later Stoics found more inviting his claim that “to enlist to tuchē what is greatest and most noble would be quite inappropriate” (NE 1099b 10-25). [1] They insulated their accounts of moral agency from the perceived threat posed by luck by reducing the domain of legitimate moral judgment to the inner life of moral agents, emphasizing self-sufficiency and claiming that one’s complete happiness depends only upon oneself. Under Stoicism the virtues would shift from being the major components of happiness to being exhaustively constitutive of it; the “indifferents” or external goods were denied as constituents of complete happiness (Sherman and White, 2002).

The Stoics also tried to extend this barrier against luck into the realm of cognition, over-running along the way certain distinctions that Aristotle conceived between the virtues of character and the intellectual virtues. Nancy Sherman and Heath White, in “Intellectual Virtue: Emotions, Luck, and the Ancients,” tell us that in trying to make this parallel the Stoics conjectured that “just as the sage’s moral virtue is sufficient for his happiness, so too his epistemic virtue is sufficient for the truth.” There need be no luck involved in believing the truth. This requires, however, a conception of the knower as essentially infallible, holding that cognitive skills honed the right way are sufficient for truth, and “that if we are non-careless and non-precipitous, non-casual and non-random in our assent to appearances, we can be infallible in our grasp of the truth.”

We can easily discern from this how the temptation of a Stoic view of knowledge has been reflected in the internalist and deontological strains of modern epistemology since Descartes’ time. “On the Cartesian view as well as the Stoics’, the availability of an infallible means for discerning the truth places knowledge squarely within the sphere of our own control.” Or as Daniel Statman (1991) puts it rather more simply, “Epistemic justification’s immunity to luck is gained by viewing it as a deontological notion.” But over the past half-century philosophy has undergone a remarkable “naturalistic turn,” leading to the reassessment of many corner-stone assumptions of the modernist tradition. For the present purposes of this essay, the most relevant challenges are those that have led philosophers to reflect on the role of luck, and to re-examine the relationship between ethical and epistemic evaluation. I refer in particular to Thomas Nagel’s paper “Moral Luck” (1979) and Edmund Gettier’s “Is Justified True Belief Knowledge?” (1963).

Since Nagel framed the question of moral luck and indicated four distinct types—constitutive, circumstantial, causal and resultant—much effort has gone into refuting his muted acceptance that luck plays a role in our assessments of moral responsibility. Following instead basic Stoic and Kantian intuitions, many philosophers have deemed it unacceptable that moral differences between agents should be affected by luck. In epistemology, consideration of Gettier cases has strongly pushed towards objective, external constraints on knowledge that can prevent a justified, true belief from counting as an instance of knowledge if its truth is merely a coincidence and is not appropriately “linked” with its causal ground. This externalist or truth-conducivist understanding of epistemic justification has had far-reaching implications for how philosophers respond to the challenge of radical scepticism. Eschewing the circular guarantee upon which Descartes pinned his return out of deep subjective doubt, philosophers have largely given up on the idea that there is any necessary connection between subjective justification and truth. We have turned fallibilist, no longer basing our response to scepticism on our ability to find a method to secure certitude for all of our beliefs. If a true belief can be unjustified, and a justified belief can fail to be true, then we may need to take another direction in looking for conditions that mark knowledge off from (mere) true belief, a direction that gives greater (or even exclusive) place to factors of which the knowing agent may not be internally aware. In epistemology this externalist path is called a “truth-linked” analysis of knowledge, because it starts from the external, antecedently specified goal of maximizing interesting true beliefs and minimizing false ones. As Martha Nussbaum points out, the Greek tradition already contained resources for such an alternative means of subduing tuchē, for the identification of “something external, clear, and antecedently specifiable that counts as an end result” of an activity provides a clear measure of that activity’s success (2002, 165).

Andrew Latus helps to further clarify the connection between Nagel’s challenge in ethics and Gettier’s challenge in epistemology when he writes in “Moral and Epistemic Luck” (2000) that, “just as we are inclined to believe that luck cannot play a role in determining a person’s moral standing, so epistemologists have been inclined to think that luck cannot play a role in instances of knowledge…It is the incompatibility thesis that drives the antagonism between epistemological internalists and externalists.” When, for instance, Gettier-style counterexamples are routinely taken to falsify an author’s analysis, it is in large part dependent upon an unquestioned assumption of the incompatibility of luck with knowledge.

In the Gettier literature each writer tries to do to their opponent’s analysis of knowledge what Gettier is perceived to have done to the standard JTB analysis. Counterexamples using the Gettier technique of double-luck are constructed and routinely taken to falsify an author’s analysis of necessary and sufficient conditions on knowledge. Meeting ‘Gettier’s challenge’ has typically been presumed to mean identifying a ‘something extra,’ a ‘property x’ that together with true belief would provide a set of necessary and sufficient conditions for knowledge. But here a deeper question emerges about whether this is really what Gettier’s challenge should be taken to be. While the Gettier literature has been a battle ground for debate between epistemological externalists and internalists, epistemologists have not always been sufficiently self-critical about the manner in which the rules of engagement presupposed a strong thesis about the incompatibility of knowledge with luck.

We expect internalistic theories to come with a denial of luck, in that they understand justification in ways that demand internal access to the grounds of our beliefs and therefore presuppose a high degree of control over whether or not our beliefs are justified. Yet as has been pointed out by Mylan Engel and others, there has in fact been a strong “anti-luck” or “incompatibilist” characteristic to the Gettier literature on both sides of the internalist/externalist divide. Carolyn Morillo makes this point well by saying that “‘no accident,’ nomological analyses are characteristically externalist, naturalistic, and anti-Cartesian” in spirit, yet—evoking Nagel’s conflict between one’s self-image and one’s actual practices of evaluation—often present to us the author’s own too-Cartesian image of the human cognizer. If this is correct then perhaps, as she suggests, the approach to constructing and criticizing an analysis of knowledge (which gave rise to this now vast Gettier literature) has not been naturalistic enough. For our fallibilist intuitions took wing only after the demonstrative conception of knowledge was abandoned as ill-befitting human knowers. But fallibilist intuitions and those that gird the incompatibility thesis push in directly opposite directions. Thus when indefeasibility or truth-tracking replaces justification in forms of radical externalism, demand for the stability and reliability of our beliefs is again set unnaturally high.

The reservation that I here express is in line with the deep ambivalence that Latus gives voice to concerning the incompatibility thesis and the role that it has played in the countless attempts to meet Gettier’s challenge. Latus points out that this ambivalence has been widely felt of late, and helps explain why what we might call the ‘driving force’ of externalist views about knowledge and justification has never quite surmounted the ‘resistance’ of internalism. Indeed, as he suggests, the lesson that we should take from these debates is not the unbridled victory of the one side. Rather, failure on both sides to meet the demands that they commonly presuppose in conformance with some version of the incompatibility thesis has apparently worked to raise the appeal of scepticism.

This problem of how to give constructive meaning to Gettier’s challenge has been heightened by philosophers who have more recently taken notice that Gettier-type counter-examples appear to be frameable for “any fallibilist theory” (Harper 1996), or again for any analysis in which the proposed additional conditions are based on “defeasible probability statements” (Craig 1990), or “do not entail truth” (Zagzebski 1996).[2] If this is correct, then again surely our assumption that we meet Gettier’s challenge only by simple expulsion of epistemic luck from our analyses of knowledge needs to be re-assessed. For any account not subject to such problems, widely construed, will then become subject to a dilemma: Either it will be one that risks returning us an all-too-Cartesian image of the human cognizer, or else it will be one that risks setting standards for knowledge so high that we would all agree they can only rarely me met.

For a variety of reasons, then, the pursuit of an analysis of knowledge providing necessary and sufficient conditions on an “incompatibilist” basis has been charged not merely with becoming a trivial pursuit, but with becoming actually a dangerous one in the philosophical sense. Scepticism has always gained a foothold in the “gaps” between subjective justification and truth through which luck can flow (Axtell 2001); but by committing ourselves on principle to discounting luck and then failing to deliver, we—all of us, externalists and internalists alike—invite the sceptic in through a broader breach. We have as Latus vividly puts it, engaged in ‘a mug’s game’ of mutual refutation played by rules too accepting of a largely unexplicated thesis of the incompatibility of knowledge with luck. Since the most basic rule of this game is that the contrasting intuitions of externalists and internalists make them mutually-exclusive competitors over a single “correct” analysis of justification, the sceptic has had only to sit back bemused while the strongest of polemical objections and sceptical arguments are unfolded one against another by the contestants themselves.

I think there is some truth in this parody of the heyday of internalist/externalist debate, but in contrast I find philosophers doing something far more interesting in their analyses today. Some are challenging the import of abstract counter-examples to the adequacy of an analysis of knowledge, and some are questioning the role assigned to intuitions in adjudicating claims to knowledge. Many now validate the intuitions on both sides of the internalist/externalist debate, and are seeking for ways to move beyond it by constructing a consistent, “mixed” account. It is here that we can start to make connections with virtue epistemology. In the final section of this paper I will return to the thesis of the incompatibility of luck with knowledge, and the question of how various positions on the issue of epistemic luck parallel or differ from those treated in respect to the problem of luck in ethics. First, however, I want to further explore the topic of epistemic luck on its own terms, and to make a special case for a virtue-epistemic approach to the problem. I can better frame this approach by referring to another paper published in Metaphilosophy.

Duncan Pritchard (2003a), in “Virtue Epistemology and Epistemic Luck,” raises what he believes to be a fundamental difficulty for aretaic or virtue-based epistemological theories. While generally quite sympathetic to virtue theory, he wonders why a virtue-epistemic approach is necessary for responding to the problem of epistemic luck. There are different groups of virtue epistemologists with divergent accounts of warrant or justification, and “they are each focusing upon a different species of luck.” The ‘externalist’ group demands that we eliminate what Pritchard characterizes as “veritic epistemic luck” and the ‘internalist’ group insists also on eliminating “reflective epistemic luck.” But, he objects, eliminating veritic luck can be equally or better achieved outside of virtue epistemology, by modalized forms of process reliabilism, and “there is no fully adequate way of eliminating reflective epistemic luck.” Hence either way virtue epistemology’s key claim, that one cannot account for knowledge and other central epistemic concepts without appeal to the virtues, is itself a dicey one: “Far from motivating the adoption of virtue-based theories, consideration of the role of epistemic luck, in both its salient forms, actually reveals that virtue epistemology is curiously ill-motivated as it stands."

I must first readily acknowledge that Pritchard’s objections, if sound, hit home with me. For in an earlier article, “Epistemic Luck in Light of the Virtues,” I argued that a strong motivation for virtue epistemology emerges when we take epistemic luck (as Pritchard himself suggests we should), as a central epistemic concern. Can the problem of epistemic luck motivate a virtue-theoretical response, or does it instead fail to motivate and perhaps even undermine such an approach? To argue the affirmative side of this question against his negative, I restate the central argument of that earlier paper in two steps, and in the course of supporting it, will try to show it capable of repelling Pritchard’s challenge. The two step argument that I want to develop is the following:

1. Consideration of the multi-fold nature of the problem of epistemic luck to an adequate account of human knowledge drives us to a mixed externalist epistemology.

2. The virtue-theoretical approach presents a particularly advantageous way of framing and developing a mixed externalist epistemology.

I will develop these two claims in the ensuing two sections. Section 3 gives Pritchard’s challenge a further short, direct response, and to conclude I re-consider attitudes towards the problem of luck as presented in both its ethical and epistemological guises.

1. FROM THE PROBLEM OF LUCK TO MIXED EXTERNALISM

The Latin expression felix culpa derives from St. Augustine’s famous allusion to one unfortunate event, the Fall of Man, precipitating a second event of good auspice, the coming of the Redeemer. Aside from the high stakes in Augustine’s allusion, this displays a degree of formal symmetry with Gettier cases, which as Pritchard points out are also constructed from instances of “double-luck.’ In Gettier-cases an agent’s belief eventually comes out true, but purportedly fails to constitute knowledge because its truth is “lucky,” having been already cut off by the first instance of luck from the evidential basis that provided its justification for the agent.

Luck is “double” in another, quite different sense that Duncan Pritchard captures in saying that a key to understanding externalism and internalism is that they “are each focusing on a different species of epistemic luck.” As Wayne Riggs (1999, 463) similarly argued “both truth-conducivist and responsibilist conceptions of epistemic justification are directed at disallowing a particular kind of chance (true) belief from counting as epistemically sanctioned.” Several authors in recent years have made directed attempts to resolve the clash of intuitions involved in the externalist/internalist debate by delineating the different species of epistemic luck that inform their contrasting intuitions.[3] While there is certainly room to debate both how various forms of epistemic luck are to be distinguished and whether and when they are indeed knowledge-precluding, I generally find this approach to epistemic evaluation through the problem of luck illuminating, and agree with Pritchard that “both forms of epistemic luck need to be responded to in any adequate epistemological theory.” So we attempt to frame externalist and internalist intuitions about justification in terms of a specific kind of luck that each takes to be knowledge-precluding. On my own account, this task is made easier if we first attend to the “pure” or “simple” or “unmixed” forms of these theories, since these are the approaches most susceptible to sceptical challenge. I will use Pritchard’s terms of “Veritic” and “Reflective” luck for the sake of simplicity, though the discussion which follows differs in significant ways and we might just as easily substitute Vahid’s terms of “truth-oriented” and “justification-oriented” luck (2001). This susceptibility to sceptical challenge is easiest to see with the internalist view, so we can start by characterizing the type of luck that it sacrifices all else to exclude; we can then better understand how sceptical problems arise for epistemic internalism from another type of luck to which it consequently renders itself defenseless.

Focusing upon a zetetic context, a context concerned with the quality of an agent’s inquiry into truths as yet unknown to her, internalists emphasize a first-person perspective on epistemic evaluation. Following from this, they have been concerned to exclude reflective luck, or luck with respect to the upstream (or inputs) of a belief-forming cognitive process. Within the context of zetetic inquiry several concerns with luck can be raised that draw our attention to the agent’s motivational states (dispositions to believe) and information-gathering behavior.[4] In terms of Nagel’s four types, these would certainly include epistemic analogues of what he termed both “constitutive” and “circumstantial” luck. The problem of constitutive reflective luck points to the affective and motivation dimensions of cognition, and insisting that justification and hence knowledge depends upon an agent that is both properly motivated and properly affected by their cognitive environment. The problem of circumstantial reflective luck points to a demand that beliefs be “based upon” or propositions accepted “in light of” good epistemic reasons. Internalists want to eliminate from knowledge all cases such as where S truly believes p, but the belief is “not prompted by” S’s evidence. Pure internalists typically confine themselves to a strictly subjective sense of justification in which the evaluation of justifying factors is restricted to evidence available to the agent (and other beliefs she holds) at a given point in time.[5] This is because, following intuitions commonplace in ethics, internalists find the highest degree of luck in consequences, and hence want to restrict justification to factors one has the most control over or reflective “access” to. If my best defense against reflective luck is to base my belief in normatively “good reasons” of which I can be reflectively aware, she reasons, then we can exclude from knowledge all cases in which it is mere (reflective) luck that the belief is true, given the evidence available to me.

Yet in the broader context of zetetic inquiry it clearly also matters what evidence may be uncovered with further effort and inquiry, and why the agent has or doesn’t have the means to bring that “objective” evidence, including potential factual defeaters, to her attention. This possibility of objective defeaters to my belief appears from the subjective perspective on justification to be something of a matter of luck. If one concedes that objective justification is necessary, then such defeaters may exist to undermine the justification of my belief no matter how well I have conducted my own mind. This objective aspect of justification also raises important social dimensions of knowledge. Are available sources of information generally reliable? Are they reliable in a particular case? Has the subject ‘blinked’ information generally known within her community, or accepted evidence on false testimony or dubious authority? How do such factors affect the quality of the agent’s inquiries and our evaluations of whether or not her true belief constitutes knowledge?[6] A second and still broader gap for the emergence of luck accrues from noting that even beliefs that have these forms of justification are not necessarily true. For the internalist such questions about the “output” of the process must be left as radical contingencies. I can have everything right at my end, that is, in how well I comport my mind, but whether or not the world complies with the output of my best cognitive efforts must ultimately be deemed, from this first-person perspective, as “beyond our ken.” Hence veritic luck or luck of the output remains a big problem for any such account, as witnessed in the charge that the internalist analysis of knowledge still leaves us quite open to radical (or brain-in-a-vat) scepticism.

Focusing upon a reconstructive context for epistemology, on the other hand, the more “pure” forms of externalism such as simple reliabilism and counterfactual analyses take a third-person perspective on epistemic evaluation. Following from this they have been concerned to exclude the veritic luck characteristic of Gettier’s cases, a kind of luck analogous to Nagel’s “resultant” luck because it concerns the downstream (or outputs) of a belief-forming cognitive process. This context as we saw draws attention to serious objections to the JTB analysis of knowledge, and thorough-going externalists attempt to leave subjective justification entirely out of the analysis of knowledge. It is the fact of p, rather, that must help us to explain the belief that p; this third-person perspective provides for a reliably truth-conducive dependency or “link” between the knower and the object known. Knowledge is true belief that meets certain objective conditions such as counterfactual conditionals or having been produced through a reliably truth-conducive causal mechanism. This may even engender the replacing of justification conditions, placing it irrevocably beyond the agent’s ken as to whether or not she meets the conditions for having knowledge. Of course, the identification of cognitive faculties as reliable proceeds on a probabilistic basis, and is relative to the field over which the faculty ranges and the normal conditions in which it operates. One must therefore avoid committing the fallacy of division that would occur from attributing success to an agent in any particular instance based only on such general reliability of her faculties. Notice also that in such an externalist approach it does not matter how fortunate we are in the cognitive faculties we come equipped with or the dependability of the information or evidence that these faculties must rely upon as inputs in any particular instance. Such “upstream” matters, including any role played by an agent’s motivational states and evidence gathering and weighing are of no direct relevance in a pure externalist analysis of knowledge; looking at knowledge from a third-person perspective these “inner” intensional and qualitative aspects of a particular person’s cognitive agency must remain ‘beyond our ken.’ This is to say that as pure externalism positions itself to expel veritic luck, it also opens itself to reflective luck.

So consequently, externalists maintain that internalism allows too much room for the operation of veritic luck, and internalists maintain that externalism has the same fault with respect to reflective luck. Both, I would argue, are right, since what is revealed here is merely a result of the limitations of their respective starting points. As each approach seeks to exclude a particular form of epistemic luck, becomes susceptible to another kind. Little wonder, then, that Nagel viewed the general structure of the problem of luck as inherent in the tension between a subjective and an objective view of action. Statman and Latus both offers explanations of this susceptibility in terms of the logic of constructing counter-example cases in both ethics and epistemology: According to Latus, the sceptic has many options to find a foothold for luck in the process of belief formation, for “It is always possible to simply pass the entrance of luck down a step in the chain, from something like Nagel’s resultant luck, to causal luck, to circumstantial luck, to constitutive luck, wherein even one’s basic dispositions and traits of character can be presented as if acquired in some manner outside all personal control.”[7]

The upshot of our discussion, (here in line with Engel, Harper, Latus, Riggs, Vahid and Axtell (2001)), is that pure externalism and internalism are necessarily “incomplete” accounts of knowledge and justification. Along with the strengths of each “pure” externalist and internalist analysis of knowledge come certain inherent weaknesses, and tantamount among these weaknesses is their respective susceptibility to a kind of luck that their opponent believes to be knowledge-precluding. This is by no means a ‘mysterious’ outcome: That the strength of each account should appear the weakness of the other is just what we might expect if the conflict of intuitions is ultimately based more on divergent interests than on strictly incompatible logics.

But of course ‘susceptibility’ is not a term that either side would choose except to characterize their opponent’s position. In the rhetoric of the externalist/internalist debate, “acquiescence” would be a preferred term. For neither denies that it would be a ‘good thing,’ or ‘desirable’ to meet the conditions that the other places on justification, but, simply because their logic cannot allow it, they deny it as a valid condition on knowledge. Each presents the form of luck to which the other is susceptible as undermining, but the form to which their own account suffers as being ‘ineliminable’ and at any rate relatively innocuous; instead it is simply “part of what we might call the ‘epistemic condition,’” and something to which we should humbly acquiesce. But without further argument this logic is circular, since it merely reiterates their respective claims that we should accept their intuitions and reject those of their opponents. On the orientation we are urging, then, remarks such as these are typically part of the rhetoric of the externalist/internalist debate, and best understood as implicit concessions to the weaknesses a theorist’s own analysis.

On the other hand, recognizing that simple or pure externalism and internalism are each deeply susceptible to epistemic luck does not imply that the kinds of luck that they are respectively susceptible to are necessarily on epistemic equal footing. We must leave this as a matter of contention, requiring detailed argument for why and in what degree each is taken by their opponent to be knowledge-precluding. But the point I have argued (so far in only a negative fashion)—the advantages of a mixed externalist strategy over both pure externalism and internalism—does not require any such thesis of strict parity in their intuitions regarding epistemic luck. Each we have found is susceptible to a form of luck that has potential knowledge-precluding implications, and precisely because each is equally committed to the incompatibility of knowledge with luck, they forfeit any possible consensus on the best way to respond to scepticism. From these considerations I believe we can clearly infer the advantage, with respect to the conjoined problems of luck and scepticism, of a “mixed externalist” strategy. Such accounts are externalist in the standard sense that not all of the constraints they place on justification need to be internally accessible. The “hybrid character” of such accounts sets them apart both from internalist theories and from the more strongly-stated externalist theories in the contemporary literature.[8] In the next section we can begin to give a positive development of mixed externalist epistemology, and explain why an aretaic or virtue-theoretical approach presents a particularly advantageous way of developing it.

2. FROM MIXED EXTERNALISM TO VIRTUE EPISTEMOLOGY

In the introduction to their recent collection Virtue Epistemology: Essays on Epistemic Virtue and Responsibility, Abrol Fairweather and Linda Zagzebski (2001, 12-13) write,

It happens that the dispute between internalists and externalists can be framed nicely within a virtue-based framework. Since some virtue epistemologists maintain that both the causal history and efficacy of a person and her motivational states are important in conferring virtue, both internalist and externalist requirements must be satisfied in the possession of virtue. Rather than ending up with radically opposed epistemic theories, this form of VE [virtue epistemology] proposes a unified framework that admits the value of both criteria.

This passage illustrates a “mixed” externalist account of justification, sometimes also called a “dual component” account. This is one that integrates constraints on the reliability of an agent’s cognitive faculties with constraints on the agent’s zetetic responsibility and motivational states (or dispositions to believe). The above passage is qualified by saying that such a mixed account of justification is proposed in “this form of VE,” but it has in fact been a widely-shared view among self-described virtue epistemologists. Of course, most epistemologists today avoid the extremes of pure externalism and internalism by espousing some admixture of internally inaccessible and accessible constraints on justification. So I am not claiming that the mixed externalistic approach is a distinguishing characteristic of virtue epistemology. Our point is rather that the manner in which this approach is given philosophical and psychological grounding in the virtues makes VE’s defense of it unique and advantageous. We can begin to see this by noting that on the Aristotelian view, neither the motivation nor the success component is singly sufficient for the presence of virtue. Greco and Zagzebski (authors sharply contrasted in Pritchard’s essay) both point out that an Aristotelian account of the moral virtues combines considerations of inner motivation and objective success:

every virtue or excellence both brings into good condition the thing of which it is the excellence and makes the work of that thing be done well….Therefore, if this is true in every case, the virtue of man also will be the state of character which makes a man good and which makes him do his own work well (NE 1106a, 16-20).

The Aristotelian conception of virtue provides an integrative naturalistic ground for affirming the complementarity of reliability and responsibility constraints on justified belief. We can further illustrate how perceived advantages of mixed externalism are brought to light through an aretaic approach by adducing representative statements from a few of VE’s best-known authors.

Sosa: “Our virtue epistemology and virtue ethics focus…on the agent and cognizer. When the agent’s actions are said to be right and the cognizer’s beliefs to be knowledge, we speak implicitly of the virtues, practical or intellectual, seated in that subject.” “Knowledge requires not only internal justification or coherence and rationality, but also external warrant or aptness. We must be both in good internal order and in appropriate relation to the external world.”[9]

Greco: “The main idea [behind agent reliabilism] is to define knowledge in terms of virtuous cognitive character, and to define virtuous cognitive character in terms of proper motivation and reliable success. This takes care of the ‘no accident’ condition on knowledge, in that true belief which is formed through an agent’s reliable character is not an accident in any relevant sense. It takes care of the ‘subjective justification’ condition as well, since there is proper motivation, and as Aristotle would say, ‘the moving principle is within the agent.’ Roughly, a belief is both subjectively and objectively justified, in the sense required for knowledge, when it is produced by a properly motivated, reliable cognitive agent.”[10]

Zagzebski: “It is important to see that while the account of virtue in terms of motivation plus success is my own, the concept of virtue has almost always combined internally accessible and (potentially) internally inaccessible elements. So the blend of internal and external aspects is something that comes with the concept of virtue and that gives it an enormous advantage in epistemology, where it is becoming apparent that it is desirable to avoid both extreme externalism and extreme internalism.”[11]

Without obscuring real differences that exist among these authors, I would like to very briefly outline three features of mixed externalist virtue epistemology that are viewed as advantages by its proponents:

1. Integrating reliabilist and responsibilist constraints on justification. The turn towards externalism allows us to take the problem of scepticism seriously, while opening new avenues to support knowledge against the sceptical challenge. Grounding mixed externalism in the motivation and success components built into the concept of a virtue allows us to avoid the standard approach that ties these intuitions to mutually exclusive and exhaustive logics. It views these components instead as falling out from an analysis of intellectual virtues as relatively stable dispositions and habits of belief-formation. If something like Greco’s distinction is correct, between subjective and objective justification as each is related to a common aretaic-basing condition, then this will be a further advantage in reconciling the legitimate insights that externalists and internalists have about knowledge-precluding forms of luck.

2. Agent-basing, or turning to the agent as the seat of justification. The failure of process reliabilism to account for the serious problem of strange and fleeting processes is one way of indicating the need for agent-basing, that is, for taking the agent himself, in his stable dispositions, as, to use Sosa’s phrase, “the seat of justification.” Virtue epistemology and virtue ethics both affect such a ‘reversal in the direction of analysis’ (Greco), where evaluation of beliefs is tied in with the evaluation of the agents who hold them. As Sosa puts it, “To praise a performance as skillful or an action as right, or a judgment as wise or apt, accordingly, is to assess not only the action or the judgment, but also the reflected aptitude or character or intelligence. This is a distinctive view with versions both in epistemology and ethics….”(1997, 194).

3. Addressing the “value problem.” The difficulty of explaining what makes knowledge more valuable than true belief is what is called “the value problem.” Knowledge could not be epistemically better than mere true belief if true belief is the only epistemic good (DePaul 2001). The independent source of value or good needed to account for the superior value that we accord to knowledge over (mere) true belief is the agent herself (Greco, Riggs, Zagzebski). The virtues add to the subject’s worth as agent or cognizer, and their value is not purely that of instrumental means to right action or true belief. Knowing, as the above authors all generally agree, “has something to do with the agent getting credit for the truth, that she gets to the truth because of something about her as a knowing agent--her virtues or virtuous acts” (Zagzebski 2002).[12] Thus, the value problem suggests that we identify knowledge precluding luck with any kind that impugns the credit we would otherwise grant an agent for the truth of her belief.

Each of these tenets of a virtue-theoretic approach in epistemology can be viewed from the perspective of advantages in dealing with the problem of epistemic luck. Connections with the first two tenets should already be clear from our foregoing discussions. The agent-basing of epistemic evaluation provides an integrated and naturalistically sound way of balancing ‘first’ and ‘third-person’ constraints on knowledge. Zagzebski (2003b) has recently referred to this as the idea that knowledge is an “organic unity,” a unity of true beliefs arising from virtuous intellectual performances. Regarding the third tenet, Riggs rightly points out that “the problem of accounting for the value of knowledge and the problem of moral luck are conceptually intertwined. Each raises the issue of what degree of credit we are due when the outcome in question is due, to a large degree, to chance” (2002a). Of course, such connections need not imply that the problem of luck is more than one way of motivating mixed externalist virtue epistemology.

Yet virtue epistemologists are sometimes divided along lines between “virtue reliabilists” and “virtue responsibilists,” a division which brings into focus differences over issues such as the supervenience of normative properties, responsibility for character, and prospects for a ‘unified’ account of ethical and intellectual virtue. These philosophical differences are interesting, but Pritchard’s manner of defining the internalism/externalism distinction is suspect if it leads one to disregard the strong consensus about the advantages of mixed externalism that underlies these differences.[13] The problem in defending mixed externalism is not that concerns for faculty reliability are somehow incompatible with concerns for agent responsibility, but that the temptation to recast differences between sub-interests within VE in terms of the dichotomous categories of the older externalist/internalist debate works to obscure the common ground of these concerns. It is important, in particular, to avoid conflating virtue responsibilism with epistemic internalism in any of its standard senses, because the key constraints that virtue responsibilism places on justification regard motivation and agency, not “access.” Vrinda Dalmiya (2001, 232) makes this point well when she writes,

…when the knowing-self moves to center stage, epistemic evaluation, whether it is of beliefs or of character, cannot function within the constraints of a strict internalism. The relaxation of internalist criteria occurs on two fronts. First, consideration of reliability and success in achieving truth become relevant, and second, a social dimension is introduced to rupture the isolationism of purely ‘internal’ looks ‘within’…Epistemic responsibility now is not a function of either not violating epistemic obligations (deontology) nor of factors purely transparent to the knower (internalism).

Aside from helping us avoid the polarizing logic of the externalist/internalist debate, an additional and often overlooked advantage of mixed externalist virtue epistemology is the response to scepticism that it allows. If the failed attempts on the part of internalists and externalists for a non-circular and “fully general theory of knowledge” has only strengthened the sceptical challenge (Sosa 1999a), then a more ‘bootstrapping’ approach is called for. And a bootstrapping response, in turn, is strengthened to the extent that our judgments concerning the reliability of our cognitive faculties and the responsibility of our zetetic efforts can be shown not merely to be logically compatible, but actually mutually-supportive or complementary. Mixed accounts that are based specifically in the concept of intellectual virtues again seem uniquely well equipped to reveal the ground of this complementarity, and in so doing to support a claim of philosophical stability in the face of the sceptical challenge. An ideal epistemic agent has beliefs formed by reliable processes, is cognitively well-integrated because ‘properly affected’ or attuned to her environment, and seeks to acquire virtuous habits or dispositions both for their own sake and as a means of appropriately basing her reflective beliefs on good reasons.

3. A RESPONSE TO PRITCHARD: OPTIMISM ON (THIN) ICE?

Duncan Pritchard offers a keen analysis of how distinctions regarding epistemic luck inform epistemological accounts, including those of virtue epistemologists. His challenge to the motivations of virtue epistemology takes the form of a dilemma. Until it is met, Pritchard suggests that virtue epistemologists keep optimism regarding their approach to questions of knowledge “on ice.” The first horn of his dilemma is meant to pose a problem for those virtue epistemologists Pritchard identifies as reliabilists: “If veritic epistemic luck is the problem, then a safety-based account of knowledge is the answer….” The second horn is meant to pose a problem specifically for those virtue epistemologists, including Linda Zagzebski, who Pritchard wants to characterize as “internalists”: if reflective epistemic luck is legitimately a supplementary problem, “then a safety-based account of knowledge coupled with the demand for internalist justification is the answer, and, again, this need make no essential reference to the epistemic virtues.” My manner of responding to Pritchard is essentially an attempt to ‘grab the first horn,’ of the dilemma by showing that the problem that is supposed to accrue for the virtue reliabilist is a non-starter. I will not have space to address the second horn directly, but given what I have just said about the mistake of confusing virtue responsibilism with epistemic internalism, I believe that this second horn of the dilemma could be similarly rebutted.

Pritchard wants to argue that “the best accounts of knowledge available which respond to the problem of veritic epistemic luck make no essential mention of the epistemic virtues at all.” By “best accounts” he means those that have a modally-framed safety principle to preclude Gettier-type cases. Yet Ernest Sosa’s much-discussed safety principle and neo-Moorean response to scepticism (1999b)—both of which are shared by Pritchard—is closely integrated into his (Sosa’s) broader virtue epistemology. While Pritchard mentions at one point that Sosa employs a safety principle, he does not examine the connections that actually exists in Sosa’s thought between the safety principle and his virtue-basing conditions, nor does he offer any independent reason to think that Sosa’s employment of the principle is in some way inconsistent with commitments entailed by his broader virtue epistemology. Rather, he attempts to put a modalized version of process reliabilism into competition with agent reliabilism, and to argue that if veritic luck is the motivating problem then a modal “safety-based” theory can supplant both, since it alone completely eliminates it.

It may still be instructive to see why Sosa has held that ‘Cartesian’ or ‘C- tracking’ and the safety principle that he derives from it is only a necessary condition on knowledge, and one which—far from standing sufficiently on its own—entails essential reference to the virtues to be properly understood. For Sosa, “A belief is safe if and only if it is based on a reliable indication” (2000a). Left by themselves, tracking accounts reduce justification to “little more than a claim about that belief’s counterfactual relation to the truth of what is believed”: “Unaided, the tracking requirements proposed suffer from a sort of tunnel vision. They focus too narrowly on the particular target belief and its causal or counterfactual relations to the truth of its content.” This account is given fuller development in “Tracking, Competence, and Knowledge,” (2002), and here we find again that

The safety requirement is after all just a necessary condition. So the mere fact that a belief satisfies that requirement does not commit us to counting it a case of knowledge. On the contrary, whether or not one knows …will plausibly also depend also on why it is that one believes as one does. And here one must look to the ‘habit’ of thought that [guides one’s belief].

So Sosa’s fuller view, of which safety is one part, is what he terms a “tracking-though-a-virtue view”: For a proper understanding of tracking it is essential and not merely ancillary that it is the agent’s constitution and positioning vis-à-vis a fact p, and the field in which that fact obtains, that determines whether a subject will or will not track the truth as to p. Although we cannot take space here to evaluate all aspects of Sosa’s account, these arguments to the effect that a safety principle is necessary but not sufficient for knowledge directly answers the question that Pritchard’s asks as the first horn of his dilemma [‘If the problem of is veritic luck, then why isn’t a safety-based view the answer?’], and show that it does not present a serious problem to the motivations for reliabilist versions of virtue epistemology. There are, furthermore, other inter-related issues on which virtue epistemology sets itself apart, such as the “value problem” defined above, where it appears that an adequate response requires a turn to agent-basing. I would suggest that Pritchard’s decision to contrast his “safety-based” analysis with both reliabilism and virtue epistemology leaves him without resources to address this problem. But if the value problem is a serious one, then mustn’t what helps us to resolve it be present as part of the very conditions for knowledge?

Epistemologists have all too often written as if the particular feature on which they themselves focus excludes the insights of others and so exhausts the analysis of knowledge and justification. Virtue epistemologists can claim no general exemption from this common academic fault, but their culpability in it seems less than that of writers generally during the heyday of the internalist/externalist debate. Given our discussion in the previous sections, I hold it as erroneous to characterize the outstanding differences between Sosa, Greco and Zagzebski in terms of the division between externalist and internalist accounts of justification. Mixed externalism rejects the position that sees these groups as mutually-exclusive competitors over a single “right” concept of justification.[14]

This can lead us to compare Pritchard and Greco more directly. Pritchard holds that “One can regard the move from a simple process reliabilist view to either a modalized or agent-based reliabilist view as a move towards something like a safety-based account,” while Greco (1999) holds that the movement in epistemology towards more sophisticated forms of reliabilism can be regarded as a movement towards “agent reliabilism.” Aiming at minimalism, each is in this sense a “three-condition” epistemologist. Pritchard leans towards the view that true belief that meets the safety principle is the essence of knowing, and that a further demand for aretaic grounding of belief is therefore superfluous. Greco, conversely, advocates an aretaic grounding condition on knowledge, and since this “takes care of the ‘no accident’ condition on knowledge,” a distinct safety principle would seem to be superfluous. Ironically, in Putting Skeptics in Their Place, Greco does not claim sufficiency for the conditions he puts on knowledge. He stays with “S know p only if” throughout the book because he concedes there that his aretaic-grounding condition is “not adequate for addressing Gettier problems” (251). Sosa’s tracking principle is seen as one attempt to provide direct aid against Gettier problems, but remaining uncommitted to such a strategy. But in “Virtues in Epistemology” (2002b) he offers essentially the same set of condition, but now with “S has knowledge regarding p if and only if S believes the truth regarding p because S believes p out of intellectual virtue.”[15] This considerably stronger stance he supports by arguing that “agent reliabilism has the resources to address a wide range of ‘Gettier problems.’”

Neither of these “three-condition” analyses secures my confidence, and I must simply fall back on a rather mundane four conditions strategy, true belief plus both an aretaic-grounding condition and a specific no-luck condition. We can simply call this the DATA account, for its doxastic, alethic, tucheic and aretaic conditions. But this suggestion to combine some version of the conditions that, if I have read them correctly, Pritchard and Greco each independently take as sufficient for knowledge, will then seem redundant to both authors. For the one, a modal no-luck provision makes any reference to virtues or processes of the agent superfluous. For the other, a correct attribution of the grounding of one’s true belief in an intellectual virtue already implies non-accidentally. I must, however, leave defense of the alternative four-condition approach to another time.

4. CONCLUSIONS

In a “Postscript” (1993, 252) to his influential paper on moral luck, Bernard Williams writes that

…the resistance to luck is not an ambition gratuitously tacked on to morality: it is built into it, and that is why morality is inevitably open to skeptical doubts about its capacity to fulfill this ambition. In this respect there is an analogy to the idea (which I also accept) that there are intrinsic features of the concept of knowledge that invite skepticism…the concept of knowledge is itself involved in discounting luck.

Williams is here concurring not only with Nagel’s earlier analogy between the problem of moral luck and ongoing debates in epistemology, but also with the idea that sceptical arguments do not depend on the imposition of arbitrarily stringent standards of knowledge, but rather, as Nagel (1979, 467) put it, “appear to grow inevitably from the consistent application of ordinary standards.” Whether or not these two authors are correct in this point of mutual agreement, our discussions have contributed to showing how the problem of luck is by its nature a metaphilosophical one, because luck threatens in similar ways our conceptions of both moral and epistemic evaluation (Statman 1991).

Adequately addressing the problem of luck as a metaphilosophical problem—that is, as a problem that ranges across philosophic subfields—requires avoiding certain extremes that we have previously touched upon. We need to eschew both the Stoics’ over-brave synthesis of the good and the true, as well as the positivists’ prosaic dichotomy between moral and epistemic evaluation. As we have seen, carrying over into epistemology an essentially Stoic conception of ethical agency proves unrealistic by setting the bar for knowledge too high. We began with Sherman and White’s discussion of the connection between Stoicism and epistemic internalism, and should now to their own conclusion that humans are somewhat asymmetrically related to the goals of happiness and truth, since “happiness can be to a larger degree a matter of our own making”: “Though the Stoics themselves argue for a unified thesis that demands a parallel between truth and happiness, we suggest that at least on this issue, we do well to separate virtue epistemology from virtue ethics” (2002).

On the other hand, neither these authors’ greater deference to the role luck of the epistemic domain, nor our own call for re-examining the role of the “incompatibility thesis” in the analysis of knowledge, implies a radical lowering of the bar, or the claim that humans have to be content with doxa and eschew episteme. The “gaps” between what one believes to be true, relying on the available evidence available to one, and what is really true, have always been a motivating source of scepticism. But our discussion of how this problem is addressed within mixed externalist virtue epistemology has also pointed out its unique advantages for responding to sceptical arguments.

Author: Guy Axtell, Department of Philosophy/102, University of Nevada, Reno. Reno, NV 89557-0056. axtell@unr.nevada.edu

REFERENCES

Aristotle. 1980. Nichomachean Ethics. Translated by David Ross. Oxford: Oxford University Press.

Audi, Robert. 2000. “Philosophical Naturalism at the Turn of the Century.” Journal of Philosophical Research XXV: 27-46.

Axtell, Guy (ed.). 2000. Knowledge, Belief, and Character: Readings in Virtue Epistemology. Lanham, Maryland: Rowman & Littlefield.

Axtell, Guy. 2001. “Epistemic Luck in Light of the Virtues.” In Virtue Epistemology: Essays on Epistemic Virtue and Responsibility, edited by Abrol Fairweather and Linda Zagzebski, 158-177. Oxford: Oxford University Press.

Battaly, Heather. 2001. “Thin Concepts to the Rescue: Thinning the Concepts of Epistemic Justification and Intellectual Virtue.” In Fairweather and Zagzebski, 2001, 98-116.

Craig, Edward. 1990. Knowledge and the State of Nature. Oxford: Oxford University Press.

Dalmiya, Vrinda. 2001. “Knowing People.” In Knowledge, Truth, and Duty, edited by Matthias Steup, 221-234. Oxford: Oxford University Press.

Dancy, Jonathan. 2000. “Supervenience, Virtues and Consequences,” in Axtell (2000), 73-86.

DePaul, Michael. 2001. “Value Monism in Epistemology.” In Knowledge, Truth, and Duty, edited by Matthias Steup, Oxford: Oxford University Press.

DePaul, Michael. and Zagzebski, Linda. 2002. Intellectual Virtue: Perspectives From Ethics and Epistemology. Oxford: Oxford University Press.

Driver, Julia. 2000. “Moral and Epistemic Virtue,” in Axtell (2000), 123-134.

Elfin, Juli. 2000. “The Structure of Virtue Centered Epistemology.” Acta Analytica 15: 73-87.

Engel, Mylan. 1992. “Is Epistemic Luck Compatible with Knowledge?” Southern Journal of Philosophy XXX, no. 2: 59-75.

Fairweather, Abrol and Zagzebski, Linda. (eds.) 2001. Virtue Epistemology: Essays on Epistemic Virtue and Responsibility. Oxford: Oxford University Press.

Foley, Richard. 1987. The Theory of Epistemic Rationality. Cambridge, Mass.: Harvard University Press.

Gettier, Edmund. 1963. “Is Justified True Belief Knowledge?” Analysis 23: 121-123.

Greco, John. 1999. “Agent Reliabilism.” In Philosophical Perspectives 13: 273-296.

Greco, John. 2002a. “Knowledge as Credit for True Belief.” In DePaul and Zagzebski, (eds.) (2002), forthcoming.

Greco, J. 2002b. “Virtues in Epistemology.” In Paul Moser (ed.), The Oxford Handbook of Epistemology. Oxford: Oxford University Press.

Greco, John (ed.). 2003. Sosa and His Critics. Oxford: Basil Blackwell, forthcoming.

Harper, William. 1996. “Knowledge and Luck.” The Southern Journal of Philosophy XXXIV: 271-283.

Katzoff, Charlotte. 2001. “Epistemic Virtue and Epistemic Responsibility.” Dialectica 55, 2: 105-118.

Latus, Andrew. 2000. “Moral and Epistemic Luck.” Journal of Philosophical Research XXV: 149-172.

McGinn, Colin. 1984. “The Concept of Knowledge.” Midwest Studies in Philosophy, IX: 529-54.

Millgram, Elijah. (ed.) 2002. Varieties of Practical Reasoning. Cambridge, Mass.: MIT Press.

Morillo, Carolyn. 1984. “Epistemic Luck, Naturalistic Epistemology, and the Ecology of Knowledge.” Philosophical Studies 46: 109-29.

Moser, Paul. (ed.) 2002. Oxford Handbook of Epistemology. Oxford: Oxford University Press.

Nagel, Thomas. 1979. “Moral Luck,” in Mortal Questions. Cambridge: Cambridge University Press, 24-38.

Nussbaum, Martha. 2002. “A Science of Practical Reasoning,” in Elijah Millgram (ed.) (2002). Cambridge, Mass.: MIT Press.

Pritchard, D. 2002. “Recent Work on Radical Skepticism.” American Philosophical Quarterly, 39, 3: 215-257.

Pritchard, Duncan. 2003a. “Virtue Epistemology and Epistemic Luck.” Metaphilosophy 34, 1/2.

Pritchard, Duncan. 2003b. “Epistemic Luck.” Journal of Philosophical Research, 28, forthcoming.

Riggs, Wayne. 1998. “What are the ‘Chances’ of Being Justified?” The Monist, 81, 3: 452-472.

Riggs, Wayne. 2002a. “Reliability and the Value of Knowledge.” Philosophy and Phenomenological Research LXIV, 1, forthcoming.

Riggs, Wayne. 2002b. “Understanding ‘Virtue’ and the Virtue of Understanding.” In DePaul and Zagzebski (eds.) (2002), forthcoming.

Sherman, Nancy and White, Heath. 2002. “Intellectual Virtue: Emotions, Luck, and the Ancients.” In DePaul and Zagzebski (eds.) (2002), forthcoming.

Sosa, Ernest. 1997. “Reflective Knowledge in the Best Circles.” In Steup (2001), 187-203.

Sosa, Ernest. 1999a. “Philosophical Scepticism and Epistemic Circularity.” In Skepticism, edited by Keith DeRose and Ted Warfield, 93-114. Oxford: Oxford University Press.

Sosa, Ernest. 1999b. “How to Defeat Opposition to Moore.” Philosophical Perspectives 13: 141-153.

Sosa, Ernest. 2000a. “Skepticism and Contextualism.” In Philosophical Issues 10: 1-18.

Sosa, Ernest. 2000b. “Replies.” In Philosophical Issues 10: 38-42.

Sosa, Ernest. 2002. “Tracking, Competence, and Knowledge.” In The Oxford Handbook of Epistemology, edited by Paul. Moser. Oxford: Oxford University Press.

Sosa, Ernest 2003. ”The Place of Truth in Epistemology.” In DePaul and Zagzebski (eds.) (2002), forthcoming.

Statman, Daniel. 1991. “Moral and Epistemic Luck.” Ratio IV, 2: 146-156.

Statman, Daniel. 1993. Moral Luck. New York: SUNY Press.

Steup, Matthius. 2001a. “Epistemic Duty, Evidence, and Internality,” in Steup (ed.), (2001b), 134-148.

Steup, Mattius (ed.). 2001b. Knowledge, Truth, and Duty. Oxford: Oxford University Press.

Umbers, Richard. 2002. “Virtue Epistemology in Linda Zagzebski.” Unpublished Dissertation.

Unger, Peter. 1968. “An Analysis of Factual Knowledge.” The Journal of Philosophy LXV, no. 6: 157-170.

Vahid, Hamid. 2001a. “Knowledge and Varieties of Epistemic Luck.” Dialectica 55, 4: 351-362.

Walker, Margaret U. 1993. “Moral Luck and the Virtues of Impure Agency.” In Daniel Statman (ed.) (1993), 235-250.

Williams, Bernard. 1993. “Postscript.” In Daniel Statman (ed.) (1993), 251-258.

Zagzebski, Linda. 1994. “The Inescapability of Gettier Problems.” The Philosophical Quarterly, 44, no. 174: 65-73.

Zagzebski, Linda. 1996. Virtues of the Mind. Cambridge: Cambridge University Press.

Zagzebski, Linda. 2001. “Recovering Understanding.” In Matthius Steup (ed.) (2001), 235-252.

Zagzebski, Linda. 2003a. “The Search for the Source of Epistemic Good.” Metaphilosophy 34, 1/2.

Zagzebski, Linda. 2003b. “Epistemic Value Monism.” In John Greco (ed.) (2003). Oxford: Basil Blackwell, forthcoming.

Notes