Prudence, Rules, and Regulative Epistemology

Abstract

Following Ballantyne, we can distinguish between descriptive and regulative epistemology. Whereas descriptive epistemology analyzes epistemic categories such as knowledge, justified belief, or evidence, regulative epistemology attempts to guide our thinking. In this paper, we argue that regulative epistemologists should focus their attention on what we call epistemic prudence. Our argument proceeds as follows: First, we lay out an objection to virtue-based regulative epistemology that is analogous to the no-guidance objection to virtue ethics. According to this objection, virtue-based regulative epistemology cannot offer us useful guidance in our deliberations, because an abstract knowledge of virtue does not tell us what we should do here and now, especially in hard cases. We respond to this objection by showing that our making good epistemic decisions cannot simply be a matter of our following the right epistemic rules. In order to reliably inquire and deliberate well, we need epistemic prudence. Thus, while virtue-based regulative epistemology fails to determine how we should inquire and resolve deliberation here and now, this is also true of norm-based regulative epistemology. The upshot of this argument is that regulative epistemologists should focus their attention on understanding the nature of epistemic prudence and on understanding how we can promote its development in ourselves and others.

1. Introduction

Following Ballantyne [1], we can distinguish between descriptive and regulative epistemology. Roughly speaking, descriptive epistemology aims to conduct conceptual or metaphysical analyses of common epistemic categories. Descriptive epistemologists ask questions like “what is knowledge?” (or “justified belief”, or “evidence”). Regulative epistemology, on the other hand, aims to guide our thinking. Regulative epistemologists ask questions about how we should conduct inquiry, how we should respond to peer disagreement, or how we should deliberate about controversial issues.

We find an analogy to descriptive and regulative epistemology in ethics. Sometimes ethical philosophers attempt to conduct an analysis of common ethical categories. For instance, in meta-ethics, philosophers inquire whether we should understand the right in terms of the good or vice versa. However, oftentimes, ethicists want to provide us guidance when we deliberate about what to do. Normative and applied ethicists try to guide our deliberation when we make decisions about our lives. Just as normative and applied ethics are supposed to guide us in our deliberations about what to do, regulative epistemology is supposed to guide us in our deliberations about what to think.

Regulative epistemology can be divided into two broad approaches. On the one hand, there is norm- or rule-based regulative epistemology. Norm-based regulative epistemologists try to better our thinking by giving us norms or rules that can guide us in inquiry and deliberation1. On the other hand, there is virtue-based regulative epistemology. Virtue-based regulative epistemologists try to better our thinking by providing us with models of excellent thinkers and their traits2. Again, the distinction between norm-based and virtue-based regulative epistemology finds parallels in normative ethics. There, consequentialist and deontological ethicists attempt to provide rules to guide us in our deliberations about what to do. Virtue ethicists, on the other hand, attempt to guide our actions, not by providing rules, but by providing models of morally excellent individuals and their traits.

Given these parallels between regulative epistemology and normative ethics, we should expect some of the debates between different schools of normative ethics to find their analogue in regulative epistemology. In this paper, we are interested in one particular debate that has to do with prudence.

We can enter into this debate by considering an important criticism against virtue ethics. According to this criticism, virtue ethics is a non-starter when it comes to normative ethics because it is unable to offer us guidance when we need it. We need guidance from philosophical ethics, not in cases in which it is obvious how we should or should not act (e.g., should we be senselessly cruel?), but rather in the hard cases, when how we should act is not obvious (e.g., should we pay to off-set our carbon footprint?).

Virtue ethics tries to guide us in our decision making by providing a model of the ethically excellent person and her character traits. However, knowing what a virtuous person should do in idealized contexts does not help us figure out what to do here and now. Perhaps the just person will be concerned with promoting the common good. To know this is fine, but this does not allow us to know what justice demands of us in our particular circumstances and whether we should pay to off-set our carbon footprint (assuming that carbon off-set payments can be a way to avoid making real changes to our behavior). What we want to know is how we should act in our particular circumstances, and what justice demands of us here and now. A sketch of what the just man is like will not help us resolve this question [6] (p. 35).

The virtue ethicists’ response to such criticism will involve an appeal to prudence. Prudence is a character trait that allows us to make good decisions; it allows us to pick out the action that virtue demands of us in each situation. The virtue ethicist will allow that virtue ethics cannot provide guidelines for an individual that would serve as a substitute for having prudence. However, she will protest that this is equally true for any school of normative ethics. Ethics simply cannot be codified so that acting well could, in principle, be a matter of grasping the right rules and applying them. What one should do in a concrete situation cannot be determined by a rule; it cannot be determined by a decision procedure that virtuous and unvirtuous agents alike can apply. Rather, to know what one must do in a concrete situation depends ultimately upon a sensitivity to reasons; on a quasi-perceptual capacity to ascertain what is to be done. (See [6] (pp. 39–32), [7] (pp. 54–105), and especially [8]).

The critic of virtue ethics is likely to see in this appeal to a quasi-perceptual capacity a retreat to obscurantism. How exactly is this quasi-perceptual sensitivity to reasons supposed to work? What underlies this power of intuition, so that those who possess it can reliably latch on to what should be done here and now? If two individuals disagree as to what is to be done, how can we decide whose quasi-perceptual faculty is properly working? If these questions cannot be answered satisfactorily, then it seems that the appeal to prudence is analogous to the appeal to the Deity in God-of-the-gaps arguments. We cannot understand how we determine what it is we should do in concrete situations; we posit a faculty that supposedly explains how we have this ability. However, the explanation offered seems to shed no light on how we reliably decide what to do. Rather, it seems to be no more than a re-packaged affirmation that some people indeed have the ability.

We can call this objection to virtue ethics the no-guidance objection. This paper does not weigh in on the virtue ethicists’ response to the no-guidance objection. We only note that if the virtue ethicist is correct and no rules can be given that can guide us to the correct decision here and now, then the no-guidance objection gives little support to rule-based normative ethics. In this case, prudence may be mysterious. However, if we need prudence to reliably figure out what to do, and careful observation attests that at least some people can reliably know what to do, then this mysteriousness is not a reason to reject prudence; rather, it is a reason to explore the nature of prudence more deeply.

A version of the no-guidance objection can be applied to virtue-based approaches to regulative epistemology. Knowledge of the epistemic virtues in the abstract can give us the most limited guidance as to what we should think here and now about this particular topic in our particular context. Simply knowing the idealized dispositions of an excellent thinker, or simply knowing that, e.g., they love the truth and are open-minded, conscientious, and intellectually humble, will not help us determine how we should inquire and resolve our deliberations. What we need to know is what it is to be a lover of truth or open-minded, or conscientious, or intellectually humble in our particular circumstances, and what we seem to want instead of abstract ideas of virtue are applicable rules that would guide us in our particular case. If such rules do exist, then we should seek them out. That is, it seems that we should abandon virtue-based regulative epistemology in favor of norm-based regulative epistemology.

The burden of this paper is to show that successful inquiry and deliberation cannot simply be the fruit of following rules. Rather, if we are to reliably know how to proceed in our inquiry and deliberation, then we need something like prudence in the epistemic realm. Thus, regulative epistemologists should further examine what we call epistemic prudence and contribute to dispelling its elusive character.

Our paper will proceed as follows. In Section 2, working from Aristotle, we give a sketch of prudence. In brief, we hold that prudence is the disposition that allows us to make good decisions when what counts as a good decision is not determined by a rule. In Section 3, we argue that we make epistemic decisions. In Section 4, we offer two arguments that good epistemic decisions are not (in general) determined by rules. Finally, in Section 5 we conclude by considering the implications this has for regulative epistemology.

Certainly, we are not the only authors to argue for the necessity of epistemic prudence. In epistemology, Zagzebski [9] (pp. 220–234) is known for her claim that we need to appeal to prudence. However, there are two reasons why her account is not sufficient. Zagzebski’s argument presupposes that we should adopt a virtue-theoretic approach to epistemology. Furthermore, Zagzebski allows that the the prudent person’s intellectual comportment could be understood in terms of implicitly following rules (see [9], p. 221). This enables the no-guidance objection to remain in full force and undermine the case for prudence. Norm-based regulative epistemologists may allow that we need prudence in general to make good decisions, but if prudence is just a matter of implicitly following the right set of rules, then the best thing we can do as regulative epistemologists is to articulate these rules so that those lacking prudence may learn to apply them. However, our argument goes further than Zagzebski’s. We hold that good epistemic decisions are not the result of following rules either explicitly or implicitly. Because of this, we need epistemic prudence to reliably make good decisions.

2. What Is Prudence?

To a first approximation, prudence could be defined as the disposition to make good rational choices, that is, choices that bring our actions in line with rationality. For example, the prudent money saver decides where to invest her 30-year savings based on the best financial advice, rather than on eye-catching internet ads, or the testimony of unreliable money savers. Searching for an authoritative source of financial advice is acting prudently. To the extent that we choose prudently, we make our life more rational.

Prudence has been a topic of philosophical interest for centuries. Thus, Plato considered it a virtue of the soul and assigned it a central place in the domain of practical wisdom. Expanding on Plato’s idea, Aristotle characterized phrónēsis —straightforwardly translated as “wisdom” or “practical wisdom”—as a “truth-attaining intellectual quality concerned with doing and with the things that are good for human beings” (EN VI 5, 1140b 21, [10]). This is, of course, a general characterization, but some of its implications can be spelt out. By saying that phrónēsis is “truth-attaining”, Aristotle is not suggesting that it is a form of propositional belief whose truth value can be empirically evaluated. He is rather pointing to the larger context in which a good choice is normally underpinned by an apt exercise of the intellect, one that leverages what is best in the context of what is wise to do in practical contexts. The concern of phrónēsis is the identification and pursuit of life-enhancing ethical choices. Only in the context of “doing what’s best” for a rational agent can this singular sense of “truth” be meaningful.

In which sense is phrónēsis an “intellectual disposition”? Aristotle understands that while virtues like courage or magnanimity are rooted in the non-rational part of the soul, phrónēsis is rooted in the rational part. This means that the exercise of practical wisdom is first and foremost a rational exercise that informs the main virtues of character. Because phrónēsis is a virtuous exercise of our rational dispositions, doing what is best cannot simply be a matter of following our instincts.

To see this, suppose that an individual has a powerful hunch about the next National Lottery. She suspects what the winning lottery ticket number will be and ponders whether she should purchase it. At the same time, her knowledge of how lotteries work tells her that winning lottery tickets is the result of stochastic probability distributions that, like a die roll, are unpredictable. Because she knows it, she knows that her hunch cannot seriously pass as a valid prediction, and so, that she should disregard her hunch. Careful analysis could give this or any other individual additional reasons not to buy lottery tickets, such as the fact that big lottery prizes are easy money and that easy money often comes with strings attached, such as generating unrealistic expectations about success, or the possibility of becoming profligate or callous of others’ needs. All potential considerations regarding the value and consequences of buying or not buying a lottery ticket and the inclination to act accordingly are rational, and they are enabled by phrónēsis. At the same time, phrónēsis is a situated form of cognition, in the sense that it is rooted in an understanding of the agent’s variable context and is always contingent on what is possible to do here and now. Aristotle argues that the person who deliberates prudently “is the one who tends to aim, in accordance with his calculation, at the best of the goods for a human being that are achievable in action” (EN VI 7, 1141b, [10]).

Aristotle distinguished phrónēsis from two other intellectual virtues: scientific knowledge (epistêmê) and art (tékhnē). In which way is phrónēsis different from them? He described epistêmê as a héxis apodeiktikè, that is, a demonstrative ability. By following the syllogistic rules of reasoning, Aristotle believed that we can produce error-proof conclusions in a robust sense. His Organon describes how to arrive at these conclusions by following the determinate procedures of the rules of good inference. As a result, any true scientific proposition inherits its truth value from an inference that relies on prior and better-known premises that are either inferred from more basic premises or self-evident. So, any valid truth must deductively be linked with these premises to be considered scientific. On the other hand, prudence distinguishes itself from art (tékhnē), the creative ability. Since an artefact does not exist prior to its creation, art is the ability to bring about an intended outcome using pre-existing raw materials like the piece of bronze from which a statue is made. While art does not follow the rigid strictures of syllogistic deduction, how the end product is created in artistic production involves techniques and rule-based procedures that make the end product more likely. For instance, subjecting metal to high temperatures increases the likelihood that a bronze statue will be successfully shaped according to a design, while ignoring this procedure most often results in failure. This implies that art does not follow arbitrary norms or random procedures, yet it differs from phrónēsis in that its result is normally an artefact.

In the domain of epistêmê, the possible ways to bring about a specific outcome are determinate. Scientific knowledge is a good example of how a determinate procedure like observing the rules of inference can lead the individual to reaching a logically valid conclusion. Departing from valid premises, the individual is only required to follow the rules of syllogistic reasoning to make good science. In this context, if the starting point is correct, then the end is reached deterministically. While artistic creation can be expected to exhibit fewer rules, the techniques and procedures required to produce artefacts are not arbitrary; works of art are usually made on the basis of common conventions about style and good taste.

In Aristotle’s perspective, practical wisdom works unlike both scientific knowledge and art. This is because, in the domain of phrónēsis, the observation of rule-based procedures cannot guarantee that the agent will make a virtuous choice. This is because a virtuous choice may be informed by rule-based procedures, but the mere observation of these procedures will not yield a prudential outcome—a good ethical decision. Only by relying on the best exercise of her prudence, that is, the agent’s capacity for making a good choice, can the best decision be made.

To see this, consider the virtue of courage. Aristotle looks at the courageous person as the one who “endures and fears—and likewise is confident about—the right things, for the right reason, in the right way, and at the right time” (EN III 7, 1115 b5, [10]). According to his doctrine of the mean, acting courageously involves avoiding two contrary extremes: on the one hand, cowardice and on the other hand, foolhardiness. According to Aristotle, the coward experiences disproportionate fear in the face of danger and flees it. In so doing, she is not courageous. The foolhardy person, conversely, experiences no fear whatsoever in the face of danger. In so doing, she acts rashly and exposes herself and her comrades to unnecessary dangers. Aristotle argues that both extremes are far from the just mean between two mutually exclusive options. However, how is this “just mean” exactly determined? This is where virtue comes in. Aristotle is careful to point out that the just mean cannot be correctly identified without consideration of the particular circumstances of the individual (EN II 6, 1106a36–b7, [10]). If virtue is a rational choice relative to what is most practically wise (EN II 6, 1106b36–1107a 1, [10]), then the courageous action must be the result of a deliberation process intended to avoid cowardice and foolhardiness, but whose identification of the correct mean involves a careful analysis of circumstances.

The process by which the circumstances that surround the individual’s actions are judged in view of the intended goal is non-deterministic. As Kraut suggests:

“The arithmetic mean between 10 and 2 is 6, and this is so invariably, whatever is being counted. But the intermediate point that is chosen by an expert in any of the crafts will vary from one situation to another. There is no universal rule, for example, about how much food an athlete should eat, and it would be absurd to infer from the fact that 10 lbs. is too much and 2 lbs. too little for me that I should eat 6 lbs.”

[11]

In other words, knowing, as a general rule, that an athlete should ingest a maximum of 5000 calories a day and a minimum of 2500 does not tell her the number of calories she should eat at any given time. In some circumstances, the athlete may be advised to significantly reduce caloric consumption in order to lose weight as the most effective way to stay competitive. Whether she should or should not do it at any point in time is a prudential choice. Expecting general rules to give the athlete the right answer to whether she should eat more or less of her recommended calorie intake is illusory.

Prudence is not the result of the successful application of rule-following procedures. This is why some authors [12] (p. 81) have proposed that prudence consists of the correct application of the available information to a particular case. How does this application take place? Several authors have noted that Aristotle’s phrónēsis has a perceptual element (see for example [8], and [7] inter alia). As Roberts and Wood [3] (p. 306) note, for Aristotle, phrónēsis is both a “power of deliberation” as well as a “power of perception”. If this interpretation is correct, then the prudent person’s ability to discern what is to be done given all relevant information involves not merely ratiocination but a sensitivity to reasons.

This need not imply that rules play no role in making virtuous decisions for Aristotle (see [13]). On the one hand, Aristotle seems to allow that there are certain universal prohibitions. He argues that acts of adultery, theft, or murder are always vicious actions, and to choose to do these actions goes against right reason. In reasoning about what to do, a prudent individual may rely on these universal prohibitions to make up her mind. For example, because she believes that downloading copyrighted material from the internet without paying is a violation of justice and this is always wrong, a prudent individual might decide against doing it. On the other hand, a prudent individual might appeal to certain rules of thumb in deciding what to do. Thus, a courageous person might use the following rule of thumb: if overwhelmed by a far superior force, then flee. Of course, in neither sort of case do the rules determine what virtue demands one do. Negative rules only tell us what we should not do, rather than what we should do, and rules of thumb are rules that hold for the most part but not always. Nevertheless, the main point stands; the prudent agent will sometimes make use of rules in deciding what to do.

We have presented Aristotle’s phrónēsis as the rational virtue that enables us to make good ethical choices. This implies that Aristotle’s epistêmê—as he understands it—has no ultimate effect on the moral quality of the individual. In other words, having a better or worse disposition to make good logical inferences does not give the individual any particular moral quality—whether good or bad—or a better or worse disposition to make good or bad moral choices. It is only the end for which one’s logical abilities are made to work that determines this quality. We think that what we might call “epistemic prudence” is akin to epistêmê in this respect since most epistemic deliberations do not carry relevant moral consequences. Because of this, Aristotle’s phrónēsis cannot straightforwardly provide a basis for epistemic prudence. And yet, later developments in ancient Greek philosophy may provide it. Quoting Wolfson [14], Zagzebski [9] pointed out that the Stoics developed a concept of “assent” (synkatathesis) that refers both to the judgement of practical goods—as those embodied by the good ethical choice—as well as judgements of truth—as those embodied by true beliefs. Zagzebski argues that the Stoics’ concept of assent can substantiate the claim that prudence is also an epistemic virtue to be predicated on decisions about theories, propositions, and attitudes. We find this suggestion illuminating.

3. Epistemic Choice

If prudence is a rational virtue that allows us to make good ethical choices, then we view epistemic prudence as the application of the virtue of prudence to epistemic choices. Some philosophers might balk at the talk of “epistemic choice”. We hope to be able to show that these concerns are not justified. In our view, whenever we are confronted with an alternative, we are confronted with a choice. Choosing simply consists of picking out an option out of a range of possible alternatives. In this sense, it is clear that we often have to make epistemic choices, that these choices are familiarly involved in all sorts of doxastic deliberation. When we engage in doxastic deliberation, we are deciding what attitude we should hold towards the credibility of a given proposition. For instance: “What should I believe about the prospects of regulative epistemology?” “Can regulative epistemologists give good guidance to regular thinkers?” If so, “does this guidance cover the vast array of possibilities that present themselves to the individual?” There are several attitudes that I might take towards these questions: I might deny that epistemologists can give thinkers guidance, or that they can do so in a general way; I might suspend judgment about the matter; I might hold that epistemologists not only can but should be involved in giving regular thinkers guidance in a wide range of situations. If I am deliberating, then I am trying to figure out which attitude I should take out of a range of possible options. So, doxastic deliberation involves epistemic choice.

Epistemic deliberation concerning a proposition p involves a wide range of epistemic decisions that go beyond which attitude we should adopt towards p. Some of these choices have to do with how we should deliberate. Should we continue down a certain line of inquiry? Or would we be better off starting afresh? Some of these choices have to do with the norms or guidelines we accept to regiment our inquiry. What counts as evidence? What kinds of evidence legitimately settle the question? What reasons are there for keeping a question open as opposed to closing it? In response to each of these questions, there will be a range of options available to us; in inquiring, we must take on one of these.

While we affirm that we make epistemic decisions, we remain neutral about doxastic voluntarism. According to the doxastic voluntarist, our beliefs, at least some of the time, are under our voluntary control. While our making epistemic decisions implies that we have some control over our deliberating and our beliefs, it does not force us to accept that we have voluntary control over our beliefs. The control we have need not involve intentions or will. As a result, no one should take the talk about epistemic decisions to imply that we could form a belief merely by intending to do so.

In this paper, we do not assume any particular view about the nature of the control we have over our beliefs, the control that is implied by epistemic decisions. Instead, we merely note that regulative epistemologists, qua regulative epistemologists, must assume that we make epistemic decisions and that we can choose between competing beliefs or bodies of belief. Otherwise, it makes no sense to speak of offering guidance about inquiry and belief formation. If we cannot pick out our doxastic attitudes out of a range of alternative options, and if we cannot exercise control over the doxastic attitudes we adopt, then there is no work for regulative epistemologists to do.

4. The Need for Epistemic Prudence

To sum up the discussion up to this point, we see that prudence is a virtue that allows us to reliably make rational decisions whenever there are no determinate rules for making such decisions. Given that there are no determinate rules for choosing a virtuous action, it follows that we need prudence to reliably make virtuous choices in the ethical realm. Furthermore, we have argued that we regularly make epistemic decisions, particularly when we deliberate about what doxastic attitudes to adopt. The question we now turn to is whether we need epistemic prudence to reliably make good epistemic decisions.

In this section, we argue that, indeed, we do need epistemic prudence to reliably make good epistemic decisions. To make our case, we show that good epistemic decision making cannot be reduced to following determinate rules for adopting doxastic attitudes. As a result, we need epistemic prudence if we are to reliably make good epistemic decisions. However, before moving on to these arguments, we wish to clarify what it means for good decisions to be determined by rules.

As we are using the term, a good decision is determined by a rule, if there is a rule or a set of rules that provide a decision procedure, such that by following these rules one will thereby make a good decision. For the sake of simplicity, we can think of rules as conditional imperatives. These imperatives have an if–then structure. If certain conditions C are met, then one is to respond in a particular way R. For example, consider the following rule of etiquette:

If you are at a social engagement, then put your phone away.

This rule specifies a condition, namely, being at a social engagement, and it commands a response when such conditions hold, namely, to put one’s phone away. We can think of rules like this as potentially comprising a decision procedure when put together with other rules. Consider someone who is deliberating whether they should put their phone away. Suppose furthermore, counterfactually, that the rule of etiquette we have been considering is an absolute, indefeasible rule. Always and without exception, if you are at a social engagement, then you should put your phone away. This rule of etiquette could determine whether or not the individual should put away her phone (in conjunction with other rules), as per the following diagram (see Figure 1):

Philosophers have long noted that following a rule or a decision procedure requires more than just acting in conformity with the rule or decision procedure. Consider the following case: A member of an indigenous group living in voluntary isolation comes across a phone and claims it as her own. Suppose, furthermore, that for whatever reason, this individual puts her phone out of sight whenever she has a social engagement with others in her community (perhaps she is embarrassed to let other people know she possesses such an oddity). Even if she happens to be acting in conformity with the rule, such a person is not following the rule of etiquette “If you are at a social engagement, put your phone away”. In the second case, a young person is well aware of the rules of etiquette but could not care less about them. Nevertheless, because she is embarrassed to be seen with her outdated phone, she puts it away when she is with others. While in both cases the individual acts in conformity with the rule of etiquette, in neither case does the individual follow the rule.

The upshot of cases like these is that to follow a rule like the rule of etiquette above, one must be able to grasp the rule and grasp that the conditions specified by the rule obtain. Furthermore, one’s grasp of the rule and the relevant conditions obtaining must be the motivating reason for one’s acting in conformity with the rule [15] (p. 32). There is no need to suppose that one’s grasp of the rule is explicit or conscious. Without ever explicitly thinking it, one might grasp that it is against the rules of etiquette to have one’s phone out when one is with company, and so put one’s phone away whenever one is socially engaged. This unconscious grasp of the rules can still allow us to implicitly follow a rule, even if we are not explicitly following the rule.

Thus, when we argue that good epistemic choices are not determined by rules, what we are saying is that it is impossible that making good epistemic decisions could merely be a matter of grasping the relevant rules, grasping that the relevant conditions obtain, and thereby responding correctly. We must arrive at our good epistemic decisions some other way.

The first argument we consider is a regress argument. We start the regress argument by supposing that making good epistemic decisions could be reduced to following rules. We then consider someone who is deliberating as to whether she should settle her belief that p, or if she should continue the inquiry. To decide this question, the individual has a rule at her disposal of the following form:

If C, then R; else R*.

C states certain conditions, R is to settle her belief in favor of p, and R* is to apply some further rule. To follow the rule, our individual will need to grasp the rule (at least implicitly) and to grasp whether the specified conditions obtain (again at least implicitly). However, this forces on her a further epistemic decision: should she accept that conditions C obtain? Furthermore, it seems that it is an epistemic decision to take the original rule as a good rule. As a result, to decide these matters, our individual must follow some further epistemic rules of the form:

If C¹, then R¹; else R¹*. If C², then R²; else R²*.

However, our individual once again is confronted with the problem of deciding whether C¹ or C² obtain and whether either of these two new rules should be accepted. To make the right decision, there must be some further rules that our individual can follow. But of course, the same problem will occur: Our individual will now need to determine whether C³ and C⁴ obtain and whether the relevant rules themselves are to be accepted, and so on ad infinitum. However, it is impossible for an individual to explicitly follow an infinite number of rules. Thus, good epistemic decision making could not consist in merely following some set of rules. Rather, some good epistemic decisions need to be made without recourse to a rule to be able to follow epistemic rules.

Why does the regress in the argument occur? To follow an epistemic rule, one needs to make further epistemic decisions, which in turn gives rise to the need for further rules, which in turn requires further epistemic decisions and so on ad infinitum.

One might try to block the regress by positing basic beliefs, the acceptance of which is not under our control. Perhaps such beliefs are self-evident and so indubitable; perhaps such beliefs are presupposed by our very practice of inquiry and deliberation. Whatever the case may be, there is no epistemic decision to be made as to whether one is to have the relevant belief. Thus, according to this objection, there are certain conditions C that obtain, and we cannot but believe they obtain. Furthermore, there are certain correct epistemic principles of the form if C then R such that we cannot but accept them. From these basic, unchosen beliefs in the correctness of certain rules and certain conditions obtain, we can then derive enough epistemic rules to allow us to come to good epistemic decisions about every possible subject matter.

This response to the regress argument needs only to be stated to see that it is as futile as Descartes’s project of building up our current system of knowledge on the basis of indubitable truths. That there are any such indubitable truths is controversial. However, even if there are some indubitable truths, it is implausible that they are sufficient in number and scope to allow us to derive from them our whole body of current knowledge, so that by following indubitable rules we could move from indubitable truths to a perfectly rational representation of the world.

We can summarize our first argument against the possibility that making good epistemic decisions, in general, could be a matter of following rules as follows. Such a hypothesis either leads to an infinite regress of rules that one must follow to make good epistemic decisions, or it requires the possibility of a Cartesian-style reconstruction of our knowledge. However, it is impossible for us to explicitly follow an infinite number of rules, and Cartesian attempts to justify our current array of doxastic attitudes on the basis of indubitable truths are doomed. Thus, making good epistemic decisions could not simply be a matter of explicitly following epistemic rules3.

The second argument that good epistemic decisions are not determined by epistemic rules is the argument from the variability of reasons. According to the variability of reasons thesis, the same consideration that in one context is a reason in favor of holding a certain doxastic attitude, might not be in another. Indeed, in a different context, the same consideration might count as a reason against holding that doxastic attitude (see [17] Section 3).

Consider, for example, what attitude you should take towards the proposition that there is a pink table in front of you. Suppose, furthermore, that it appears to you that there is a pink table before you. In many contexts, appearing to see a pink table would be a good reason for believing that there is a pink table in front of you. However, in certain contexts, it might be a good reason to think that this table is not in front of you. (Suppose, for instance, that you have good reason for thinking that your visual experiences of color are hopelessly misleading; if it seems to you that there is a pink table in front of you, then there most likely is not a pink table in front of you, but a table of some other color [17].

Given the variability of reasons, that is, that reasons are context-dependent, we should expect the epistemic rules that we may follow in forming our doxastic attitudes to be defeasible. This is because epistemic rules seem to express the reason relation between certain states of affairs obtaining and having certain doxastic attitudes. Take for instance the following epistemic rule put forward by Nathan Ballantyne:

No Trespassing: If we are experts in one field and believe proposition p, and we recognize that (i) we lack another field’s evidence concerning p or (ii) we lack another field’s skills to evaluate the p-relevant evidence, then we should suspend judgment that p…

[1] (p. 207)

We can understand No Trespassing as expressing the idea that the recognition that we lack another field’s evidence concerning p or that we lack another field’s skills to evaluate the p-relevant evidence is a reason to suspend judgment about p.

Given the variability of reasons thesis, however, we should also expect a rule like No Trespassing to be defeasible. Sometimes, becoming aware that experts in another field have evidence bearing on something we believe is a reason to suspend judgment, but sometimes it is not, and this is exactly how Ballantyne presents matters. According to Ballantyne, we should suspend judgment if we recognize that either (i) or (ii) are obtained unless we have a “defense”. Ballantyne thinks that we have a defense whenever “we have reason to accept there are facts indicating we do not violate any epistemic norm by accepting p” [1] (p. 207). For instance, if we have reason to believe that the field whose evidence we lack “is ruled by degenerate or pseudoscientific research programs” [1] (p. 208), then we may be justified in holding onto our original belief, despite not having the evidence or skills provided by this field.

Can defeasible rules like No Trespassing determine for us what epistemic choice we should make? Consider the following decision procedure expressed by the indefeasible epistemic rule: if C, then R (see Figure 2).

Figure 2 represents an indefeasible decision procedure. Suppose that If C, then R is a defeasible rule. In this case, we will have the following sort of decision procedure chart (see Figure 3):

This is because even if we can ascertain that conditions C are obtained unproblematically, it is not clear whether we should adopt the doxastic attitude described in R or apply a different epistemic rule. If the rule is not defeated, then we should adopt the relevant epistemic attitude. However, if the rule is defeated, then we should apply a different epistemic rule.

One might hope to solve this problem by putting a no-defeater condition in the original epistemic rule. So, for instance, one might formulate the rule as:

If C, and there are no defeaters for the rule, then R.

For this new formulation, we have the following decision procedure (see Figure 4):

There are problems with this new formulation, however. Whether there are defeaters for a given rule in one’s particular context is not something that is unproblematically obvious. This is because whether or not one has a defeater will depend on the content of one’s total set of reasons. However, the contents of our total set of reasons are not transparent to us. We could not, even in principle, list out all the reasons we have for adopting doxastic attitudes. Nor can we pretend to have an ineffable grasp of our total set of reasons in its detail, so that we can unproblematically know if a rule is defeated or not. Rather, our beliefs about our total set of reasons depend upon further reasons. If I accept that there is an undefeated defeater for a rule in my total set of reasons, or if I accept that there is not, then I must do so for some reason; this is not self-evident.

But now, given the variability of reasons, defeasible epistemic rules cannot provide us with an indefeasible decision procedure. The rule that we add to decide whether our original rule is defeated will itself be defeasible. Thus, it will not give us a decision procedure for when we should think that the original rule is defeated or not. However, without a decision procedure to tell whether the original rule is defeated, the original rule gives us no decision procedure for deciding which doxastic attitude to adopt. Thus, since we are stuck with defeasible epistemic rules, no epistemic rule can determine which epistemic decisions we should make.

One might object that some reasons for adopting doxastic attitudes are invariable. For instance, factive reasons, such as seeing that p, always count in favor of believing that p. This is because one cannot have such a reason without p being true. Thus, the rule, “if you see that p, then believe that p”, and others like it are indefeasible epistemic rules. If we can have indefeasible rules, however, then the argument above fails.

In response to this objection, we note that first, accepting that one sees that p presupposes that one accepts that p. This is because seeing that p is factive. In this case, we cannot follow the rule “if you see that p, believe that p”. Recall that following a rule involves responding in a particular way because one grasps the relevant rule and the relevant conditions obtain. However, in this case, grasping that the relevant conditions obtain requires that one has already responded as the rule prescribes. This is because grasping that the relevant conditions obtain involves believing that p; however, believing that p is what the rule tells one to do. As a result, one’s belief that p in this case cannot be the result of following the indefeasible rule, “if you see that p, believe that p.” Rather, following the rule presupposes that one believes that p.

This is not to say that we cannot have factive reasons for our beliefs. Our point is that we should not understand our responsiveness to such reasons in terms of following a rule. In the view we prefer, what motivates us in adopting the relevant doxastic attitudes are the reasons themselves and not the grasp of some further rule apart from these reasons. For instance, suppose that an individual sees that p and, on this basis, forms the belief that p. In our view, seeing that p might provide a factive reason for that individual to believe that p. However, what motivates the individual to adopt the belief that p is simply her seeing that p. She is not motivated by the grasp of some further rule “if you see that p, then believe that p”, and her belief that she sees that p4.

Even if we grant that in some cases which doxastic attitude we should have is determined by an indefeasible rule, our argument above is unaffected. It remains true that, in general, our reasons for adopting an attitude are variable. In some contexts, they are reasons for belief, and in other cases they are not. Thus, in general, we will be forced to work with defeasible rules in forming our doxastic attitudes. It follows that for most of our doxastic attitudes, there are no epistemic rules that determine what epistemic decisions we should make.

It is worth noting that the arguments above do not depend on one explicitly following the rules. It is just as impossible for us to implicitly follow an infinite series of rules as it is for us to explicitly follow them. Likewise, defeasible rules will fail to provide us with a decision procedure whether we suppose that we are implicitly or explicitly following them. Thus, our arguments for the necessity of epistemic prudence go further than Zagzebski’s. While Zagzebski allows that prudence might be a matter of implicit rule following, we have shown that this is not the case. We need epistemic prudence, not just because our good epistemic decisions cannot be the mere result of explicitly following epistemic rules, but because they cannot be the mere result of following rules full stop.

5. Conclusions

In this paper, we have formulated an analogue of the no-guidance objection for virtue-based regulative epistemology. According to this objection, virtue-based regulative epistemology can offer us no guidance when we need it most, as abstract knowledge of the virtues gives us little insight into how we should inquire and resolve deliberation about the hard cases facing us here and now. The burden of this paper has been to deflate the no-guidance objection. Making good epistemic decisions is not a matter of just applying epistemic rules. In order to reliably make good epistemic decisions, we need epistemic prudence.

We have shown that mere abstract knowledge of intellectual virtues is no substitute for epistemic prudence, but neither is knowledge of whatever epistemic rules that norm-based regulative epistemologists can provide us. If we have succeeded in our arguments, then the no-guidance objection is no reason for favoring norm-based over virtue-based regulative epistemology.

The no-guidance objection and our response to it warrant a mitigated skepticism regarding the ultimate prospect of regulative epistemology. What we need to reliably make good epistemic decisions is epistemic prudence; neither virtue-based nor norm-based regulative epistemology can provide a substitute for this. Thus, regulative epistemology will let us down when it seems most needed: when it is hard to see how we should proceed in inquiry or how we should resolve deliberation.

However, just because regulative epistemology cannot guide us to a successful issue of deliberation regarding hard cases does not imply that regulative epistemology has no role to play. Regulative epistemologists might guide our thinking in broad ways; they might provide us with epistemic ideals we should aim for, or they might orient our thinking through the articulation of some defeasible rules. However, the guidance regulative epistemologists can offer is limited. Thus, while we might expect some guidance from regulative epistemology, we should not expect too much.

Furthermore, if our argument is correct, we take it that investigation into epistemic prudence should be at the center of regulative epistemology. To make epistemic prudence understood, we should confront these questions: How is it that we can have a disposition to reliably make good epistemic decisions? What underlies any quasi-perceptual sensitivity we have to epistemic reasons? How might we reasonably distinguish between real sensitivity to reasons as opposed to merely apparent sensitivity? Our inability to answer these pressing questions with regard to epistemic prudence might tempt one to embrace a different (and we suppose) less healthy kind of epistemological skepticism.

Moreover, if we need epistemic prudence to reliably make good epistemic decisions, it behooves regulative epistemologists to understand not only its nature but also how we can promote its development in ourselves in others. How can we become good epistemic decision makers, and help others do the same? This question seems to be at the heart of regulative epistemology. If the argument of this paper is correct, this amounts to the question: How can we become epistemically prudent5?