1 What Is the Problem?

In short, the problem is this: Any modern, moderately intellectually mature believer in God faces a variety of challenges to the rationality/reasonablenessFootnote ¹ of their belief in God. These sorts of challenges are called “epistemic defeaters.” These challenges may be strong enough to prevent them from being able to know various things they initially thought they knew about God (such as that God exists). Some of the challenges have been around for centuries, and others have arisen or become prominent more recently. Many of these challenges are threatening even if we assume that believers in God typically have strong justification for their belief.

A modern, moderately intellectually mature (MMIM, for short) believer in God is a believer who

1. (i) moderately understands their belief in God (and related beliefs),

2. (ii) understands to some moderate degree how those beliefs fit together,

3. (iii) is moderately aware of the dominant scientific understanding of the world and the place of humans in it,

4. (iv) is moderately aware of the outlines of the range of diverse religious stances in the world, and

5. (v) is moderately intellectually competent at forming and maintaining their beliefs (and thus possesses, moderately, intellectual virtues such as curiosity, carefulness, humility, and open-mindedness).Footnote ²

The term “moderate” is intended to depict a degree of possession that is typical of an educated young adult who has the qualities described. A MMIM believer is an idealization, to be sure. Many believers are not MMIM – children, many adolescents, some particularly sheltered adults, and some people with mental disabilities. But I wager that most believers are MMIM (or approximately so), and most of those who aren’t will one day become (or once were) MMIM. And many of the rest fail to be MMIM because they suffer from various intellectual vices.

Consider the plight of a MMIM believer in God.Footnote ³ As most people do, they of course think their beliefs are true, care about their beliefs – especially core beliefs such as religious and moral beliefs – being true, and think of themselves, and wish to be thought of, as rational in holding their beliefs (Wuthnow Reference Wuthnow2012). However, it is hard to be a MMIM believer. In particular, MMIM believers face various challenges to the rationality of their belief in God (and related beliefs).

Take Kristi, a typical MMIM Christian believer in God.Footnote ⁴ Kristi believes that God is all knowing and in some sense in control of every facet of the universe, but Kristi also believes that she is responsible for her actions and has sinned against God. Kristi notices a tension here (a classic tension that has long been discussed in philosophical/theological literature, although Kristi probably knows little to nothing of this literature): How can she be responsible for her actions if God knows exactly what she is going to do ahead of time? Does God’s foreknowledge conflict with her freedom, and thus her responsibility for her actions? Kristi also notices that the world is full of evil and some of it is hard to reconcile with her beliefs about God: Her uncle, a devout Christian and good man, died in a car wreck leaving behind a family; natural disasters kill many people and destroy the homes and livelihood of others; and diseases strike communities with no apparent regard for their moral qualities. Kristi here wrestles with the problem of evil. Kristi notices that the Bible seems full of miraculous events and God or angels speaking to people, and yet she doesn’t see much of that today. When she does hear of such claims she tends to be somewhat skeptical. But if she’s skeptical of such claims today, why should she so readily accept miraculous claims from ancient texts? Kristi here wrestles with the problem of whether it is rational to believe in miracles. She sees that some of the distinctive moral claims made by her brand of Christianity are deeply contested by others in her society and she herself feels that some of her society’s objections to these claims have force. These are all potential problems for Christianity that arise from the specific claims made by Christianity.

There are other more general problems that Kristi is aware of. She knows that other people reject Christianity, and many people accept other religions. And she knows that many of these people seem reasonable, devout, and good. What reason does she have to think she is correct in believing in Christianity whereas all these other people are incorrect? She is aware of the fact that science is very successful at explaining events in the world – even events that for much of human history seemed mysterious. She wonders: Can science explain every event in the world? If so, maybe God isn’t as active in the world as she thought. She’s even vaguely aware that some scientists have given evolutionary explanations for why humans hold religious beliefs – including belief in God! Could it be, she wonders, that her beliefs about God don’t reflect reality at all, but that she holds those beliefs for reasons entirely unrelated to the truth of the matter?

Charles Taylor argues that most people in the West live in a secular age – a society where belief in God “is understood to be one option among others, and frequently not the easiest to embrace” (Taylor, Reference Taylor2007: 3). The things mentioned above are part of what makes theism difficult to embrace, but Taylor diagnoses the difficulty in a view modern westerners have about the self. They view their self as “buffered.” A buffered self privileges itself over the world outside the self by treating things as meaningful, valuable, or as ends worth pursuing only insofar as the self bestows meaning/value/worth (Taylor, Reference Taylor2007: 38). A buffered self views itself as choosing to be open to other things and their value. This is to be contrasted with what Taylor calls a “porous self” which views itself as innately, without any control, vulnerable to the influence of foreign forces that can both move the self and bestow or bring meaning/value/worth. Christianity – and other theistic views – fits better with seeing the self as porous as God himself is viewed as the source of all goodness, value, and ultimate meaning and God knows everything about all human selves and can influence them directly. Assuming that Taylor’s views are correct, Kristi likely views herself as buffered, and so finds it difficult to accept the Christian God, which appears to threaten her view of herself. This isn’t just a psychological obstacle for Kristi, but also an epistemological obstacle if she has good reason to think that selves are buffered. She surely has good testimonial reasons for thinking this as most folks in her community (even modern Western religious folks, Taylor argues), have such a view.

This is but a brief survey of some of the challenges Kristi faces in embracing Christianity.Footnote ⁵ Some amount to psychological or social pressures pushing against Christianity, or certain aspects of Christianity, or pushing toward other views. But others are epistemic – that is, they are challenges that indicate there is something bad about Kristi’s religious beliefs when evaluated from the perspective of attempting to accurately apprehend the way the world is: either her religious beliefs are false, or she lacks good reason to think they are accurate (or she has less reason than she used to have, or thought she had), or she does not know them to be true.

The notion of a defeater is useful for categorizing and understanding the force of these different epistemic challenges to religious belief. After describing defeaters in the next section, we will then categorize a variety of defeaters for belief in God. We will then evaluate in greater detail just a few of these defeaters – in particular, some defeaters that pose a challenge to a wide range of beliefs about God from many different religious traditions.

2 Defeaters

The notion of a defeater originates in the work of twentieth-century epistemologist John Pollock on defeasible reasoning. He defines it as follows:

Pollock’s notion of defeaters is defined in a background view that centralizes reasons as the determinants of justification.Footnote ⁷ In this view, one’s beliefs are justified provided they are supported by adequate reasons. Defeaters are states that, combined with one’s original reasons for believing Q, prevent those reasons from remaining as reasons. Pollock describes two sorts of defeaters. The first, a rebutting defeater, is a reason for believing not-Q. If M* is a rebutting defeater with sufficient strength, then M – a reason for believing Q – and M* will together not be a reason for believing Q as M* will balance against M. The second, an undercutting defeater, doesn’t directly provide any evidence against Q; rather, undercutting defeaters are reasons to doubt or deny that one would not be in M unless Q were true. Others describe them a little differently: either M* indicates that M does not adequately support Q in the circumstances, or M* indicates that M does not make Q likely to be true in the circumstances.

In this view, reasons are sort of like forces that push around where one’s belief should be.Footnote ⁸ A strong reason for believing p is like a fan blowing a small plastic toy across a track and the movement of the plastic toy in a certain direction is like belief in p being justified (how justified is sort of like how forcefully the toy is moved). A defeater, then, is something that prevents the motion of the toy. A rebutting defeater would be like having another fan on the other end of the track, blowing directly with equal force against the first fan. Now the toy doesn’t move at all, and so belief in p isn’t justified. However, note: The first fan is still on. It is still providing a force in a certain direction; it’s just that the force vector is countered by an equal and opposite vector. Likewise, rebutting defeaters leave the strength of the initial reason intact. They just counter it with equal and opposite reasons. (Of course, there can be partial defeaters, which are opposite in direction but of lesser force.) An undercutting defeater would be like turning off the first fan, or rotating it so that it isn’t pointing down the track. Here, no additional force is applied to the toy; rather the force of the initial fan is modified (removed or redirected).

The distinction between rebutting and undercutting defeaters is fairly intuitive from simple examples. Suppose I’m wondering where I placed my keys. I seem to remember putting them in the drawer of the stand by the front door – which I know, from memory, is where I usually place them. These memories are reasons to believe my keys are in the drawer. Now let’s imagine two different cases. In the first case, I go look in the drawer and I do not see the keys there. This perceptual experience is a reason to think my keys are not in the drawer, thus it is a rebutting defeater of my memory-based reason for thinking my keys are in the drawer. In the second case, I realize that when I last came in the house I had a bunch of stuff in my hands and I was in a hurry and tired and I know that, in those circumstances, I am prone to putting my keys down absentmindedly in various other places but – because I usually place my keys in the drawer – my mind later falsely appears to remember placing them in the drawer. All this evidence about the special circumstances in this scenario provides reason for thinking that my apparent memory isn’t a good indicator that the keys are in the drawer. This evidence thus is an undercutting defeater of my memory-based reason for thinking the keys are in the drawer. Notice that in this latter case, my knowledge about the special circumstances is not evidence that my keys are not in the drawer. Maybe they are, maybe they aren’t; maybe I actually do have a memory of putting them there; maybe given the circumstances my memory is faulty. I can’t tell which it is and I don’t know how likely it is that the keys are in the drawer.Footnote ⁹

Defeaters can themselves be defeated. Suppose, to modify the latter example above, my spouse tells me, “Yes, you were in a hurry and tired, but nevertheless you were clearly really on top of things. Those factors didn’t seem to be affecting you at the time.” My spouse – who knows me well – provides testimony that rebuts my reason for thinking my memory may have been faulty in the circumstances. This restores my apparent memory as a reason for thinking my keys are in the drawer.

Defeater defeaters could be defeated by a further defeater, which could also be defeated, and so on. A reason for a belief held by a certain person is ultimately undefeated provided it retains its justificational force (or perhaps a reduced, but still positive justificational force) after considering all other relevant evidence/reasons possessed by the person.Footnote ¹⁰ A person’s belief will be propositionally justified if and only if it is supported by sufficiently strong ultimately undefeated reason(s) possessed by the person. Propositional justification concerns whether a proposition is supported by adequate reasons for a person. It is distinct from doxastic justification, which concerns whether a person’s belief is based on adequate reasons. A person’s belief is doxastically justified if and only if the person’s belief is based on sufficiently strong ultimately undefeated reasons.

Shortly we’ll discuss a few other possible kinds of defeat, but first we should address a couple of objections facing Pollock’s account of defeat. First, Chandler (Reference Chandler2013) provides a counterexample wherein a defeater D defeats the force of a reason R for believing p, while providing its own support for p. The combination of D&R constitutes a reason to believe p, and yet D seems to defeat R, contrary to Pollock’s definition of defeating a reason. This appears not to present a devastating problem for Pollock’s general approach to defeaters since an alternate definition seems acceptable:

In Chandler’s case, R is not a reason that would justify belief Q, although there is another reason to believe Q: D itself. This definition also allows that undercutting and rebutting defeaters are two different ways of M not being a reason that would justify Q in light of a defeater: because M has no justificational force in M&M*, or because M’s justificational force is rebutted by M*’s force (respectively).

The second objection is more concerning. Pollock’s approach to defeaters takes reasons to be what grounds justified beliefs. This approach is conducive to evidentialist, internalist approaches to epistemology such as Conee and Feldman’s (Reference Conee and Feldman2004). However, many contemporary epistemologists are externalists, and non-evidentialist. In their view, reasons do not play an essential role in justifying a belief; rather, a belief is justified when it is the result of a process that is suitably truth-conducive (e.g. reliable, safe, sensitive, or competent). These externalist epistemologies must provide an alternative account of defeat. Goldman’s (Reference Goldman1979) reliabilist account of defeat is particularly notable, although it also has faced objection (see Beddor Reference Beddor2015). Nevertheless, the worry for Pollock’s definition is that it weds the notion of defeat too closely to a general epistemological approach that is highly contested, even though the notion of defeat should have a place in any epistemological theory.

Although Pollock’s account of defeaters fits well with internalist evidentialist theories in epistemology, his account can also be built to fit with externalist non-evidentialist theories. Beddor (Reference Beddor, Brown and Simion2021) defends what he calls “reasons reliabilism” according to which reasons for belief are identified with the inputs to reliable or conditionally reliable available belief-forming processes. Reasons reliabilism can simply import all of what Pollock says about defeaters, while giving a reliabilist account of reasons. Graham and Lyons (Reference Graham, Lyons, Brown and Simion2021) propose instead to define defeaters in terms of warrants (rather than reasons), where to have a warrant to believe p is just to have prima facie propositional justification to believe p. Reasons can provide warrant, but there can be warrant without reasons, as with (in their view) non-inferential perceptual beliefs. Then they provide an externalist account of propositional justification. Their account of defeaters maintains the important structural features of Pollock’s account, simply replacing the notion of reasons with the notion of warrants. These theories may face their own objections, but their existence and promise indicate that Pollock’s general approach to defeaters isn’t clearly hostile or unfavorable to externalist, non-evidentialist epistemologies. Furthermore, we can flip this objection on its head. Pollock’s general approach intuitively accounts well for the phenomena of defeat, and thus a good epistemological theory should be able to accommodate it (or one much like it).

As we’ve seen, Pollock defines the notion of a reason being defeated; however we can also talk about our justification, our belief, and even our knowledge being defeated. We will also want to distinguish between the defeated state (e.g. a reason, justification, belief, and knowledge) and the mode of defeat – that is, the way the state is defeated. Pollock focuses on one defeated state – reasons – and mentions two modes of defeat: rebutting and undercutting. Definitions of other defeated states will reveal additional modes of defeat.

Here are some Pollock-inspired definitions of defeat of propositional and doxastic justification:

These notions are broader than reasons defeat because one of one’s reasons could be defeated while others remain undefeated and thus one could remain propositionally and doxastically justified while having a reasons defeater. Thus the state of being justified will be defeated only if one’s total reasons are taken into consideration (this is why J is defined the way it is). We could also define notions of propositional justification or doxastic justification being defeated. These notions allow that you can have justification without being justified, and that a source of justification can be defeated while you remain justified. It should be fairly clear how to define these notions given the definitions of reasons defeat and defeat of being justified.

Rebutting and undercutting defeaters are also modes for defeating the states of being propositionally or doxastically justified. But there is another mode of defeat for doxastic justification: defeat that changes the basis for one’s belief. One could cease to be doxastically justified based on J if one ceased to base one’s belief on J. The state that caused this to occur would count as a defeater according to this definition; we can call it a “base defeater.” The nature of the basing relation is disputed (see Korcz 2021), but it is fairly easy to see how base defeat could occur. A person could hold a belief on the basis of good reasons, but then later encounter poor reasons for the same belief that are, for some reason, much more attractive to the person and so their basis for the belief shifts to the new, poor reasons. Imagine a person who already has a good reason for thinking a certain politician is corrupt, but who is convinced by poor reasons persuasively presented for believing that the politician is corrupt because of their participation in some elaborate conspiracy. As a result, the person fixates on the conspiracy theory, which guides her belief while the previous good reasons cease having any continuing effect. She may remain propositionally justified in believing that the politician is corrupt (her overall reasons still support this claim), but she is no longer doxastically justified.

Knowledge can also be defeated. One could acquire strong evidence against a proposition that one formerly knew. One could acquire an undercutting defeater: Some information that, together with one’s former grounds for belief, now no longer supports that belief in a way that amounts to knowledge. Different theories of knowledge will offer different accounts of when and how this occurs. And one could acquire a base defeater where one’s new base does not ground knowledge.

Note that the term “defeater” is used in certain theories of knowledge to describe something quite distinct from how we are using the term here. In the “No Defeaters” theory of knowledge, one knows p (roughly) provided there is no fact such that – if one were aware of it – one would cease to be justified in believing p. A defeater, in this sense, prevents knowledge from being had, rather than eliminating a prior state of knowledge. Defeaters, in this sense, also can be completely outside of the subject’s mind whereas defeaters, as we are characterizing them, are mental states (although we’ll shortly examine a kind of defeater that isn’t a mental state but is something one should be aware of).

We can also speak of a belief being defeated. Sometimes we might say “your belief is defeated” as shorthand for “your doxastic justification is defeated” or “your knowledge is defeated,” expecting then that the person will go ahead and give up their belief (or reduce confidence if it is a partial defeater). But there are situations where a person has a belief that is not justified or known, and yet we can say “your belief is defeated.” Plainly in these situations, we mean something different than “your doxastic justification/knowledge is defeated.” We may neither know nor care whether their belief was initially justified/known; we may think that the person nevertheless has a defeater and as a result they should give up their belief. Here the defeater doesn’t take away an initial positive epistemic status (i.e. being doxastically justified or having knowledge); rather, the defeater indicates that the person lacked justification/knowledge.

I’ll call this mode of defeat, status-revealing defeat. Status-revealing defeat comes in many flavors. Something like rebutting, undercutting, and base defeaters can all provide status-revealing defeat. You might possess a reason that would rebut or undercut your grounds if your grounds were on their own to provide the justification/knowledge you thought they provided. Here no actual reasons or positive epistemic states are defeated – because you didn’t have any to start with – but hypothetical reasons and positive epistemic states are defeated: the ones you have considered to be good grounds, hypothetically taken to be as good as you think they are. Alternatively, you might possess a reason for thinking that your actual grounds are some part of G and that part doesn’t amount to good grounds for B.

Another variety of status-revealing defeat resembles undercutting defeat: having a reason to think that G is not good grounds for positive epistemic status. If G in fact was not good grounds, then no positive epistemic status was undermined (nor was a reason undermined). However, once you acquire this defeater, G is revealed to not be good grounds and should you rationally respond to the defeater, you no longer will rely on G. This defeater (and the undercutting-like defeater mentioned in the previous paragraph) and undercutting defeaters all show that the agent should not base her belief on some grounds by showing that those grounds do not suffice for justification/knowledge. The difference is that, with an undercutting defeater, the grounds were good grounds (the defeater set aside), whereas with this defeater, the grounds were not good grounds (even setting the defeater aside).

There is another mode of defeat that has been at the center of some controversy in epistemology: what Nathan Ballantyne (Reference Ballantyne2019: 104ff) calls a competence defeater. Competence defeaters do not address your first-order evidence/reasons/grounds about the proposition in question; rather they address your competence in evaluating your evidence/reasons/grounds. Plainly competence defeaters can at least sometimes result in doxastic justification defeat. Consider a case wherein a person has misevaluated their evidence: they think their evidence supports p when in fact it does not. However, the person has followed what they are justified in taking to be the canons of evidence evaluation to the best of their ability and they have reason for thinking they are generally reliable at evaluating where the evidence points when they follow these canons. This person has good reason to think they have evaluated the evidence properly; as a result they are doxastically justified in believing p, even though their first-order evidence about p does not support p. Their second-order evidence about their competence, together with their considered judgment about what the first-order evidence supports, supports p. In this case, should the person acquire a reason, D, to think that in fact, in this situation, they have not evaluated the evidence competently, their second-order evidence would be defeated, and thus their doxastic justification would also be defeated. D would thus function as a competence defeater.Footnote ¹³

Arguably, reflective subjects – including those who are moderately intellectually mature – will have developed a sense of their competence on various issues and situations and their assessment of their competence is part of the grounds for their beliefs. If this is correct, reflective subjects will always need to be mindful of potential competence defeaters, as those defeaters can potentially knock out some of their grounds for their beliefs (even if their beliefs are true and have positive support).

However, many epistemologists think that competence defeaters do more than this. They don’t just knock out a person’s belief in their competence with respect to p, but they rationally should knock out belief in p as well. Suppose I have an undefeated competence defeater with respect to p: that is, I have good reason for thinking that I am not competent at evaluating p in my current circumstances, and no defeater for this good reason. I then have no reason to think that my belief that p has been arrived at competently. I should thus think that my belief is not justified/rational/known. But then I shouldn’t believe p. How could I, rationally, keep on believing p if my evidence indicates that I am not justified or rational in believing p?Footnote ¹⁴

If this reasoning is correct, then even in cases where a person in fact has good overall reason for believing p and has competently evaluated the reasons, if that person were to acquire good reason for thinking he was not competent (of course, this reason would be misleading), then the person’s belief that p would not be justified/rational/known – despite in fact having good reasons, competently evaluated.Footnote ¹⁵ Let’s call these “misleading competence defeater cases.”

Here is where controversy arises. Some philosophers reject the above description of people in misleading competence defeater cases. They say that people who have good overall reasons to believe p and have competently evaluated their reasons know (and may be justified in believing) p, despite having good reason for thinking they are not competent at evaluating p. They embrace what has become known as “level-splitting”: you can know/be justified in believing p while being justified in believing that you are not justified in believing p. This view has been supported a couple of different ways. Lasonen-Aarnio (Reference 67Lasonen-Aarnio2012) argues that there is no plausible, coherent way to have a rule-based epistemology that embraces (1) a belief is rational only if it is the result of following the correct rules, and (2) a belief is rational only if one lacks evidence that it is rationally flawed. Lasonen-Aarnio (Reference Lasonen-Aarnio2010) argues that some plausible externalist theories (e.g. safety theories) of knowledge cannot account for various putative cases of defeat (including misleading competence defeater cases), and that this should not be seen as problematic. Baker-Hytch and Benton (Reference Baker-Hytch and Benton2015) argue that all plausible theories of justification/knowledge either have internal problems or cannot account for various putative cases of defeat – including misleading competence defeater cases – and thus we should think that some of those putative cases of defeat are not genuine cases.

Despite the arguments of their defenders, level-splitting views are not widely accepted, in part because they reject what seem to many to be very clear cases of defeat and because of the apparent inconsistency of at the same time believing p and believing that one’s belief that p is irrational. However, for our purposes we needn’t resolve this dispute, for defenders of level-splitting grant that there is something epistemically bad about continuing to believe p when one has good evidence that one is not competent in evaluating p. Lasonen-Aarnio (Reference Lasonen-Aarnio2010, Reference Lasonen-Aarnio, Brown and Simion2021), for example, says that in these cases one believes unreasonably: that is, roughly, one believes in a way that is not knowledge conducive. And believing in this way is often criticizable. So even if we grant that knowledge and justification are not defeated in misleading competence defeater cases, people in those cases do not believe reasonably. They lack an epistemically valuable state that we ideally would like to have and typically think that we possess. And realizing that you lack that state will typically indicate that you should change your belief and how you act when your actions rely on that belief. This counts as a form of defeat for our purposes here.

The final mode of defeat that we shall discuss expands on what counts as defeating evidence. All the modes so far have presumed that, in order to defeat a person’s belief/reason/justification/knowledge, a piece of evidence needs to be possessed by that person – that is, they need to be aware of the defeating evidence.Footnote ¹⁶ Several philosophers have argued that evidence one does not possess may also serve as a defeater. For instance, suppose that Jane has evidence that her daughter gets off school early today (her calendar says it is an early release day and she remembers putting the early release days in her calendar). So, Jane believes her daughter has an early release day today. However, the school changed the day to a full day to make up for a past cancelation, and they sent a robocall to parents’ phones informing them of the change. Jane has ignored her phone and the messages she has received on it, so she is unaware of this information. Many people judge that in this case, Jane is not justified in believing that her daughter has an early release day today.Footnote ¹⁷ As Goldberg (Reference Goldberg2018, Reference 66Goldberg, Brown and Simion2021) explains, Jane should have had the evidence from the robocall, and evidence that you should have had can defeat just as well as evidence you do have. Following Goldberg, we’ll call these normative defeaters.Footnote ¹⁸

It is somewhat controversial whether there are normative defeaters – that is, whether the evidence you should have possessed defeats your belief, reason, justification, or knowledge. But it is a common enough view that we shall take it seriously here as a potential defeater. Furthermore, even if normative defeaters don’t defeat belief, reasons, justification, or knowledge, they seem epistemically significant. For we typically view our justified/known beliefs as non-fluky: that is, our being justified/having knowledge doesn’t depend on being sheltered, disabled, ill-informed, or missing out on information we should know. If we have a normative defeater for a belief that p, then this ideal epistemic view of ourselves regarding believing p is inaccurate. Normative defeaters are thus threatening to our epistemic pictures of ourselves.

3 Map of Defeaters for Belief in God

In Section 1, I argued that a MMIM (modern, moderately intellectually mature) believer in God will wrestle with various defeaters for their belief in God. Section 2 has clarified the notion of a defeater, distinguishing various epistemic states that can be defeated and distinguishing various modes of defeat. Now we will map out a variety of defeaters for belief in God, categorizing them as we go by the modes of defeat they present.

Before presenting the map of defeaters, a few general comments are in order. First, some of the defeaters on the map will plainly be reason defeaters – that is, they purport to defeat a reason in favor of God’s existence. Reason defeaters, as mentioned in Section 2, do not necessarily produce belief, justification, or knowledge defeaters, as a person might have other excellent grounds for their belief even if one of their reasons is in fact defeated. Whether reason defeat produces the other sorts of defeat depends on whether there are any undefeated reasons/grounds for thinking that God exists that remain after considering all the purported reason defeaters.

Second, many of the defeaters on the map will purport to be belief, justification, or knowledge defeaters. Which of those sorts of defeaters a believer has will depend upon the prior epistemic state of the believer. If the believer is a mere believer – that is, they are unjustified and lack knowledge – then the only state that could be defeated is their belief. If the believer knows that God exists and is justified in believing that God exists, then both their knowledge and justification may potentially be defeated. On most views of knowledge, justification defeat will also entail knowledge defeat, but the reverse is not true. That is, one’s knowledge could be defeated while leaving one’s justification intact. This could happen if enough of one’s reasons are defeated to drop the level of propositional justification below the knowledge threshold, but high enough to remain justified.

Third, whether a purported defeater actually defeats a person’s belief in God will depend upon the believer’s background mental profile – that is, their various mental states aside from the defeaters. This is so for at least two reasons. First, as mentioned above, which epistemic state is defeated depends on the prior epistemic state of the believer – whether, before considering the defeater, they knew or had a justified belief that God exists. Whether they knew or had a justified belief that God exists depends on their mental profile – their various reasons and experiences that may serve to ground their knowledge or justification. Second, as mentioned in Section 2, defeaters can themselves be defeated. And whether a believer has a defeater for the defeaters also depends on their various reasons and experiences.

Therefore, fourth, I will not be able to say here whether any of the purported defeaters on the map will in fact defeat any particular religious believer’s epistemic state. That is because to do so I would have to know each aspect of the mental profile of that believer that can ground their belief in God or serve as a defeater–defeater. Most of us do not know other people’s mental lives that intimately. Probably many of us do not have that intimate a knowledge of our own mental lives! In addition, I would have to know whether the believer initially justifiably believed or knew that God exists, and to do that I’d have to evaluate the various positive grounds for belief in God. I am not able to do that here either. The best I can do is this: describe, in a fairly general way, the relevant mental states of a typical MMIM believer and discuss whether and how the purported defeaters might defeat this typical believer’s belief assuming, for the sake of argument, that their initial grounds sufficed for having a justified belief/knowledge that God exists. If, instead, we assume that their initial grounds do not render their belief justified or known, then the putative defeaters will instead at best be status-revealing.

Fifth, there are many religious beliefs regarding God that face putative defeaters, but here we are going to focus on defeaters for the claim that God exists. For those interested in evaluating other religious claims, the discussion here will provide a model that may prove useful for analyzing and evaluating defeaters for the religious claims that interest them.

Lastly, before presenting the map of defeaters, I want to describe, very broadly, a typical MMIM believer’s grounds for believing that God exists. Work from the sociology and psychology of religion informs the picture I will paint.Footnote ¹⁹ However, for our purposes here it isn’t necessary to have an empirically accurate picture of the typical MMIM believer. All we need is a picture that is roughly in the ballpark so that we can describe the defeaters faced by someone who comes close to matching this picture. We can then go on to discuss whether a person who matches the picture already has the resources to respond to the putative defeater, or whether there are any resources they might be able to acquire to address the defeater. It is less important, for our purposes, to accurately describe what resources the typical MMIM actually has.

A typical MMIM believer in God will have a variety of grounds for their belief.Footnote ²⁰ These grounds fall into three broad classes: explanatory, experiential, and testimonial. Explanatory grounds indicate that God’s existence and activity help explain various facts about the world. Cosmological arguments, design arguments, moral arguments and the like all fall under this class of grounds. Typically MMIM believers are not philosophically sophisticated. They are not aware of the various forms these arguments take or the history of discussion of them. However, typical MMIM believers are aware of the grounds that these arguments aim to articulate. Evans (2010) argues that these arguments are based on what he calls natural signs of God’s existence: the existence, ordered structure, and moral character of the world. The philosophical arguments aim to articulate how these signs indicate that God exists, but you don’t have to be aware of the arguments to recognize and follow the signs. Evans’s view fits well with the fact that typical MMIM believers recognize these signs as reasons even though they often cannot (or cannot easily) articulate compelling arguments based on the signs. However, if these signs genuinely are signs of God’s existence, Evans argues that it is rational to believe on the basis of an intuitive recognition of the signs, regardless of whether one can articulate an argument based on the signs.

Some historical arguments for miracles also count as explanatory reasons as they argue that certain historical events are best explained by the miraculous intervention of God. The most famous such arguments are Christian arguments for the resurrection of Jesus (Craig Reference Craig1989; McGrew & McGrew 2009; Swinburne Reference Swinburne2003). The historical evidence, such as it is, for miracles is of course not a natural sign of God’s existence. Such evidence may amount to a sign, but only for those who are aware of the evidence and various background evidence needed to appreciate it. As any cursory scan of the literature on arguments for the resurrection will show, they rely on information that it is doubtful many MMIM believers possess. However, MMIM believers are often aware of some of this evidence, as it is often partially encapsulated in religiously framed retellings of particularly significant putative miracles.

Experiential grounds include a wide range of experiences: of apparent miracles, of God’s will or love or some other trait in a rich mystical experience, and of peace, comfort, love, guidance, and forgiveness that often come while praying. Mystical experiences are often difficult for the experiencer to describe and are fairly uncommon,Footnote ²¹ while experiences of peace, comfort, guidance, etc. while praying are common. Indeed, it is quite common for MMIM believers to seek out such experiences while praying and there is a long history of discussion about how to discern God’s guiding will in these experiences (Alston Reference Alston1991; Luhrmann Reference Luhrmann2012, Reference Luhrmann2020; Willard Reference Willard2012).

Testimonial reasons are testimonial sources that in some way ground or purport to support belief in God. These can include the testimony of trusted people who report having had various religious experiences, the testimony of trusted people who have spent many years thinking through and practicing a religion, or the testimony of a religious text such as the Bible. Provided one has reason to trust these sources – and on most views of testimony it is fairly easy for us to acquire reason to trust testimony – one has testimonial grounds for one’s belief.

I suggest that a MMIM believer will typically have some combination of the above three grounds. I do not assume that any of these grounds are more basic, psychologically or epistemologically. Psychologically, it is plausible that most MMIM believers begin with primarily testimonial grounds and later acquire experiential and explanatory grounds as they learn more about their religion and gain experience practicing it. Once a believer becomes MMIM, it seems plausible that their belief will come to rest on some combination of these three sorts of grounds. Their belief may later rest simply on their sense that, as far as they can tell, God’s existence and providential action best “[make] sense of human life” (Donagan, Reference Donagan and Perovich1999: 13), where all the above – and no doubt other – grounds are aspects of human life that God’s existence explains well. Surely most MMIM believers do not explicitly, consciously, run an inference to the best explanation style argument for God’s existence from the above features of human life. But we do have an intuitive sense of how well a set of beliefs hangs together and whether a given belief fits with other things we believe. This intuitive sense plausibly involves a sensitivity to how well God’s existence explains, or makes sense of, these features of human life.

This description of the grounds a typical MMIM believer has for believing that God exists is intended to be neutral about epistemological disputes regarding how a person is justified in believing, or knows, that something is the case. I make no assumptions about whether these grounds should be understood in an internalist or externalist way. Evidentialist internalists like Feldman and Swinburne will understand the grounds as evidence that may or may not support the proposition that God exists. Externalists like Goldman, Plantinga, and Sosa will understand the grounds as playing some key role in a process that may or may not be reliable, properly functioning, or apt in producing a true belief.Footnote ²² Internalist and externalist epistemologies both allow for the possibility of justification and knowledge defeaters, although they tend to give different analyses of how defeaters defeat.

Likewise, this description stays neutral regarding Reformed epistemology – the view that belief in God is (or can be) properly basic.Footnote ²³ A belief is properly basic when it is epistemically proper (justified, rational, or known) and it is not based on an inference drawn from some other belief. In the description I gave of the typical MMIM believer, their grounds for belief in God may or may not be properly basic. Some of the grounds could in principle serve to justify belief or ground knowledge non-inferentially. Religious experience, in particular, is often argued to be a potential ground for properly basic belief. In Evans’s view of natural signs, religious belief can be properly basic when based on awareness of a natural sign of God’s existence, which needn’t involve inferring that God exists from beliefs about the sign.

Let’s now turn to the map of defeaters, which can be found in Table 1. I should note that I cannot here list every specific defeater that has been offered. Rather, I will describe various families of defeaters.

Rebutting defeaters aim to provide reason to think that God does not exist. They may defeat partially or fully; if they provide enough evidence that God does not exist to approximately balance out the strength of the evidence in favor of the existence of God, then they will fully defeat the reasons and produce justification defeat and, perhaps, also knowledge defeat (depending on one’s view of knowledge).

There are three prominent sorts of rebutting defeaters. The first comes from arguments that the divine attributes are incoherent – either in themselves or in combination. For instance, there are arguments that the notions of omnipotence and omniscience are self-contradictory. There are also arguments that some of the divine attributes are not compossible – for instance that perfect goodness precludes freedom.Footnote ²⁴ If any of these arguments succeed and the notion of God is incoherent, then there cannot be a God – thus providing a rebutting defeater of belief, reasons, justification, and knowledge. The second is the venerable problem of evil, which comes in many varieties. The basic idea is straightforward: there is much evil in the world, but we should expect God to prevent such evil for, being omnipotent, God has the power to prevent evil, being omniscient, God knows how to use their power to achieve their ends, and being perfectly good, God would want there to be no evil (or a less amount, or no evils of a certain sort). Sometimes instead the problem is stated as a probabilistic comparison of two potential explanations for the evil we see in the world – atheistic naturalism and theism. This version argues that atheistic naturalism better explains and predicts the patterns of evils we see than does theism, and thus the evil we see supports atheistic naturalism over theism.Footnote ²⁵ The third, first rigorously presented and popularized by Schellenberg (Reference Schellenberg1993), is the problem of divine hiddenness. In short, given God’s goodness and love, we should expect God’s existence to be more evident than it is if God really did exist. Thus, we have reason to think that God does not exist from the fact that it is not particularly evident that he does exist.Footnote ²⁶ These are not the only rebutting defeaters for the existence of God, but they are the most prominent, widely discussed, and widely accepted.Footnote ²⁷

Undercutting defeaters aim to undercut the force of the grounds supporting belief in God. They broaden the evidential base such that the total, broader base, no longer supports or properly grounds belief that God exists. As with rebutting defeaters, undercutting defeaters can be partial or full and can produce reason, belief, justification, and knowledge defeat. Additionally, some undercutting defeaters may show that the grounds one had for believing that God exists never were good grounds to begin with. In that case, they produce status-revealing defeat.

There are three main types of undercutting defeaters. The first is a very big family, comprised of all responses to the positive grounds for theism that aim to neutralize the force of those grounds. Examples include arguments against the principle of sufficient reason, evolutionary explanations of the appearance of design, and arguments that religious experience has little to no justificatory force. The name for the second undercutting defeater – the Superfluity argument – comes from Van Inwagen (Reference Van Inwagen, Dole and Chignell2005), although the objection is much older. Indeed, it was discussed by Aquinas. It goes like this: we can account for all that we see and know about the world at least as well using only naturalistic resources as we can using theistic resources. God’s existence and activity are explanatorily superfluous. Thus, explanatory and experiential grounds do not support theism over atheistic naturalism. Testimonial grounds presumably will be epistemically useless if the grounds testified to do not support theism over naturalistic atheism. This argument will presumably overlap somewhat with the first family as individual potential grounds for theism will have to be assessed, but what distinguishes the superfluity argument is that it focuses on undermining the grounds for theism by showing those grounds do not support theism over other explanatory hypotheses.Footnote ²⁸ The third is the problem of unpossessed evidence: people have been discussing whether God exists for thousands of years and have written tons evaluating the very grounds had by a typical MMIM believer. Most believers know very little of these discussions. Nobody knows all of them. Surely there’s a good chance that there are defeaters lurking in the mass of evidence a believer does not possess, and they have no idea whether those defeaters are any good. This evidence of the existence of defeaters itself functions as an undercutting defeater.Footnote ²⁹

Base defeaters can either change the base of one’s belief in God to something that is evidently not a good ground for theism, or they can reveal that one’s belief in fact is based on different grounds than one thought – grounds that are evidently not good. In the latter case, the base defeater is also a status-revealing defeater, thus defeating belief. In the former case, one’s original grounds/reasons aren’t defeated per se; rather, one’s doxastic justification and knowledge are defeated. As far as I am aware there are no standard, widely known base defeaters of the former variety. Individual believers may acquire one when they fall for charismatically presented but evidently faulty reasons. There are, however, several varieties of the latter sort of base defeater.

Certain sorts of genealogical debunking arguments offer one kind of base defeater. These arguments offer evidence for a genealogical account of one’s (or a group’s) religious beliefs that reveals that the base of religious belief is some process or ground that is plainly not epistemically supportive. Freudian and Marxist explanations of religious belief are good examples. They purport to show that belief in God stems from wish fulfillment, such as a deep need to feel loved by a father figure, or a desire for comfort in a threatening, chaotic world. Today the most plausible genealogical arguments come from the cognitive science of religion (CSR) – a discipline that draws on cognitive science and evolutionary theory to explain religious belief and behavior. Researchers in CSR explain belief in God as either the byproduct of other evolved cognitive traits, or as selected for because of certain benefits individuals or groups get from believing in God. Either way, these real, cognitive/evolutionary explanations for belief in God are not reliable indicators of God’s existence. The real grounds for human belief are not epistemically supportive. A variety of arguments of this sort have been developed in the literature, drawing on various genealogical explanations and epistemic principles (Thurow Reference Thurow, Fuqua, Greco and McNabb2023b). Another sort of argument – rationalization arguments – purport to show that people’s stated or apparent grounds for their belief are not their real grounds. Rather, they are mere rationalizations. Rationalization arguments are sometimes combined with genealogical accounts of a believer’s true grounds, which are then judged to be evidently epistemically poor.Footnote ³⁰

Competence defeaters aim to show that a believer is not competent at evaluating the force of their reasons/evidence/grounds for their belief in God. Competence defeaters aren’t reason defeaters, but they can be justification and knowledge defeaters (though see the discussion in Section 2, again, for qualifications). They can also defeat belief as a status-revealing defeater if they show the believer was never competent to begin with.

There are four prominent kinds of competence defeaters for belief in God. First, some genealogical debunking arguments aim to show not that the true genealogy for one’s belief in God is rationally defective, but rather that the believer cannot tell whether the true genealogy is not rationally defective.Footnote ³¹ Second, the problem of historical variability notes that if Kristi Christian had been born in India, she’d probably be a Hindu or a Buddhist. If she’d been born in Saudi Arabia she’d probably be Muslim. There are different ways to try to turn this doubtlessly true observation into a defeater for theistic belief. The most plausible, in my view, offers a competence defeater: the lesson to draw from this observation is that humans’ cognitive tools and methods for evaluating religious claims are not generally very good and we have no reason to think that our own present tools and methods are any better than any other believer’s (or any better than those we would have used had we been born elsewhere in other conditions).Footnote ³² Third, there is the venerable problem of disagreement, which also can be developed in various ways. It can be understood as a rebutting defeater: testimonial evidence from people who disagree with us about p who are, as far as we can tell, epistemic peers (i.e. just as intellectually virtuous and well-informed as ourselves) just is evidence for us against p. It can be understood as an undercutting defeater: the fact that so many people who are otherwise epistemically competent come to different views about what the evidence about p supports is evidence that the total evidence doesn’t support p (or not-p). This higher-order evidence about the evidence seems to undermine, for those aware of it, whatever support the evidence may have originally given. There are situations and beliefs for which disagreement would, I think, best be understood as defeating in one of these two ways. However, for theistic belief, I think that disagreement is best understood as a competence defeater: widespread disagreement among people who are otherwise epistemically competent demonstrates the class of methods used to form beliefs about God’s existence are collectively unreliable, and nobody has any reason to think the process they use is more reliable than the others. Somebody may in fact be properly responding to the grounds they possess, but nobody can tell who that is (if anybody is). What is called into question is everybody’s competence at evaluating the evidence/grounds – not what those evidence/grounds support. Lastly, all these grounds (as well as others) for generating a competence defeater can be combined into a cumulative case argument.Footnote ³³

Ingredients used in one defeater can be used in other defeaters as well. We already observed how disagreement can be used in different sorts of defeaters. Similar things can be said for evil and for historical variability; both can be used to support the problem of divine hiddenness, for example. Furthermore, although multiple defeaters can be combined independently to increase the net strength of defeat, some defeaters can increase in their own strength in combination with another defeater. For instance, a supporting defeater may knock out an objection to, or support a key aspect of the supported defeater. We’ll see examples of this in later discussion.Footnote ³⁴

I suggest that most typical MMIM believers will be aware of at least some defeaters on the chart. By “awareness” here I mean de re awareness – that is, they are aware of the phenomena that form the basis of defeaters. They may or may not recognize these phenomena as defeaters. There may be an evaluative epistemic difference between people who merely have a de re awareness of a defeater and those who also recognize the defeater as a defeater (or putative defeater). I won’t tease out these subtle epistemic differences here. Mere de re awareness of a defeater is surely sufficient to produce some amount of defeat and in any case many MMIM believers are aware of some of these defeaters as putative defeaters.

Typical MMIM believers are most likely aware of the existence of puzzling instances or amounts of evil, the existence of widespread disagreement by otherwise seemingly rational people, and the tendency of science to explain more and more of reality including, potentially, human religious beliefs. Just as I do not assume MMIM believers could present the grounds supporting their theistic belief in an argument or syllogism, I do not assume they could present these defeaters in an argument or syllogism. Most likely, their awareness of the defeating phenomena is accompanied by felt tension with their theistic belief, although that tension is surely not always present in their mind and is in most circumstances overwhelmed by their confidence that God exists.

However, MMIM believers who are ignorant of some (or perhaps even all) of the defeaters on the chart may be responsible for their ignorance. If they should have known some of the defeaters, then they will possess a normative defeater – the defeaters they should have known can defeat their theistic belief simply because they should have been aware of those defeaters. What makes it the case that a person should have known something? Goldberg (Reference Goldberg2018: 160ff) argues that we humans have general epistemic responsibilities to play our social-epistemic roles properly. For instance, a medical doctor is expected to know about the latest professional standards regarding how to treat various conditions. This isn’t just a predictive expectation but a normative expectation, for if a doctor is unaware of some relevant professional standard and thus fails to treat a patient’s condition in accordance with how that standard recommends, the patient has grounds for complaining and seeking redress. Non-doctors are not expected to know these things. We might predictively expect a particularly well-informed layman to be aware of the professional standards, but if they aren’t and we nevertheless take the layman’s advice on how to self-treat a condition, we have no grounds for complaining and seeking redress (at least, assuming the layman was simply sharing his honest opinion, was not trying to deceive us, and fulfilled his epistemic duties). There are many kinds of roles that place us under epistemic expectations: professionals, experts, service providers, jobs, and various interpersonal relationships. If I have a social-epistemic role, the proper fulfillment of which would require that I know p, then I ought to know p.

MMIM believers do not in general have professional or expert duties that would require them to be aware of these defeaters. (Some might happen to be professional philosophers of religion; they do have a duty to know these defeaters!) But most – certainly the typical – MMIM believer will be a member of a religious community and religious communities have social-epistemic expectations of their members. At the very least, most religious communities expect their members to know the doctrinal basics of their religion and to have more than a passing familiarity with their authoritative texts. For theistic religions such as Christianity, Judaism, and Islam, fulfilling these expectations would require that one be aware of the problem of evil as their religious texts directly and explicitly address the problem of evil. Members of theistic religions have an obligation to continue to grow in their faith and service to God, which is often taken to require that they come to ever-deeper understanding of God and his will. Acquiring a deep understanding will unveil some of the putative tensions in divine attributes and perhaps the problem of divine hiddenness. Many theistic religions have an evangelistic imperative – they expect their members to be prepared to spread their faith. Effectively carrying out the evangelistic imperative requires some knowledge about the people one might evangelize and some of the obstacles facing their conversion, which surely includes some defeaters (although it may be indefinite which of the defeaters one might need to know).Footnote ³⁵

Many MMIM believers live in a religiously diverse, secular society and they may possess social duties as members of those societies – duties that produce normative expectations to know certain things. Since many diverse secular societies are also democratic, and citizens in democratic societies have social duties to participate in the political process by at least voting, many MMIM believers will have social duties to vote. Voting well requires that one be well-informed about the general features of one’s society. Religious diversity is presumably one of those general features. Thus, many MMIM believers will be normatively expected to know about religious diversity, and the problem of disagreement arises right from religious diversity. Some societies include a basic knowledge of religions and their place in society on the state-approved K-12 (or equivalent) school curriculum. People in those societies can be expected to know about religious diversity and the historical variability of religious belief.

In short, many MMIM believers will be aware of at least some putative defeaters on the chart. Many believers will be normatively expected to know about some of the defeaters. Assuming these putative defeaters are genuine (and normative defeat occurs), MMIM believers’ belief in God is (at least partially) defeated. Perhaps some MMIM believers manage both to be unaware of any of these defeaters and to lack any normative expectation that they would know of any of them. Their belief in God will not face defeat. However, I suggest (i) few MMIM believers will be in this position, and (ii) many of those who are in this position will very likely one day learn of or acquire a duty to learn of some of these defeaters. And in any case, it would be epistemically sad for theism if the only way to be an epistemically well-off believer is to avoid hearing about or having any duties to hear about the defeaters. Thus we now turn to a new question: are these defeaters any good? If they provide any prima facie defeating force, can they themselves be defeated? The rest of this Element focuses on beginning to answer these questions for a few of the defeaters on the chart.

There are plainly too many defeaters to discuss in the space available to us! Instead, I’ll focus mostly on some defeaters that present a threat of justification/knowledge defeat without requiring an evaluation of the strength of theistic grounds. These sorts of defeaters – including genealogical debunking arguments and the problem of historical variability – are cleaner defeaters in that, if they work, they defeat theistic belief without having to wade into complicated disputes about the strength of specific theistic grounds. I will not be able to discuss all the defeaters of this sort. For instance, the problem of disagreement will receive only brief discussion here; however, it has its own Element to which I refer the reader for further discussion.Footnote ³⁶

Section 4 will be devoted to making a few comments about the rebutting and undercutting defeaters on the table. Section 5 will discuss the base defeaters, and Section 6 some of the competence defeaters.

4 Rebutting and Undercutting Defeaters

4.2 Undercutting Defeaters

Most undercutting defeaters are aimed at undercutting specific grounds for theistic belief. Each of the grounds mentioned in our description of MMIM believers is the subject of centuries-long debate, some defending the grounds, some offering defeaters for the grounds, others offering defeaters for those defeaters, and so on. We cannot examine any of these defeaters here.Footnote ⁴⁸ Instead, we’ll focus on the superfluity argument and the problem of unpossessed evidence, as both defeaters aim to be general and less tedious. They are general in the following sense: they aim to undercut justified belief/knowledge of theism without knocking out the theistic reasons one by one. If they succeed in doing so, they’re less tedious: figuring out whether theistic belief is justified/known will not require us to carefully examine a bunch of grounds and their putative defeaters one by one.

4.2.1 The Superfluity Argument

The superfluity argument aims to undermine the putative force of grounds supporting theism by presenting an alternative explanation of the grounds. The more grounds we can explain naturally, the less overall support is provided for theistic belief by the grounds. Bayesian reasoning can be used to make this argument. According to the relative odds form of Bayes’ theorem,

$\frac{\text{Pr}\left(\mathrm{T/e}\right)}{\text{Pr}\left(\mathrm{N/e}\right)}\text{=}\frac{\text{Pr}\left(T\right)}{\text{Pr}\left(N\right)}\times \frac{\text{Pr}\mathrm{(e/T)}}{\text{Pr}\mathrm{(e/N)}}$

The ratio on the left compares the posterior probability of the hypotheses T and N in light of evidence e (that is, the probability of the hypotheses given e), the first ratio on the right compares the prior probability of the hypotheses (that is, their probability prior to considering e), and the second ratio on the right compares the likelihoods of the hypotheses (that is, how probable the evidence is on the hypotheses). We can view the various grounds for theism as different bits of evidence e: e1 = various religious experiences, e2 = natural facts that appear well-designed, etc. Let T = theism (roughly, God exists and is the creator of the universe) and N = naturalism (roughly, there is no God or other supernatural beings, but only the entities within nature and laws governing their behavior). A given evidential ground will favor T over N only when the likelihood ratio is greater than 1 – that is, when e is more probable on T than it is on N. Whether T ends up being more likely than N depends of course on the prior probabilities as well, but the change from the prior probability contribution to the posterior probability comes entirely from the likelihood ratio.

We might initially think – and have good reason to think – that a given piece of evidence, e, supports theism. Perhaps e provides some pro tanto evidence in itself (or perhaps in combination with some other information) for theism. However, using a Bayesian approach, e will support T over an alternative hypothesis only if theism better predicts e than the alternative considered. Now the naturalistic defender of the superfluity argument comes along and suggests that naturalism predicts e as well, or perhaps better, than theism. If they are correct, then the likelihood ratio will be less than or equal to 1, which means the evidence does not favor theism over naturalism, and so it cannot suffice as grounds for believing theism over naturalism. Thus, the evidential force of e has been defeated. If the naturalist can pull this off for all the grounds of theism, then a believer will no longer know or be justified in believing that God exists (i.e. justification and knowledge defeat result).

Some believers, depending on their grounds, will have their belief in God at least partially defeated. For example, theists whose grounds include the apparent design of various biological traits. The theory of evolution by natural selection offers a naturalistic explanation for why organisms are well-adapted to their environment. Without this theory, naturalists do not have a great explanation of why organisms are well-adapted, but theists do: God, in his goodness, wants things to live and thrive and so gives them the traits to do so. If e = organisms are well-adapted to their environment, then Pr(e/T) is fairly high and, prior to knowledge of evolution by natural selection, Pr(e/N) is low. The likelihood ratio favors theism (thus, the design argument has some force prior to knowledge of evolution by natural selection). But, once we become aware of evolution by natural selection, we now must examine Pr(e/T&E)/Pr(e/N&E), where E = organisms evolved through natural selection. This ratio looks to be around 1, as now Pr(e/N&E) is fairly high.Footnote ⁴⁹

At this point defenders of intelligent design will push back, pointing out various biological structures that, they claim, cannot be explained by the present understanding of the theory of evolution (which allows for other explanations of traits aside from natural selection). This isn’t a Element about the design argument, so we won’t pursue this debate any further.Footnote ⁵⁰ However, even if the defenders of intelligent design are correct that evolution cannot explain some biological traits, it has succeeded in explaining other traits and the general adaptiveness of organisms, so some of the evidential base of the classic design argument has been knocked out. Theists who believed based on that evidential base thus possess partial defeat. Whether this partial defeat amounts to much depends on the strength of their other grounds.

Another ground that is at least partially defeated is evidence in favor of miracles. Consider evidence e = there are many reports of miraculous events – that is, events that cannot as far as we can tell be explained in any naturalistic way and that have some sort of religious significance. This evidence may, assuming the reports come from reliable sources, provide some initial evidence in favor of theism. However, once we consider naturalism’s explanatory capabilities, its evidential force is defeated. For, Pr(e/N) is actually pretty high. In any large, complex system of laws and events, many events will occur that, in fact, have a purely naturalistic explanation even though humans have little to no idea what that explanation might be. Surely at least some of those events can be interpreted to have religious significance – as signs of what God wants one to do or as signs of his blessing or judgment. So even if Pr(e/T) is fairly high, so is Pr(e/N) and so the likelihood ratio for this evidence does not favor theism over naturalism.Footnote ⁵¹

Some evidence about miracles and apparent design is not as easily defeated. Consider e = there are intelligent creatures with the power to reflect on morality and modify their will based on their reflections. Pr(e/T) is arguably a fair bit higher than Pr(e/N) even if we enrich N with the theory of evolution. It is quite unlikely that these sorts of intelligent creatures would have evolved, even on a stable, habitable planet like Earth assuming naturalism, whereas it doesn’t seem at all unlikely that there would be such creatures if God exists. So, even if naturalism has an in-principle explanation for the evolution of humans, the likelihood ratio may still favor theism. Thus, e will constitute evidence for theism (and thus will not be defeated by the superfluity argument). Now consider e* = the various events surrounding Jesus’s purported resurrection and the testimony concerning it. Although we cannot get into the details of arguments for and against the resurrection of Jesus, we can at least observe an initially strong case to be made that Pr(e*/T)/Pr(e*/N) is much greater than 1, even if we enrich N and T with lots of background information about the Middle East at that time and with various potential explanations of hallucinations and deceptions. These events are the sorts of things you might expect a good God to do, particularly if we assume that this God has interacted with the Jewish people throughout history. But these events are highly unlikely on naturalism, as resurrection of this sort is nearly impossible assuming naturalism, and it seems unlikely that people would honestly and sincerely testify to Jesus’s resurrection at great risk to themselves if naturalism were true. It would have been far more likely for people to have just assumed that Jesus was like all the other failed revolutionaries: perhaps inspiring, but still dead.Footnote ⁵²

My intent here is not to endorse this argument for the resurrection. Rather, my intent is to point out that there may well remain cases for individual miracles, provided we have sufficient evidence about the events surrounding the purported miracle, where the likelihood ratio favors theism. And this is so even if naturalism predicts that there will be many reports of miraculous events. A theist’s belief grounded on evidence where the likelihood ratio favors theism may not be defeated by the superfluity argument.

There is one more piece of evidence worth noting that seems to resist the superfluity argument: eu = the existence of the universe, where the universe is understood to include all positive contingent facts (including the laws of nature and the initial conditions of the universe). Pr(eu/T) seems fairly high. God doesn’t have to create anything, but given his goodness and love, it would not be in the least surprising if he decided to create something to love and care for. Whereas Pr(eu/N) seems really low, as naturalism will tend to treat eu as a brute fact and there is simply no way to predict or explain a brute fact. It’s hard to know how likely this brute fact would occur, but it seems quite unlikely, especially if we add to eu the fact that the universe is habitable for life. At any rate, we do not here have a strong case that, with respect to eu, theism is superfluous.

The above set of observations points to a deeper lesson about the superfluity argument: evaluating it will require a careful examination of several different grounds for theistic belief, just as independent evaluation of those grounds requires. So, we don’t get a quick route to a defeater here. The superfluity argument helpfully points out that to fully evaluate the evidential force of the various grounds, we need to consider whether alternative, compelling explanations exist for these grounds. But we can’t assume naturalism is an all-purpose explainer that will explain away all, or even most theistic grounds. Once again, we will have to examine the grounds one by one to determine the best explanation in each case.

Peter van Inwagen offers a different reply to the superfluity argument. He claims the argument assumes that belief in God is an explanatory hypothesis – that is, that people believe in God because God’s existence offers the best explanation of a set of factors that can be presented in “statable reasons or publicly available arguments” (2005: 145).Footnote ⁵³ And van Inwagen rejects this assumption. He thinks many of our beliefs – belief in external objects, other minds, that the moon exists, that women are intellectually equal to men, and that God exists – are either hard-wired or are “based on a lot more than what [we are] able to put into words” (Reference Van Inwagen, Dole and Chignell2005: 147). So the fact that some alternative hypothesis can explain the statable, publicly available evidence on these propositions tells us nothing about whether our belief in these proposition is justified. Why? Because our belief isn’t based on the statable, publicly available reasons. Put in terms of the terminology we’ve developed here, van Inwagen seems to grant that the superfluity argument might defeat some potential reasons, but he thinks it doesn’t defeat propositional or doxastic justification or knowledge because we have no reason to think it defeats our actual grounds.

There are a couple of problems with van Inwagen’s reply. I grant that a typical MMIM theist will base their belief in God on more than they can state or make publicly available. Belief in God, much like the belief that your mother loves you, can impact your mind and life in a wide variety of ways. Some of those ways may partially ground your belief even though you can’t state them or make them publicly available. Many religious experiences are like this, I presume. However, we saw earlier that a typical MMIM believer will ground their belief at least in part on many of the reasons that the superfluity argument attempts to neutralize, such as that the world looks designed, or that there must be an explanation for why the universe exists. Those who aim to debunk religious experiences with naturalistic explanations can run a superfluity argument. If those grounds are explained as well on naturalistic terms as on theistic terms, it does seem that at least some of a typical MMIM theist’s reasons are at least partially defeated. If enough are defeated, there could be at least partial justification for defeat and knowledge defeat.

Could van Inwagen grant this point and insist, nevertheless, that not all of a believer’s grounds will be touched and so they may well remain justified and have knowledge? I suppose this is possible. Perhaps we are default justified in maintaining our strong beliefs that impinge on our mental life in more ways than we can count, and so we are justified in holding on to these beliefs even if some of our grounds get defeated. But surely there’s a limit. Default justification is quite a different matter from a license to maintain no matter what defeating evidence you come across. If all my reasons I can think of – after giving it a lot of thought – are undercut by the superfluity argument, and the remaining reasons I have are likely to be of a type (say, religious experiences) that has generally been undercut, then I have reason to think that my grounds – even the ones I can’t quite state – are all (or likely all) undercut. This is itself an undercutting defeater. Whether this defeater is strong enough to completely defeat justification and knowledge will depend on the force of the different defeaters.

Perhaps here van Inwagen will retreat further and say that we don’t need any grounds if the belief is hard-wired, it isn’t defeated, and hard-wired beliefs can be justified. I see two problems with this response. First, even if belief in God is hardwired, it can (and apparently does) acquire additional justification from reasons such as those mentioned in the above psychological description of a typical MMIM believer. Those reasons can still be defeated even if the belief is hardwired. Second, hardwired beliefs can be defeated too if one were to learn that the process leading to the beliefs being hardwired does not, in some epistemically important sense, track the truth regarding the belief in question and related beliefs.Footnote ⁵⁴ We will examine these kinds of defeaters in Section 5. Many of them argue that we can give naturalistic explanations for why humans believe in God – explanations that imply that human religious beliefs do not track the truth. If the superfluity argument is intended to explain not just religious reasons but also religious attitudes, then the superfluity argument includes these sorts of base defeaters. A full evaluation of the superfluity argument would then await an evaluation of these base defeaters.

4.2.2 The Problem of Unpossessed Evidence

The problem of unpossessed evidence points out that the typical MMIM believer (and perhaps even the atypically well-informed believer) is aware of only a small fraction of the evidence regarding God’s existence that humanity in total is aware of. Others possess evidence that leads them to different judgments: atheism, agnosticism, or an embrace of different theistic beliefs. And many of these people appear otherwise intellectually competent. Thus, there appears to be a good chance that the evidence a given person does not possess will contain defeaters – that is, will be such that, if the person had possessed it, their belief about God’s existence would be defeated. But the fact that there is a good chance such defeaters exist is (once one becomes aware of this fact) itself a defeater for one’s belief about God’s existence.

There are different ways of analyzing this intuitively compelling argument form. I’ll present a modified version of Ballantyne’s (Reference Ballantyne2019) Doubtful Fairness version of the argument, as applied to theistic belief. This version captures well the reasoning of the argument. It also closely matches Schellenberg’s (Reference Schellenberg2007) subject mode argument for religious skepticism, which is to date the most extensive argument for religious skepticism based on unpossessed evidence.Footnote ⁵⁵

Here is the argument:

Evidence of unrepresentative evidence argument

• EUE1. If S believes p on the basis of some evidence/grounds E and S has either (i) prima facie reason to disbelieve that E is representative of the total relevant evidence/grounds, or (ii) prima facie reason to suspend judgment whether E is representative of the total relevant evidence/grounds, then S has a prima facie undermining defeater for believing p.

• EUE2. Many MMIM believers believe that God exists on the basis of some evidence/grounds E and have either (i) reason to disbelieve that E is representative of the total relevant evidence/grounds, or (ii) reason to suspend judgment whether E is representative of the total relevant evidence/grounds.

• EUE3. They have no defeater for the undermining defeater for believing that God exists. Therefore,

• EUE4. Many MMIM believers have an undefeated undermining defeater for believing that God exists.

The “total relevant evidence/grounds” refers at least to all the evidence or grounds that are relevant to justifying the belief in question that are available to humanity. This would include evidence that one is in some sense aware of but hasn’t adequately accounted for, and evidence had by other people that one isn’t aware of. Schellenberg’s (Reference Schellenberg2007: 23–7) version of this argument also includes undiscovered and undiscoverable evidence: that is, information or facts that nobody in fact is (or could be) aware of, but that would evidentially bear on p, if one were aware of them. A given body of evidence or grounds, E, is representative of the total relevant evidence/grounds, T, with respect to p when (and this is just a stipulation) the doxastic attitude justified by E regarding p approximately matches the doxastic attitude justified by T regarding p. So long as both E and T justify the same general doxastic attitude – belief, disbelief, or suspension of judgment – then E will be representative of T.Footnote ⁵⁶ T and E may justify somewhat different levels of confidence regarding p, and yet E still counts as representative of T. E would be representative of T even if E omits large classes of entirely novel evidence (i.e. evidence very unlike evidence already possessed by the thinker in question), provided the novel evidence wouldn’t, taken as a whole, affect which general doxastic attitude was justified.

Premise EUE1 says that if you believe p and have reason to think that your grounds for your belief are not representative of the total evidence (or you have reason to suspend judgment about its representativeness), then you have an undermining defeater for believing p. This principle plausibly explains many examples. Suppose Joe sees, for the first time, a can of Coke and notices that it is a red can. He then notices that all Coke cans in the package are red. He has some evidence that all Coke cans are red. But this evidence doesn’t justify belief that all Coke cans are red. After all, his sample of evidence is quite narrow. For all he knows, cans in other towns, states, or countries are given different colors. And he doesn’t have any evidence of the color of cans anywhere else. Joe has reason to suspend judgment about whether his evidence is representative, and he is not justified in believing that all Coke cans are red. It is easy to construct similar examples (e.g. concerning attributing intentions to people and explaining physical phenomena that are not directly observable). Ballantyne (Reference Ballantyne2019: 173) offers another familiar example: ten years ago you did some research on an economic issue (say, whether minimum wage increases lead to job loss), and you came to a fairly firm belief. But you haven’t had time to keep up on the issue for the last several years. You now decide to investigate it once again and you find, after doing a literature search online, that hundreds of books and articles potentially relevant to this issue have been published since you last examined the issue. And a cursory glance through them reveals many of them argue against, or present evidence that raises doubts about, your belief about this issue. You haven’t had the chance to look carefully through any of this, and yet it seems already you should suspend judgment. Why? You have good reason to suspect that your evidence is not representative. For all you can tell, this new evidence may contain something that would tell against your belief.

Often we have no reason to doubt whether our evidence is representative. For instance, when I judge someone’s shirt to be red. I have every reason to trust my own perceptual faculties and no reason to think anybody else’s experiences would provide any contrary evidence. I have no reason to think or even suspect that I’m hallucinating or that the lighting is odd.

Premise EUE2 seems quite plausible. Many MMIM believers may be aware, in general terms, of much of the evidence regarding God’s existence that humanity has discussed – cosmological reasons, design reasons, religious experience, evil, etc. However, they are aware that there is considerable discussion and dispute about these reasons. They may know about some of these disputes, but they know (or can easily discover) that there are many objections to the various reasons that they simply haven’t considered. Many MMIM believers are also aware that many people embrace quite different religious beliefs (or reject religious beliefs entirely) and that these people have their reasons. They may not know what those reasons are, but they have good reason to think those reasons exist and have some chance of being good reasons. Some Theravada Buddhists have enlightenment experiences that seem to them to reveal that selves are nothing and that ultimate reality is nirvana – a state of undifferentiated bliss. Some MMIM believers may be aware of these experiences, others may not – but they know that Buddhists have some grounds for their beliefs. And finally, many MMIM believers know that atheists and agnostics have their reasons and offer naturalistic accounts to explain the evidence we see. They may not know much about the details of these accounts, but they know the accounts exist and that there is some chance those accounts undercut positive reasons for theism.

What about EUE3? For EUE3 to be false, MMIM believers would need some reason to think that the reasons/grounds that they have – despite being limited in the ways just mentioned – are nevertheless representative. That is, they’d need some ground for thinking that all the other reasons they are unaware of will likely not make a difference as to whether their belief that God exists is justified. As far as I can see, there might be two kinds of grounds of this sort. First, the grounds they have are so strong that it is unlikely any other set of reasons would be strong enough to undermine them. Second, they have reason to think that someone else has surveyed the reasons they are unaware of and, according to the surveyor, those reasons don’t make a difference. Can MMIM believers have either of these sorts of reasons?

Plantinga famously defends the first sort of move (albeit in reply to other defeaters). He motivates it with an example involving memory.Footnote ⁵⁷ Imagine George applies for a fellowship and bribes a colleague to write an inaccurate recommendation letter inflating his qualifications. The colleague refuses and writes a letter to the department chair informing them of George’s inappropriate request. However, the letter goes missing and another colleague reports seeing George trying to enter the colleague’s office through a window. In fact, George did not attempt to enter the colleague’s office; indeed, he made no attempt to intercept the letter. He clearly remembers what he was doing that day: he was hiking by himself in the mountains. A third person – say, the department chair – may well be justified in believing George attempted to steal the letter. Even if the chair initially had some reason to trust George, the circumstantial evidence undermines those reasons in this case and provides the chair with good reason to doubt George’s testimony that he was out hiking by himself. However, George himself remains justified in believing he didn’t attempt to steal the letter. He clearly remembers being out hiking by himself. He can recognize that the circumstantial evidence gives some reason for thinking he stole the letter, but this evidence is nowhere near strong enough to defeat the justification he has from his clear memory of having been out hiking by himself. Thus we see the following lesson: potential defeaters may fail to defeat a person’s belief if their belief is grounded in a sufficiently epistemically solid way.

We could apply this lesson to the evidence of unrepresentative evidence argument. Suppose George has reason to think there is evidence about his potential guilt that he does not possess: he hears there are other testifiers, he’s sure his past unsavory request has left further traces he can’t recall, and the university has requisitioned his computer. No matter. He still clearly recalls hiking alone in the mountains on the day in question. His memory is a sufficiently strong ground to give him reason to think that (just about) any evidence that he was not out hiking alone is faulty. His memory (or his memory experience, depending on how to understand the epistemology of memory) is itself a defeater–defeater for the unpossessed evidence. Surely the defeating evidence could mount so high that eventually memory is unable to function as a defeater–defeater, such as if “the letter turns up in [George’s] back pocket; [his] fingerprints are all over the file it was kept in; the mountain [he] thought [he] was hiking on that afternoon was destroyed by a volcanic eruption the preceding morning” (Plantinga Reference Plantinga2000: 372n8). But absent such strong evidence undermining his assumption that his memory has been functioning properly, memory can serve as a defeater–defeater for unpossessed evidence.

Plantinga suggests that a person’s belief in God could be grounded in religious experience, or in the proper functioning of the human mind’s sensus divinitatus (in response to the prompting of the Holy Spirit), and that these grounds could serve as defeater–defeaters just as memory does in George’s case.Footnote ⁵⁸ So even if the believer is aware that there is relevant evidence they do not possess, if their belief is grounded as indicated by Plantinga, they have a defeater–defeater for the evidence of unpossessed evidence. EUE3 is false for such a believer.

This reply is, in one way, plausible. The problem of unpossessed evidence seems especially challenging when we investigate issues that cannot be resolved by direct observation or experience. I can’t see microscopic particles, or the insides of stars, or the distant past or future. For complicated systems, I often can’t just discern the right or best way to organize that system. For these issues, we gain knowledge by gathering indirect evidence and then working out the best explanation/account of that evidence. Learning there is unpossessed evidence for these issues is like learning the puzzle you’re working on is missing many pieces: you have good reason to doubt you’re putting it together well. But for matters that can be resolved by direct observation (or memory of direct observation), we are rightly less concerned about unpossessed evidence because the fact that we can (apparently) directly tell that p is true indicates to us that the unpossessed evidence will likely make no difference to whether p is true and whether I am justified. If some people can just tell that God exists, through religious experience or through whatever is involved in the sensus divinitatus, then they can thereby know clearly – “see” almost – that God exists and have good reason to think that the unpossessed evidence will likely make no difference to whether God exists and whether they are justified.

However, as Plantinga grants, we can acquire defeaters even in cases where we can apparently just directly tell that p. And some of our faculties are more vulnerable to certain kinds of defeat in this way than others. For instance, although George’s memory of having been walking alone on the mountain isn’t defeated by the circumstantial evidence mentioned in the example above, it is more easily defeated by those with contrary memories of the same event. Suppose Steve seems just as clearly to remember walking the same path on the mountain that day, at the same time, and not seeing evidence of anyone else. Think, more generally, of how easy it is for our apparent memory of past events to be defeated by others’ (who we have decent reason to trust) apparent memory of the same events. We might initially have reason to trust ourselves over others, but with enough evidence from others whom we have reason to trust, defeat comes quickly. This is especially so when there are plausible explanations for why we might have falsely seemed to have remembered p.

Grounding for belief in God may be similarly vulnerable to defeat. Many people apparently have radically different religious experiences; many Buddhists claim to experience fundamental reality as qualityless. Other people lack experiences at all, despite apparent openness to God. And, as we’ll see, there are plausible explanations for why it might clearly seem to people that God exists.

However, if all this defeating evidence were unpossessed by a theist, Plantinga’s argument plausibly diagnoses such a theist’s epistemic state. Their religious experiences or the proper functioning of their sensus divinitatus gives them apparent direct awareness that God exists. That apparent awareness is reason to think their evidence is representative. However, once they acquire the above defeating evidence, they have decent reason to doubt that they are directly aware of God. Once in that state, the problem of unpossessed evidence arises. Here we see how two defeaters can work together: defeat for the claim that one is directly aware of God’s existence enables the defeater from unpossessed evidence to become operational.

Let’s turn to the second potential ground for rejecting EUE3: reason to think that someone else who has reliably surveyed the relevant evidence judges that it wouldn’t make a difference to one’s epistemic state. A typical MMIM believer will have at least two reasons of this sort. First, typical MMIM believers are embedded in extended religious communities that typically contain scholars or priests who have a fairly good grasp of the relevant evidence. A typical MMIM believer may not personally know such a scholar or priest, but they will usually have good reason to think that their community contains such people.

One might doubt whether there are any religious experts who are aware of all the relevant evidence. If by “all” we mean, literally, all arguments and evidence ever presented, then there are no religious experts. But we don’t need super experts like these in order to justifiably think that the evidence has been surveyed adequately. So long as the experts are aware of a sizable sampling of the evidence – including the supportive and potentially defeating evidence – then they can have reason to think they have adequately sampled the evidence. Furthermore, this work may be divided among many experts. Perhaps some are experts about the problem of evil, others are experts about various evidence for theism and defeaters of that evidence, others about other religions and defeaters that may come from awareness of other religions, and so on. So long as a MMIM believer has reason to believe that there are religious experts in his community who have collectively adequately sampled the relevant evidence and who judge the potential defeaters that arise as not making a difference to one’s epistemic state, he has reason to think that his grounds are representative, and that EUE3 is false.

Some kinds of defeating evidence can, in principle, knock out this reply. For instance, if the MMIM believer becomes justified in believing that the problem of religious disagreement undercuts the rationality of relying on the religious experts in his community, then he will no longer have this sort of reason for thinking that his grounds are representative. And then the defeating strength of the problem of unpossessed evidence will be restored. Once again, we see that two defeaters can work together: the defeater from religious disagreement enables the defeater from unpossessed evidence to become operational.

A typical MMIM believer will have another reason to think his grounds are representative: he has grounds for thinking not just that God exists, but that God, an omniscient being, has revealed God’s existence to humanity. God, being omniscient, knows all the relevant evidence; God is the greatest conceivable cognitive expert. So, the MMIM believer’s grounds for believing in God are rooted in testimony from the most reliable possible testifier who is aware of all (actual and potential) defeating evidence. And clearly – because God is testifying to God’s existence – none of the evidence unpossessed by God’s audience makes a difference epistemically to their belief. A person whose belief is grounded in testimony from the greatest conceivable cognitive expert on all matters thus has every reason to think their grounds are representative (and thus that EUE3 is false).

This reply might seem to be viciously circular, for doesn’t the unpossessed evidence call into question the evidential force of the very evidence that grounds the MMIM believer’s belief that God, an omniscient being, has revealed Godself? No, the reply is not viciously circular; unpossessed evidence doesn’t undercut by undermining the source of the evidence, E, one presently has regarding p; rather, it provides evidence that it is inscrutable what the total body of evidence, including E and much one does not possess, indicates regarding p. That E provides evidence for p isn’t undermined. So, if E provides not just evidence that p, but also evidence that one’s evidence is representative, E can both support p and deflect the potential defeat coming from evidence of unpossessed evidence. The following analogy illustrates this point. Steve visits a library and sees shelves upon shelves of books on the life of Abraham Lincoln. He glances through a few books and sees that there is disagreement among authors on the nature of Lincoln’s relationship with his wife. Now imagine two possible ways Steve’s research could go. First, suppose he reads one book which tells a certain story about their relationship; the story is coherent and intrinsically plausible and the author provides evidence for the story, but the book gives no reason to think this version of the story is better than any others that Steve knows exists in the library. Here, the unpossessed evidence defeats. Second, suppose instead he reads a different book that describes their relationship the same as the first book did; however, this book also contains information about the author’s mastery of scholarship on Lincoln (there’s a brief biography describing her training and laurels and there are copious footnotes demonstrating her awareness and understanding of a vast expanse of Lincoln scholarship). Here, the unpossessed evidence doesn’t defeat. Why? In both cases, the same story is told of Lincoln, and good support is given for that story. In both cases, Steve is aware of unpossessed evidence. The difference is that in the second case, the evidence Steve happens across also provides evidence for its representativeness. And note: the evidence for the story about Lincoln and the evidence for the representativeness of the evidence come from the very same source. The MMIM believer is in a similar circumstance: his evidence that God exists comes from a source that also provides evidence of representativeness – namely, evidence that God has revealed Godself.

There is a potential way to defeat this reply: acquire evidence that God has not revealed Godself or that one has reason to think that the source for belief in revelation isn’t reliable. As with the potential defeaters for the other replies, unpossessed evidence would thus provide a defeater only if accompanied by another set of defeating evidence. So these three replies all teach us the same lesson: the problem of unpossessed evidence does not itself provide a defeater for a MMIM believer’s belief in God. However, unpossessed evidence regarding God’s existence isn’t thus rendered completely inert. It can be activated as a defeater, but only if accompanied by other evidence that defeats the above replies. Unpossessed evidence can thus, at most, function as what we might call a supplemental defeater. In the next two sections, we’ll examine defeaters aimed at undermining the sources believers employ to form and maintain theistic belief. If those defeaters succeed, unpossessed evidence will then become an active supplemental defeater. If those defeaters fail, unpossessed evidence will remain inert (unless and until some other defeater emerges that can knock out the above replies).

5 Base Defeaters

In this section, we’re going to focus on two prominent sorts of base defeaters for theistic belief: a family of genealogical debunking arguments, and rationalization arguments. Both arguments aim to show that theistic belief in fact is based on grounds (or processes) that are different than believers expect, simply based on first-person reflection on the grounds of their belief. And the real grounds for theistic belief are epistemically defective. Thus if these base defeaters work, they defeat theistic belief via status-revealing defeat. That is, they show that theistic belief lacked some important epistemic status (knowledge, justification) all along.

6 Competence Defeaters

Competence defeaters of theistic belief aim to show that a believer is not competent at evaluating the force of their reasons/evidence/grounds for their belief in God. They are not reason defeaters – that is, one’s reasons are not shown to be defective as reasons or grounds. Hypoxia offers a classic example of competence defeat.Footnote ⁶⁶ Those suffering from hypoxia – a lack of sufficient oxygen supply to the body’s tissues – can suffer from impaired judgment wherein deeply flawed reasoning will, on the contrary, seem evident. If you have reason to think you are suffering from hypoxia, you have a competence defeater for other kinds of reasoning you perform, such as in math problems. You should not trust your judgment based on reasoning; beliefs formed when in the presence of evidence of hypoxia are not justified or known. Of course, it is possible, while possessing evidence you are suffering from hypoxia, that in fact you are not suffering from hypoxia and in fact have reasoned flawlessly (or despite suffering from hypoxia, you in fact have reasoned flawlessly in this case) – the defeat doesn’t indicate that your reasons are flawed. Rather, it undermines your rational confidence in your ability to rationally respond to reasons.

In this section, we’ll build up to a cumulative case competence defeater for theism. We’ll begin with genealogical debunking arguments and then build to the cumulative case defeater by adding considerations from historical variability and disagreement.