FAITH AND EPISTEMOLOGY

Abstract

I offer an epistemic framework for theorising about faith. I suggest that epistemic faith is a disposition to believe or infer according to particular methods, despite a kind of tendency to perceive an epistemic shortcoming in that method. Faith is unjustified, and issues into unjustified beliefs, when the apparent epistemic shortcomings are actual; it is justified when the epistemic worries are unfounded.

Virtuous faith is central to a great deal of epistemology. A rational agent will manifest faith in their perceptual abilities, in determining which experts and testifiers to trust, in their a priori reasoning, and in the epistemic capacities that are specific to their social environment. To ignore faith is to ignore a crucial element of our social and individualistic epistemic lives.

One exercises faith when one forms beliefs despite a kind of apparent epistemic shortcoming, which may or may not correspond to a genuine weakness in evidential support. For example, standing on a bridge one knows to be safe, despite one's natural but irrational fear, can manifest a kind of epistemic faith. So too can forming perceptual beliefs, or engaging in logical inferences, despite lacking a dialectically satisfying response to skeptical arguments. The same goes for beliefs that are informed by one's ideological stance – these too count as manifestations of faith, and under some circumstances, such faith is epistemically appropriate. One upshot of my project will be that an intuitively appealing neutrality ideal for education and discourse is untenable. I'll conclude with some discussion of practical questions about whether, when, and why it can be worthwhile engaging seriously with people who have radically opposed views and frameworks.

INTRODUCTION

Faith, a central topic in the philosophy of religion, rarely receives focus in mainstream epistemology. When it does, it is most often disparaged as a kind of wishful thinking. Skeptics about faith are right that faith is sometimes a kind of wishful thinking; exercises of faith can be epistemically problematic. But although faith can be misplaced, an exclusive focus on epistemically vicious faith would be as much a distortion as would be an exclusive focus on epistemically proper faith. In this paper, I'll offer an epistemic framework for theorising about faith. I'll suggest that epistemic faith is, to an approximation, a reliance upon certain epistemic procedures, despite their apparent epistemic shortcomings. Faith is unjustified, and issues into unjustified beliefs, when the apparent epistemic shortcomings really do undermine reasonable belief; it is justified when the epistemic worries are insufficient or unfounded.

Virtuous faith is central to a great deal of epistemology. Rational agents will manifest faith in their perceptual abilities, in determining which experts and testifiers to trust, in their a priori reasoning, and in the epistemic capacities that are specific to their social environments. To ignore faith is to ignore a crucial element of our social and individualistic epistemic lives.

One upshot of my project will be that an intuitively appealing ‘neutrality’ ideal for education and discourse is untenable. I'll conclude with some discussion of practical questions about whether, when, and why it can be worthwhile to engage seriously with people who have radically opposed views and frameworks.

1. FAITH AS GOING BEYOND THE EVIDENCE

Faith is often thought of as a kind of epistemic overreach. To exercise faith is, on one common way of thinking, to go beyond the evidence. Given this conception, it is not surprising that many writers – especially those with broadly naturalistic sympathies – consider faith to be an epistemic error. Archie Bunker famously quipped that ‘faith is something that you believe that nobody in his right mind would believe’, from which it deductively follows that nobody in their right mind has any faith. Sam Harris writes that ‘faith is generally nothing more than the permission religious people give one another to believe things strongly without evidence’.¹ John Locke embraced many of the central theological commitments of his culture's dominant religion, but he too was careful to characterise faith as that which goes beyond ‘reason’, and emphasised the epistemic priority of reason over faith in most matters.²

Although faith is widely associated with religion, the standard conception of faith is applicable in non-religious instances, too – along with the thought that it extends beyond the evidence and is suspect for that reason. One might manifest faith in one's spouse by ignoring evidence of infidelity.³ Søren Kierkegaard is among faith's advocates, but he holds that much of the value of faith is in its transcending of reason. The orthodox divorce between faith and evidence is endorsed by such unlikely allies as Richard Dawkins (faith is ‘belief that isn't based on evidence’⁴) and the author of Hebrews (faith is ‘the conviction of things not seen’⁵).

In her contemporary treatment of faith, Lara Buchak builds in the idea of risk:

A person has faith that X, expressed by A, … only if that person performs act A, and performing A constitutes taking a risk on X.⁶

It is natural to understand risk in terms of evidence – an action is risky to the extent the evidence is consistent with its going badly. So what makes an action risky is that the evidence doesn't fully support it. While I agree that faith can oppose or transcend the evidence, I think it's a mistake to suppose it must do so. In what follows I'll try to make this case that going beyond the evidence is not the only way to manifest faith. A central upshot will be that we must reject simple ideas about the relationship between epistemology and psychology.

2. CASES OF FAITH

Suppose you are starting a business, and considering including Ranjit as a business partner. Bringing in a business partner is a momentous decision – a poor choice will result in tremendous future stress and negative financial implications. You need a business partner who is honest, and diligent, and even-keeled, with whom you can comfortably and frankly discuss sensitive financial matters. Because you take these considerations seriously, you investigate Ranjit thoroughly. You talk to his former colleagues and partners. You check in with your own more experienced mentors, who provide you with long lists of things to check and consider. After a thorough process, you've gathered a lot of evidence, and – let's stipulate – that evidence rationalises proceeding with Ranjit as a business partner. We don't have to suppose that the evidence justifies certainty that he will work out well. But let's stipulate that you have enough evidence to justify the risk.

Having enough evidence doesn't always establish confidence. Especially when the costs of going wrong are high, humans are likely to experience what Jennifer Nagel has called epistemic anxiety. Epistemic anxiety is a feeling of uncertainty.⁷ Under the circumstances, you probably won't feel sure about whether you should go into business with Ranjit. But let's suppose you do it anyway. Good decision! (Remember, I've stipulated that your evidence was strong enough to rationalise partnering with him.) You've proceeded as the evidence dictated; you've done the rational thing. But because you overcame your own very natural doubts, it is easy and proper to describe this as an exercise of faith. You put faith in your process, in your ability to make rational decisions, and in Ranjit himself. This, even though it is what the evidence suggested you should do. So faith is consistent with responding properly to the evidence.⁸

The business partner case is similar to cases discussed by Buchak (2012) in favour of her own conception of faith, which links faith closely to the practical decision to decline to seek out further evidence. (Seeking out additional evidence is closely connected to epistemic anxiety.) But I don't think Buchak's approach can handle all cases of faith. As I mentioned above, it is essential to Buchak's view that faith involves risk. In the Ranjit case, there really was a risk that things would go terribly wrong, but that risk was small enough to justify proceeding. But there are similar cases in which there is no risk – cases in which one's evidence is conclusive against the bad possible outcomes. Consider for example a case Tamar Gendler has discussed, that of the Grand Canyon Skywalk.⁹

The Grand Canyon Skywalk is a U-shaped glass-floor bridge extending over the Grand Canyon. One can walk onto it and look thousands of feet down to the canyon below. It is extremely safe. It is the product of skilled engineers, at a famous site, in a litigious country. Hundreds of thousands of people have walked out onto it; none have fallen. The reliability of the Grand Canyon Skywalk is not like that of Ranjit, where the evidence merely makes it likely enough to rationalise taking the risk; here, the evidence available establishes that stepping out onto the Grand Canyon Skywalk will not send you hurtling to your death. There is no risk of falling through the glass.

Nevertheless, it would be extremely natural to approach the Grand Canyon Skywalk with epistemic anxiety. I myself am afraid of heights. I don't know whether I'd walk out onto that bridge. Certainly I wouldn't do so in a state of confident calm. Although I know it's safe, I would not feel like it's safe. Automatic, low-level psychological processes, over which I have at best minimal control, tell my body to panic when I look straight down and see rocks thousands of feet below.¹⁰ Suppose I walk out onto the bridge anyway. We can even suppose (wildly, I confess) that I walk out to the most extended bit and jump up and down with as much force as I can muster. If I do so, I'm putting faith in the bridge, even though there's no risk.

One might fuss with my characterisation that there's no risk to walking out on the bridge. One might deny that the evidence entails that the bridge won't collapse. Maybe all the safety records I just read about on the internet were fabricated. Maybe a saboteur will destroy the bridge just as I go out on it. This is controversial territory, and not the point I want to fight about. For the record, I do disagree with this line of thought; I think that I know that the bridge is safe, and that part of what it is to know something is for there to be no chance, consistent with my evidence, that it's not so.¹¹ But one needn't insist on this point. If one takes a more restrictive view of evidence, then practically everything we do has a risk of turning out disastrously. (Maybe an evil demon will strike down everyone who isn't on the Grand Canyon Skywalk!) Then the question of whether there is “some” risk of being wrong becomes deeply uninteresting.

There are also cases where, even by stricter standards for evidence, one has conclusive evidence but requires faith to rely on it. These are a bit more contrived, but they can be constructed. Suppose, for example, that our starship is under attack, and we are abandoning ship. I am tasked with assigning groups of people to life pods. The pods come in different sizes, and there aren't enough for everyone; some people are going to die, but I can minimise that number by assigning groups as efficiently as possible. Here comes a group I know to have thirteen subgroups of six people each. Should I send them to the medium pod, which can handle 80, or the large pod, which can handle 100? Thirteen times six. That's sixty plus three times six, which is eighteen, so seventy-eight. Fewer than eighty. So the medium pod is the best choice. Then again, this is a high-pressure situation, and I'm only so-so at arithmetic. I might've done the math wrong. I have three choices: send them to the medium pod now (they'll fit); send them to the large pod now (even if I made a mistake, they'll still fit – I don't make mistakes that big); wait and double-check the math (it's good to be sure; then again, that'd delay moving on to the next group, and seconds are lives). Under these circumstances, I think, it would be an exercise of faith if I sent them straightaway to the medium pod. I am putting faith in my arithmetical abilities. But those abilities are no part of the evidence – the evidence is that there are thirteen squads of six, combined with the background knowledge about the pods. The evidence mathematically entails that sending them to the medium pod is the best choice.¹²

For reasons like these, I disagree even with Daniel Howard-Snyder's suggestion that faith requires that one's evidence be “sub-optimal”.¹³ Even optimal evidential bodies leave open the kind of feeling of epistemic dissatisfaction to which faith can be the proper response. Optimality of evidence is an epistemic notion; epistemic anxiety is a psychological phenomenon.

So I do not think it's necessary for faith that there is a genuine risk of being wrong. Faith is a kind of confidence that can be manifested by the overcoming of felt riskiness, whether or not there is real risk. The examples of faith given in the previous section were those with genuine risk; the broader phenomenon extends to a broader feeling of epistemic dissatisfaction.

Here are some more examples that I think are helpfully explained by invoking a kind of faith. In §3 I'll make a more specific suggestion as to what it is that they have in common.

2.1 Modus ponens

Suppose I wonder whether Sandy's at home. I have enough information to answer my question. In particular, I know that Sandy is in Europe, and I also know that if Sandy is in Europe, then he isn't at home. A simple exercise of modus ponens would give the answer to my question: Sandy is not at home. Someone in this state will typically conclude that Sandy is not at home, but there's no guarantee that one will. We don't always infer all that we can.

It is not trivial to establish the rational status of modus ponens. One can give a soundness proof, pointing out that, given the truth conditions for conditional statements, any time a conditional is true and its antecedent is true, its conclusion must also be true. So reasoning according to modus ponens will never take one from true premises to a false conclusion. One can give such a proof, but the proof will involve liberal use of deductive reasoning, including modus ponens. Against a modus ponens skeptic, who is suspicious of this form of reasoning, any justification for modus ponens will beg the question.

In fact, as Lewis Carroll pointed out over a century ago, it's not even enough that one know or accept that modus ponens is reliable. This inference

1. 1. A

2. 2. A ⊃ B

3. 3. For all X, Y, if X and X ⊃ Y are both true, then Y is true

   ∴ B

is compelling only to one disposed to infer according to modus ponens, just as much as the simpler inference omitting the third premise is. Lewis's Tortoise nicely illustrates the possible gap between epistemology and psychology I am focusing on. This passage begins with Achilles speaking:

“And at last we've got to the end of this ideal race-course! Now that you accept A and B and C and D, of course you accept Z.”

“Do I?” said the Tortoise innocently. “Let's make that quite clear. I accept A and B and C and D. Suppose I still refused to accept Z?”

“Then Logic would force you to do it!” Achilles triumphantly replied. “Logic would tell you ‘You can't help yourself. Now that you've accepted A and B and C and D, you must accept Z!’ So you've no choice, you see.”¹⁴

Achilles's picture of the relationship between rationality and psychology is attractive, but naive. To the extent that people are sensitive to the rational facts, they are likely to accept what their evidence implies. But it is a contingent matter to what degree one is sensitive to the rational facts. We may grant that any time a conditional and its antecedent are true, its consequent will also be true; this still won't compel one to infer according to modus ponens. Logic won't force someone to infer according to modus ponens any more than it will force me to remain calm while standing over the Grand Canyon.

It is dissatisfying to find oneself in this position. We're taught in critical reasoning courses that we should find neutral ground with our dialectical opponents, and work outward from there. But the case of a modus ponens skeptic shows that there are possible cases lacking neutral ground on which to justify some of our epistemic commitments. This is, I think, just an epistemic fact of life.¹⁵ This doesn't imply that skepticism about modus ponens is correct. Instead, it implies that one can and should accept some forms of reasoning, such as modus ponens, even though one has no non-question-begging justification for them.

This too can involve a kind of faith. There is or can be a feeling of going out on a limb – one would feel more secure if one could establish on neutral grounds, shared by a skeptic, that modus ponens is a good practice. But what most of us do, upon recognising the situation, is to put our faith in modus ponens anyway. This is what we do, and it's what we should do. Suspending judgment on modus ponens for this reason would constitute an epistemically vicious lack of faith.

It follows that sometimes, manifesting faith is epistemically virtuous.

2.2 Reasoning and induction

What goes for modus ponens goes for reasoning more generally. In fact, the case is more simply and directly made. Suppose you are challenged by a reasons skeptic to justify the practice of reasoning. Trivially, you cannot do so without begging the question. To justify the practice would be to offer reasons; but whether there are any reasons is exactly what your skeptic is denying.

People who aren't skeptics about reasoning – i.e., pretty much everybody – put faith in reasoning anyway. It counts as faith because circular reasoning is uncomfortable, even when it is cogent.

In an exactly parallel way, David Hume famously challenged the justification for inductive reasoning – what reason is there to suppose that past observations are a good basis for future predictions? The best we could do, Hume suggested, would be to cite the observed fact that, in past instances, such inferences have tended to be pretty reliable. (After all, science has worked pretty well so far.) But this is circular reasoning – whether the past performance of an inferential pattern is a good guide to its future performance is just the sort of thing that is at issue.

One might give in to skepticism for this reason; or one might exercise faith in induction, trusting it despite having no “neutral” – non-question-begging – justification for it. The latter, would, I think, be an epistemically justified kind of faith.

2.3 Science and religion

According to legend, Cardinal Bellarmine declined Galileo's invitation to look through his telescope on the grounds that he already had, in the Bible and his religious tradition, a better way to discern astronomical facts. While Galileo might have cited the strong track record of scientific observation as a way to form beliefs about the nature of the world, Bellarmine could easily have replied that this argument was question-begging; for his view precisely was that his religious tradition has a better track record. Science has a stronger track record as measured against science; but Bellarmine's Christianity had the superior track record as measured against Bellarmine's Christianity. Each stance was a kind of faith. (This is not to say that each stance was equally rational. As I'll argue below, not all faith is equally rational.)

2.4 Alternative facts

This kind of faith-involving dialectical situation also obtains with respect to deeply contingent and empirical knowledge. We often make assumptions that are substantive and controversial. As a demographic/statistical matter, my present beliefs that the Earth is billions of years old, that human activity contributes substantially to climate change, and that the MMR vaccine is unrelated to the rise of autism diagnoses, are all reasonably controversial. These beliefs are very well-supported – I'm not being hasty in accepting them. I think most of the people who disagree with me are handling the publicly available evidence in an irrational way.

There is overwhelming evidence against the hypothesis that MMR vaccines cause autism. This is something I know, even though I am not an expert on medicine. I have only a vague idea about how medical researchers investigate such questions. Perhaps if I devoted myself to the study of epidemiology, I could eventually become able to cite evidence that makes belief in a connection between autism and vaccines irrational. But this would take quite a bit of time – and the evidence I'd cite would itself make heavy reliance on others’ work. As an autobiographical fact, I have not undergone this study. As a practical matter, I couldn't undergo such a study for all of my controversial beliefs. If faced with someone – one of the many such people in the actual world – who insists that the MMR vaccine does cause autism, I would be unable to explain, from a neutral point of view, why my view is better-supported than theirs.

Although my faith is more vivid in a case like this one where I don't actually have the medical evidence at my fingertips, the same dialectical situation, and need for faith, would occur even if I were an expert in epidemiology, ready to cite the actual studies that make my belief rational. For there's every reason to expect the motivated skeptic to refuse to admit some of the data as legitimate – they might say, for example, that any study funded by the pharmaceutical industry is likely to be biased in favour of vaccine safety. I'd stick to my belief anyway, and my opponent might well say, in the pejorative tone described earlier in this paper, that I'm just exhibiting faith in the medical community. They'd be right that I'm exercising faith, but wrong that I'm thereby irrational.

Richard Dawkins has argued that science is quite unlike paradigmatic instances of faith on this score, because of the public availability of the scientific evidence, writing that

the evidence that makes me believe in evolution is not only overwhelmingly strong; it is freely available to anyone who takes the trouble to read up on it. Anyone can study the same evidence that I have and presumably come to the same conclusion.¹⁶

Whether or not this is true of Dawkins, it is certainly not true of everyone. Most people do not have the time or the expertise to take advantage of the public evidence that would rationalise accepting the theory of evolution – even if we set aside worries about agreeing on what the evidence is. (If we do not set such worries aside, then no one at all can work it out for themselves. Even Dawkins must put faith in the scientific community to identify the facts about e.g. the fossil record.) Nevertheless, many of us do accept evolution, not because we've examined the evidence ourselves and concluded that it best supports that theory, but because we put our trust in the experts who do. We may or may not have anything pertinent to say to convince, say, a creationist, that our view is better-supported by the evidence than theirs. That doesn't make us waver.

This sort of dialectical situation is not a comfortable one; it smacks a bit of circular reasoning. Ordinarily, we feel better about our beliefs when they're ones we can convince others to share.¹⁷ So although the perceived epistemic shortcoming here takes a different form than in the cases above, I do think this kind of dialectical situation can induce epistemic anxiety – especially when we consider the fact that our dialectical opponents likely feel just the same way about us that we do about them. I'll consider more precisely the connection between anxiety and faith in §3.

2.5 Standpoint theory

According to feminist standpoint theorists, all knowledge is socially situated. Different social locations correspond to different epistemic abilities; the perspectives of members of marginalised communities, such as women and racial minorities, make available an enriched epistemic status that more privileged individuals are often unable to share. Standpoint theory involves both an epistemic claim and a methodological suggestion: since marginalised perspectives correspond to superior epistemic resources (the epistemic claim), they should be emphasised and prioritised in, e.g., scientific inquiry (the methodological suggestion).

Some philosophers have worried about an epistemic circularity involved in standpoint theory. Here is a statement of the worry by Helen Longino:

If genuine or better knowledge depends on the correct or a more correct standpoint, social theory is needed to ascertain which of these locations is the epistemologically privileged one. But in a standpoint epistemology, a standpoint is needed to justify such a theory. What is that standpoint and how do we identify it?¹⁸

Longino's worry is that standpoint theory has no non-circular resources for identifying privileged standpoints; one requires standpoint theory to establish the status of the theoretical claim that marginalised social positions can come along with superior epistemic capabilities. (It is also plausible that one may need to invoke standpoints to identify which perspectives are the marginalised ones. See the next section.)

Critics of standpoint theory have argued that this implies a kind of epistemic relativism. By the lights of one perspective, that very perspective is epistemically superior, but by the lights of another, it isn't. Standpoint theorists typically respond by attempting to articulate neutral grounds on which standpoint theory may be established.¹⁹ These attempts may well ultimately be successful. However, if the broad thought of this paper is correct, then this project may not be mandatory for standpoint theorists, at least insofar as they are seeking to form their own beliefs in the best way. Finding neutral ground on which to make one's claim is often important for convincing others, but it is not always epistemically required, as at least some of the cases above make clear. Perhaps an attitude of faith in standpoint theory could manifest epistemic virtue.

I shall return nearer the end of the paper to the relationship between these epistemic remarks and the more socially and politically complex realities that are involved in cooperation in society. (For now, it may be helpful to remember that whether you can reasonably believe something yourself is a different matter from what you can reasonably expect others coming from different perspectives to think.)

2.6 Perception and ideology

It is possible, through empirical training, to gain perceptual abilities. For example, some people have learned to recognise, immediately and directly, whether a given combination of tones constitutes an augmented seventh chord. Other people lack that perceptual ability; if they hear an augmented seventh, they either must mentally decompose it to work out what chord it is, or they're unable to identify it at all. If someone who does have the ability is challenged for their reasons for thinking a certain combination of sounds constitutes an augmented seventh, they might have nothing non-question-begging to say. This doesn't oblige them to suspend judgment. One might have faith in one's perceptual abilities, even in the face of a challenge that defies a neutral answer.

Few people are passionate about augmented sevenths, and most who are are able to recognise them. So the perceiver of augmented sevenths won't often encounter disputes of the sort I'm imagining. But exactly the same epistemic and psychological pattern applies equally to many cases in which many skeptics are deeply invested in their skepticism. This happens most conspicuously in beliefs that are tied up with competing ideologies.

I recently witnessed an exchange between an acquaintance and her white friend. He was dismissing her frustration over unarmed black men being killed by police officers. He described his own non-lethal encounters with police, and said that if people would just listen and comply with police officers’ instructions, they wouldn't be shot. My friend recognised these comments as the racist products of a white supremacist ideology. He – I'm sure you won't be surprised to hear – did not. He did not think his comment was racist. This is not at all strong evidence that he wasn't being racist, because he's someone who is terrible at recognising racism. In general, if you don't know how to recognise something, your belief that it's absent isn't going to be very strong evidence. And unlike people who are terrible at recognising augmented sevenths, people who are terrible at recognising racism often have powerful social and psychological incentives to maintain their ignorance.²⁰ In such cases, one's epistemic deficiency will be very difficult to rectify via anything like the kind of neutral, common-ground-emphasising discourse that features in informal reasoning textbooks.

Like most people, I often witness sexist actions. Not everyone who witnesses sexist actions recognises them as such; indeed, some people are deeply committed to the idea that contemporary sexism is an imaginary left-wing bogeyman. If I observe e.g. a contribution to rape culture in the form of victim-blaming, skeptics are both able and motivated to explain my observations away as hysteria, or self-serving sanctimony, or animus against particular men. Attempting to engage with such people without begging central questions can be – can be – as difficult as would be similar conversations with Cartesian skeptics about perception, or Humean skeptics about induction, or what have you. (They needn't always be so; I'll return to this question below.)

I occasionally witness – and very occasionally join – conversations among professional philosophers about whether there is sexism against women in professional philosophy. I take it to be obvious that there is; I've observed it firsthand. I've heard many stories from many credible people about their experiences. But there are people who are rather deeply committed to the denial of this reality. There are people who think that the recent high-profile cases of sexual harassment were either fabrications or innocent misunderstandings to which we've all hysterically overreacted; that the power-hungry feminists who run our profession decide which philosophers will be employed; that most of the stories posted at “What is it like to be a woman in philosophy?” are outright lies.²¹ If you are one of these people, of course I don't expect you to take my word for it that there is sexism against women in philosophy. The cases I take to be proof of sexism, you will take to be proof of a feminist conspiracy to create the illusion of sexism. Although you will not appreciate this example, I think and hope you can still be on board with much of the general framework of my paper.

Even if I manage to convince someone about an instance of a sexist or racist action – the sort of thing I'd consider to be strong support for my generally well-confirmed belief that sexism and racism are things – my dialectical opponents still have the option of allowing for exceptional cases that do not confirm a broader trend.²² Again, this is unsettling. Sometimes, disagreement of this sort will cause one to suspend judgment, even if their evidence was sufficient for knowledge – even if they began the conversation with knowledge. This amounts to gaslighting.

So even in controversial cases, one may possess perceptual abilities that nevertheless put one in contact with the truth, even though one is unable to provide a non-question-begging argument in their defense. For an extreme example of this point – one involving frank first-personal reflections on experiences of sexual assault – see Freedman (2010). Freedman – herself a survivor of sexual assault – describes the way that her experience has altered her general perception of the world. She now sees it as less just than she once did; she now feels less in control over her life than she once did. Freedman argues that the unpleasant lens through which she now sees the world constitutes an epistemic virtue. People who see the world as she does, she argues, are more accurate, and better justified. This, even though she's not in a position to cite the evidence that induces her to see things this way. (The traumatic nature of her experiences plays central roles here; she doesn't have clear memories about many of the details.)

3. A THEORY OF FAITH

What these last several examples of faith have in common is a reliance on one's belief-forming capacity, even absent an ability to justify it on neutral grounds. Reliance on contested grounds is discomforting; we prefer to be able to justify our beliefs in a neutrally recognisable way. So this kind of phenomenon too fits in with the cases discussed in §1, where one exercises faith by overcoming a (perhaps non-rational) psychological tendency to doubt. What seems to unify faith is confidence in spite of a temptation to epistemic anxiety. When I look down in terror at the Grand Canyon, I don't feel secure in my belief that the bridge is safe; when I notice that my justification for trusting deduction, or my eyes, or my ability to recognise racism, relies on rule-circular reasoning, I feel like I'm taking an epistemic risk. If I rely on my judgments anyway, I exercise faith. Suppose I observe someone chastising a victim of sexual assault for not being more careful at that party. Even if I can't explain what's wrong with the comment without begging questions, I may form the judgment that it's a sexist instance of victim-blaming. If so, I have exhibited faith in my ability to detect sexism.

What is the exact relationship between faith and an actual felt temptation to epistemic anxiety? To a considerable degree, this is a terminological question, not particularly central to my main project; however, it's useful to be a bit precise here, both for specificity and to make clear the space of options. Ordinarily, we'd expect the recognition that one has no non-question-begging justification to dispose one to anxiety; in such cases, I'd characterise faith as reliance on the methods in question despite this disposition. But what if the temptation is not present? Suppose someone is entirely unbothered by the observation that their view cannot be defended without begging central questions. There is some temptation to say that no faith is exercised in such a case – believing comes easy, so no faith is required. But there is also, I think, a reasonable case to be made that, if one doesn't even feel anxious, that can be a manifestation of an even greater form of faith. Suppose that after repeated exposure to the Grand Canyon Skywalk, I stop even feeling anxious, and do walk out (and even jump up and down) in a state of calm confidence; maybe this is a case of faith without even the feeling of anxiousness.²³ My own linguistic preference is to allow this latter, more inclusive, sense of ‘faith’, although I don't think I have much of a substantive dispute with someone who prefers to reserve the term for cases where one actually overcomes a feeling of anxiety. In other words, I think the condition in (1) is too strong:

1. (1) A subject manifests faith iff they rely on their epistemic judgments despite their felt temptation to epistemic anxiety.

If the temptation to doubt isn't necessary for faith, we may also ask which kinds of confident belief count as faith. Should we go so far as to say that any time one relies on a conclusive judgment, one manifests faith, regardless of whether one has or would be expected to have considered any kinds of skeptical pressure? Definition (2) has a certain attractive simplicity, but it posits a rather weak notion of faith:

1. (2) A subject manifests faith iff they rely on their epistemic judgments regardless of whether they feel tempted to epistemic anxiety.

It sounds a bit odd to say that any time someone makes an unreflective judgment, it must be an exercise of faith. Maybe manate Moorean epistemologists exhibit faith by affirming the presence of their hands in the face of skeptical arguments, but it's more of a stretch to extend that judgment to non-philosophers who've never even considered the possibility that perception could be radically misleading. And do we really want to say that ordinary, unchallenged perceivers of augmented sevenths are exercising faith in their perceptual abilities? Perhaps instead we should say that faith only happens in particular kinds of situations – perhaps ones where we'd typically expect someone to feel a tendency towards doubt. The latter fits better with the intuitive usage of the word “faith”, but it also invites further questions: just how should we understand this notion of “typical expectation”? We could try to index the notion to ordinary human psychology, as in definition (3):

1. (3) A subject manifests faith iff they rely on their epistemic judgments, despite the fact that a psychologically ordinary human would be likely to experience a felt temptation to epistemic anxiety.

But I am dubious that the notion of a “psychologically ordinary person” is robust enough to do the work needed. Individuals vary with respect to which kinds of apparent epistemic shortcomings they find troublesome. Furthermore, it seems like a mistake to characterise faith in general in terms of ordinary human psychology; we want our notion of faith to be applicable to hypothetical Martians, etc., too. So I am more inclined towards a more flexible treatment of “faith” language, according to which we describe something as faith when someone relies on judgments, where the anxiety-disposing potential epistemic shortcomings are relevant or salient to us. In other words, I prefer a contextualist treatment of “faith” discourse:

1. (4) A speaker accurately describes a subject as relying on “faith” iff the subject relies on their epistemic judgments, despite the fact that the speaker's conversational context treats temptations to epistemic anxiety as natural for someone in the subject's position.

This treatment of “faith” discourse is similar to popular contextualist treatments of knowledge ascriptions.²⁴ On this account whether one counts as exhibiting “faith”, in relying on one's judgment about something, will depend on which kinds of worries are treated as important in the speaker's conversational context. In a conversation in which the basics of racism in America are treated as obvious, and where no one is taking seriously the possibility that it might all be a phantasm of P.C. culture run amok, someone's firm commitment to the recognition of racism in America will not be described as “faith”; in other contexts, where an inability to establish it from common ground is a salient liability, it will be accurately described as “faith”.

But again, I want to emphasise that to describe something as “faith” is not thereby to disparage it. I deny that faith is a matter of going beyond the evidence. On the picture I am sketching, one's faith can manifest in a commitment to treating the evidence as the evidence it is, even given some natural temptation to doubt, perhaps because one lacks independent evidence to treat it as such.²⁵ As the cases I've been discussing show, faith sometimes manifests epistemic virtue. Skeptical philosophers attempt to create an unfair playing field – in challenging methods, they propose rules that the methods themselves invalidate. (So the skeptic is just as question-begging as the non-skeptic.) It is a mark of rationality to refuse to play by unfair rules.

4. RELIGIOUS FAITH?

What I've said so far is grist for the mill of many religious epistemologists – particularly that of ‘reformed epistemologists’ like Alvin Plantinga. My picture suggests a kind of symmetry between religious faith and ordinary non-skeptical stances. I've argued that sometimes, it is rationally permissible – perhaps even rationally obligatory – to accept certain claims or inferences, even though one is in possession of no non-question-begging justification for them. This, of course, is exactly what someone like Plantinga will want to say about his own religious beliefs.

But the psychological symmetry I have been positing does not imply epistemological symmetry. Not all faith is epistemically proper faith. There is a natural temptation to suppose, given the role of faith in epistemically proper reasoning, that a positive treatment of faith implies that just about anything can be held rationally on faith. Some interpreters of reformed epistemology, such as Keith Parsons, have argued that this is the best way to implement the Plantinga-style view. Parsons describes Plantinga's methodology as one according to which one begins with what is obviously properly basic, and generalises appropriately in a kind of reflective equilibrium from there. But different candidate basic beliefs might be obvious to different people. Plantinga thinks the belief that God loves all His creations obviously basic²⁶; someone else might think obviously basic the Kelx belief that our world is the fictional creation of a condemned criminal, attempting to demonstrate that there is the potential for good within him. A third person might consider a naturalistic worldview obviously basic. Parsons observes that if one selects one's own paradigms, it's not hard for non-Christian frameworks to be just as well-justified as Christian ones. Parsons interprets Plantinga's proper basicality as a relativist notion.²⁷ I agree that there may be no mutually agreeable standard by which we can adjudicate these disagreements, but it doesn't follow that there's no neutral framework in the sense of an objective, mind-independent criterion that renders one approach right and the other wrong. We mustn't confuse the metaphysics of rationality with its epistemology.²⁸

Consider logic again. Inferring according to modus ponens is plausibly properly basic. But Carroll's Tortoise is unconvinced. A relativist treatment would say that I'm rational in inferring according to modus ponens, but the Tortoise, who starts with different paradigms of basic commitments, is rational in declining to do so. The problem with this view is that it is insufficiently objective. The rational norm that says to infer according to modus ponens applies to everyone, regardless of whether they feel it obvious. This is somewhat controversial; it goes against views that emphasise the epistemic importance of seemings. I have defended the objectivity of the norms of rationality at length in other work.²⁹

Correspondingly, the objectivist Plantinga would hold that there are objective facts about which things should and shouldn't be treated as basic. If somebody doesn't treat them as basic, this constitutes a kind of irrationality, whether or not they have, deep down, some appearance as of God's Christian character.

This kind of line has a bit of the feel of a kind of epistemic externalism, but it's not obvious that it must be so interpreted, depending on what exactly epistemic externalism amounts to. Notice for example that it is consistent with the supervenience of justification on the intrinsic. (Notice also that its correlate about logic is an implication of a kind of rationalism that has many internalist features.) Internalism is sometimes described as a matter of access, but access itself is an epistemological notion.³⁰

So Plantinga, or someone like him, could say that faith in one's recognition of a Christian God is objectively a good matter, just as faith in modus ponens or a feminist ideology is. One might say it, but is it plausible? Reformed epistemologists like to emphasise the good company in which their foundationalist approach sits. But as far as the idea of putting faith in beliefs or methods that aren't defended on neutral grounds goes, there is as much bad company as there is good. There are after all many possible cases of faith in irrational methods.³¹ Consider the obviously-irrational counter-induction, the inference that supposes future observations will be the opposite of what has been observed so far. In the past, this has been an extremely unreliable inference form – so far, things have usually tended to continue on in the same way – so, by counter-induction, it will probably work well in the future. This is a blatantly and viciously circular argument, but formally it is very similar to inductive justifications of induction. (Similarly: the gambler's fallacy has been extremely unreliable so far. So it's due!)

By using reasoning, including modus ponens, one can give a soundness proof for modus ponens, guaranteeing that it is a deductively valid argument form. But so too can one, by affirming the consequent, give a soundness ‘proof’ for affirming the consequent. Begin with these uncontroversial premises: (1) if affirming the consequent is a valid form of reasoning, then there is at least one valid form of reasoning; (2) there is at least one valid form of reasoning. From these obvious facts it ‘follows’, by a straightforward application of affirming the consequent, that affirming the consequent is a valid form of reasoning.

The same goes, in its more psychologically complex way, to misplaced faith in flawed ideologies, for example by those in the grip of a white supremacist ideology, who are quick to ignore or explain away compelling evidence for racism. Through the lens of their ideological stance, the fact that black Americans are five times likelier than white Americans to be incarcerated – what to me is strong evidence of an entrenched structural racism in America – looks like evidence that black people just tend to be violent criminals. I think this stance is both morally and epistemically repugnant, but I think that it is sometimes held in a way that is internally coherent and self-justifying.

Since there are clearly good, and clearly bad, methods that could only be justified by using those methods themselves, whether a method is good or bad is not a question that can be settled by these formal tests. Faith sometimes manifests epistemic virtue – as when one puts faith in the deliverance of one's perceptual faculties – and sometimes epistemic vice – as when one puts faith in affirming the consequent. These facts are objective in the sense that there is no guarantee that rational errors will be correctable by means recognisable to those making them.

My own view is that Plantinga is mistaken to consider beliefs about God's attributes basic. They do not seem to me at all plausible starting points in epistemic inquiry. I think that if one is to have a justified belief in the existence of God, one must have some independent evidence for thinking that there is a God. But this isn't something for which I can make a case without begging central questions against someone of Plantinga's faith. To Plantinga, I'm like a skeptic who insists that one shouldn't trust perception unless and until they can give a non-circular justification for doing so. The difference between that skeptic and myself (I think!) is that my stance is aligned correctly with the rational facts, and the perception skeptic's is not.

5. IMPLICATIONS: HOW (AND WHEN) TO ENGAGE WITH FAITH-BASED DISAGREEMENTS

There are possible cases where one is responding rationally to the available evidence, but where deep disagreements, irresolvable without begging questions, persist. In such cases, a neutral approach to discussion and debate, whereby one finds common ground and proceeds from there, won't resolve the issue. If Plantinga and I really do have such deeply different core epistemic commitments, there's going to be nothing we can say to one another to change one another's minds. Just the same goes for me and a modus ponens skeptic, or me and a sexual harassment denier. This is a somewhat disquieting conclusion – much as we like to think the light of reason can always show the way to proceed rationally, we also like to think that working carefully and patiently from common ground is the way to resolve disagreements. But both ideals are idealizations. Neutrality won't always work. Sometimes some people just know better than others.

The framework raises several important and practical questions about how to conduct discourse. It suggests, for instance, that since a neutrality ideal is not always applicable, it would be a mistake to give ‘equal time’ to all points of view in public discourse. Given the role of controversial ideologies in setting out these questions, there's a serious question about what it would even mean to present things ‘neutrally’. If certain ideological stances are incompatible with the recognition of racism, for example, then a policy by a news media outlet not to beg questions against those stances would amount to a failure to report certain known facts. The idea of a neutral educational system is also called into question. I don't start in my philosophy courses from a neutral stance on whether perception is a reliable guide to reality, or whether modus ponens is a rational inferential form. Is there more reason to start from a neutral stance on whether there is a God, or whether there is such a thing as systemic racism? These become pressing questions.

One reason it might be worthwhile not to beg central questions against, say, white supremacists, is that ideological commitments are often partial. I have argued that sometimes, certain irrational combinations of views are held in a way that preserves their internal consistency, but they don't always do so. When I meet someone who is skeptical about racism, I don't jump immediately to the conclusion that they're a hopeless devotee of a carefully thoroughgoing racist ideology. Maybe they just haven't had sufficient exposure to anti-racist patterns of thought. If so, finding the common ground and working from there might be very useful in bringing them to the truth. Notice also that one might be rather deeply committed to a flawed ideology and also rather committed to an inconsistent, epistemically proper, ideology. Conversation with such conflicted individuals might help bring them to the truth. Some people shift their ideological stances when they come to recognise them to be inconsistent with their other commitments. So there can be pedagogical uses to engagement on neutral terms with people involved in such deep disagreements.

Another reason to engage in a non-dogmatic way is that exposure to diverse commitments can help people recognise commitments of their own that they may not have noticed. My conversations with deniers of sexism in professional philosophy, for example, have helped to bring into focus some of my own commitments about the epistemology of testimony, and who is likely to be best acquainted with which kinds of facts.

We should also not underestimate the arational effects of conversation. Sometimes, one shifts one's stance on controversial matters, not because of an argument about which view is the right one, but because one spends one's time imbibing a certain sort of outlook on the world. Spending a lot of time with feminists is a pretty good way to end up with a feminist ideology.³² Refusing to find common ground makes it harder to reach people in this sort of way. And as Karen Frost-Arnold (2016) has observed, treating someone as a sympathetic interlocutor can help motivate them to be one.

There is some risk, given my approach, of a kind of epistemically vicious closed-mindedness. The framework gives a clear avenue for refusing to take seriously people with which one disagrees. There's a weak sense in which this is sometimes the best thing to do – when you're doing the rational thing, the opposing thing is irrational, and it would be an error to seriously consider adopting it for oneself. But as fallible humans, we're very liable to misapply this advice. We humans love to feel right, even when we're wrong; so we sometimes dismiss the truth out of misplaced dogmatism – out of bad faith. A kind of open-mindedness is also a virtue; this involves the ability to see things from someone else's point of view, even when you strongly disagree with it. It is an epistemic virtue to be open to understanding others. A tendency to listen charitably and sympathetically, even to one who is saying things that seem outlandish, will serve an epistemic agent well in the long term.³³

Finally, finding neutral ground – or even playing on someone else's turf – can be useful for understanding alternate perspectives. This can be valuable even if you already know the perspective to be objectively mistaken – it can be useful to know why someone is wrong about something. Being too dogmatic about one's own ideological commitments can interfere with one's ability to do that.

(There are also, of course, often non-epistemic reasons it's useful so to engage. As members of a shared community, our ability to understand and get along with one another, even in the face of extreme disagreements, has important pragmatic benefits. Even if I disagree with the local climate-change deniers on fundamental issues, their lives and mine will go better if we find harmonious ways to coordinate and co-exist in the same town.)

All these advantages must be weighed against the possible disadvantages of finding neutral ground for civil discourse. Here are two risks. First, just as people can be induced, non-rationally, into a correct stance, so might you, if you open your mind too much to a flawed ideology, yourself go astray via non-rational means. Second, there is a time and a place for civil discourse, but there's a good case to be made that some stances are just beyond the pale. Engaging in conversation from neutral starting points can, in such cases, be harmfully legitimising of those stances.

Just which cases are like this is a deeply contingent and contextual matter. It's not just a matter of how odious or harmful the position is; it's also a matter of how prevalent it is in one's society. Some positions are intrinsically harmful, but if I refused to engage in civil discourse with everyone who held them, I'd have very few people left to talk to in the world. Consequently, this status changes with the prevailing standards of one's culture. It was not all that long ago when, in the United States, it was considered perfectly normal to assume that homosexuality was an immoral lifestyle choice; even among those who knew better, there would have been a very high cost to refusing to take that stance to be worth thinking about, even if only to argue against it. In the circles I run in today, that view is a fringe one; many of us are now comfortable with the stance that in many cases it's not even a view worth refuting. (This will also depend on the particular individual we're considering engaging.) It also seems to make a difference whether a given view occupies a socially powerful position. As I write this paper in 2017, during the Donald Trump administration, some ideas that recently seemed unworth refuting now demand to be taken seriously.

These are of course very difficult and sensitive questions. My point in raising them is to illustrate that the relationship between my remarks on the role of faith in epistemology must interact in a rather complex way with our practical decisions about how to treat one another. I do think that an important step to understanding what is happening in cases of deep disagreement is to recognise the role of faith in everyone's beliefs. To call a stance one of faith is neither to denigrate nor to celebrate it. It all depends on whether the faith is well-placed. Unfortunately, this is something that one may only be able to recognise if it is. This can be a disquieting state of affairs – but that's exactly why faith is called for.³⁴