1 Introduction

Suppose that, while deep in prayer, you have an experience that seems to come from God: you seem to feel God’s presence, hear God’s voice, perceive a divine light, or something of the sort. It’s natural to think that such an experience could be involved in causing you to form a religious belief. An intense experience that seems to be of God’s presence might result in a belief that God is present in the room (and, if you didn’t already believe in God, it might cause you to believe he exists). Or, if you were praying about some important life decision you were facing, seeming to hear God’s voice telling you how to act could result in a belief about what God is telling you to do.

This paper examines the relationship between religious experiences and religious beliefs. At first, it’s natural to think this relationship is straightforward: that one simply forms beliefs based on religious experiences in the same way we form other perceptual beliefs. However, I’ll argue that the relationship between the two is often much less straightforward than this. Empirical evidence reveals that memories of religious experiences often play a more central role in religious belief formation than the experiences themselves. To date, however, this central role for episodic memory has been overlooked by philosophers of religion.

To bring out this overlooked role for episodic memory, this paper will consider two, related questions. First is what I’ll call the “Psychological Question.” It asks: what psychological role does religious experience play in the formation of religious beliefs? More specifically: by what process do religious experiences causally bring about religious beliefs? We’ll see below that many philosophers of religion have taken religious experiences to play some such causal role (as have social scientists of religion—see, e.g., Dawson, 2006; Luhrmann, 2012).¹

The second question I’ll consider is what I’ll call the “Reliability Question.” It asks: for a given answer to the Psychological Question about how religious beliefs are formed, are those beliefs formed using generally truth-conducive cognitive mechanisms or patterns of reasoning? Or, are they formed through generally unreliable processes? (Note: this isn’t equivalent to asking whether religious beliefs are true, since generally reliable processes can sometimes output false beliefs.)²

This pair of questions is of interest in that it’s natural to ask how the cognitive mechanisms involved in religious belief formation relate to the ones we employ outside of religious contexts (cf. Barrett, 2000; van Mulukom & Lang, 2021). Do beliefs about the supernatural involve mechanisms we also use when forming beliefs about more everyday, mundane topics, or is religious cognition somehow peculiar or atypical? If the former, do religious beliefs tend to result from instances of everyday cognition that are biased or unreliable, or do they result from belief forming methods that typically yield true beliefs? We can make progress on these sorts of questions by considering the Psychological Question and the Reliability Question.

Here’s the plan for addressing these questions. In §2, I consider the most obvious answer to the Psychological Question, one that’s dominant in contemporary epistemology of religion: the “Perceptual Account,” according to which religious experiences cause beliefs in the same way ordinary perceptual experiences do. I’ll argue that, under this answer to the Psychological Question, it’s very difficult to make meaningful philosophical progress on the Reliability Question. So, if we assume the Perceptual Account gives a complete answer to the Psychological Question, we should be skeptical about answering the Reliability Question.

§3 then draws on empirical literature about religious belief formation to argue that the Perceptual Account is incomplete. Specifically, it fails to account for the fact that religious beliefs often aren’t formed in the immediate, in-the-moment way that perceptual beliefs are formed. Instead, there’s often a delay between some religious experience(s) and the time when a subject forms a religious belief. In light of this, this section introduces a supplement to the Perceptual Account, the “Memory/Reflection Account,” on which religious beliefs are formed via reflection on episodic memories of religious experiences.

Finally, §4 explores how we should answer the Reliability Question under the Memory/Reflection account. I’ll ultimately argue that adopting the Memory/Reflection Account opens up room to make more meaningful progress on the Reliability Question.

2 The Perceptual Account

When I use “religious beliefs” in what follows, I mean any beliefs about God or some analogous supernatural entity (hereafter, I’ll continue to use “God” as a catchall, but you can substitute your favourite entity). This could be the belief that God exists, or it could be a belief about some property God has—that God is loving, is telling you to act or live your life a certain way, is currently present in the room with you, etc. By “religious experience,” I mean any experience in which one seems to have an experiential awareness of God. As in my descriptions in the opening of the paper, this could take various forms: seeming to feel God’s presence, hear God’s voice telling you to do something, perceive a divine light coming from God, etc. Often, such experiences occur during religious rituals, whether that’s individual, private prayer or high arousal group rituals (as in, e.g., charismatic Christian practices of speaking in tongues, ecstatic dancing, or being “slain in the spirit”).

To answer the Psychological Question, the Perceptual Account says that religious experiences cause beliefs in the same way ordinary perceptual experiences do. Proponents often argue that religious experiences are like ordinary perceptual experiences in terms of their presentational phenomenology. Perceptual experiences present the world as being a certain way: an experience of an apple on the table presents the apple as existing, presents the apple as being on the table, etc. On the basis of how things are presented, one then forms beliefs about the apple and its properties. According to the Perceptual Account, something similar goes for religious experiences. One might have a feeling of God being present in the room while one prays; in that case, one’s experience presents God as being present. On the basis of this experience, one would then form the belief that God is present. Or, one might have an experience as of God’s voice telling one to do something; in that case, the experience would present God as telling one to act a certain way. On the basis of one’s experience, one then forms a belief about how God is telling one to act.

This sort of account has been developed in particular detail by Alston (2004), though various prominent philosophers of religion hold similar views. Plantinga (2000) argues that we have an innate sense for gaining immediate knowledge of God, one which sometimes operates via experiences with presentational qualities (though he thinks it can also work in other ways).³ Similarly, Swinburne (2018) argues that religious experiences involve an apparent awareness of God’s presence, one which, like ordinary perception, makes us inclined to believe God exists unless we’re aware of any defeaters. Other philosophers at least gesture at this sort of view, even if they don’t explicitly develop it. For example, both Broad (1953) and Gellman (1994) make use of comparisons between ordinary perceptual belief formation and beliefs formed on the basis of religious experience.

What does this answer to the Psychological Question imply about the Reliability Question? Answering this would depend on first settling some further questions.

First, how to answer the Reliability Question depends on whether we take perception and hallucination to be instances of the same cognitive process. When one has a religious experience, there are two options for what’s really going on: either one is actually perceiving God, or one is hallucinating. If hallucination and perception are instances of the same process, this would imply that religious beliefs result from a process that’s typically reliable, since perception is typically reliable on the whole. However, if hallucination and perception are distinct processes, then how we answer the Reliability Question depends on settling the further question of whether those who undergo religious experiences genuinely perceive God (a reliable process) or merely hallucinate (an unreliable process).

So, under the Perceptual Account, conclusively answering the Reliability Question at least depends on settling the further question of whether perception and hallucination are distinct processes. If they are, then conclusively answering the Reliability Question also depends on settling the further question of whether those who undergo religious experiences actually perceive God. The result is that it’s very difficult to make any meaningful philosophical progress on the Reliability Question under the Perceptual Account. That’s because answering these further questions depends on settling some very large background issues, issues which are the subject of deep disagreements between philosophers of different theoretical persuasions.

It should be clear why this applies to the question of whether religious experiencers actually perceive God. In order to determine this, we’d first have to determine whether God exists. Of course, the debate between theists and atheists is one of the most difficult on which to make any kind of conclusive philosophical progress.

Something similar goes for the question of whether perception and hallucination are instances of the same process, since answering this question again depends on settling some deep background disagreements. On one hand, a traditional, “internalist” approach to epistemology often individuates processes in terms of how things experientially seem for a subject; under this approach, perception and hallucination turn out to be the same kind of process, as long as they’re indistinguishable from the subject’s first-person perspective. However, various philosophers defend the opposite view. Lyons (2019) defends it on the grounds that perception involves bottom-up stimulation of sense organs, while (most) hallucinations are generated via entirely endogenous, top-down processes. Nagel (2021) argues that belief forming processes which produce knowledge are distinct from those which don’t, and thus that successful perceiving is a distinct process from hallucination. Various embodied and enactivist approaches plausibly also suggest that perception and hallucination are distinct processes: if genuinely perceiving objects in one’s environment necessarily involves embodied interaction with those objects, then internally generated hallucinations wouldn’t count as genuine perceiving (cf. domínguez2019perception?).

Conclusively settling this individuation question therefore depends on first settling some much larger background issues: whether all that matters for individuating processes is how things seem from a subject’s personal level perspective, versus whether physical brain processes also matter; whether knowledge has any place in our theory of how the mind works; and whether perception is constitutively active and embodied. These are again issues about which it’s extremely difficult to make conclusive philosophical progress.

So, under the Perceptual Account, answering the Reliability Question depends on first settling some very large, difficult background issues in philosophy of mind and philosophy of religion. This makes it difficult to make meaningful progress on the Reliability Question. Of course, anyone with a settled view on these background issues will be able to give an answer to the Reliability Question that satisfies them and those who agree with them. But for the many people who don’t have a settled view, it will be difficult to come up with a satisfying answer. Furthermore, among those who have different background views, it will be difficult to achieve any kind of philosophical consensus.⁴

Of course, I don’t mean here to object to the Perceptual Account, since the fact that it makes answering the Reliability Question difficult doesn’t mean it’s false. Instead, my point is that, if we take the Perceptual Account to provide a complete answer to the Psychological Question, then we should be skeptical about making progress on the Reliability Question. In the next section, though, I’ll argue that the Perceptual Account fails to provide a complete answer to the Psychological Question.

3 The Memory/Reflection Account

I grant that the Perceptual Account may accurately characterize some cases of religious belief formation. However, §3.1 appeals to empirical literature to argue that the Perceptual Account leaves a large gap. §3.2 then proposes the Memory/Reflection Account to fill this gap.

3.1 Cases which the Perceptual Account leaves out

Notice that the Perceptual Account implies that the relevant religious beliefs are formed in an automatic way as one is undergoing the religious experience: one has an experience as of God, and one automatically takes the content of this experience at face value. This is necessary to preserve the putative analogy with ordinary perception. When, for example, one perceives an apple on the table, one automatically comes to believe that there’s an apple on the table. Without this, one’s belief about the apple wouldn’t be perceptual. If, instead, there was a delay between the experience and the belief formation, the belief would be based on a memory of the experience rather than on the experience itself. So, if religious experiences cause beliefs in the same way ordinary perceptual beliefs do, it must be in an automatic, in-the-moment way.⁵

Again, I grant that this may sometimes be how religious belief formation works. However, empirical evidence suggests that the relationship between religious experience and religious belief is often more complex than this.

First, consider cases of conversion to a religion, which presumably often involve first adopting beliefs in that religion’s claims about supernatural entities. Studies of conversion, through means like firsthand interviews with converts, suggest that many people begin actively participating in religious communities and their rituals before becoming full believers (Dawson, 1990; Galanter, 1999; Iannaccone, 2006; Luhrmann, 2020). People often initially get involved either because they’re curious to try out a religion, or even for purely practical reasons rather than because they actually believe—for example, because they find a sense of community and belonging. After participating for some time, they’re eventually won over and become genuine believers.

Now, many studies relevant to this point are specifically about faith communities that put a great focus on vivid, personal experiences of God—for example, Christian evangelical communities (Luhrmann, 2020) and smaller, more esoteric “cults” and new religious movements (Galanter, 1999; Iannaccone, 2006). If someone is participating in such a group and their rituals, we’d expect them to be engaging in activities that are aimed at inducing religious experiences. In doing so, they’d have experiences that seem to present God as being present while they’re engaged in these ritualistic activities. Yet, we often see this delay between the time they initially start participating and the time they actually adopt their religious community’s belief that God exists.

This suggests that these subjects don’t form beliefs on the basis of religious experiences right away—if they did, they’d form the belief that God is present (and, therefore, exists) at the time that they’re having the apparent experience of God. Yet, empirical studies of conversion suggest that belief in the existence of God often isn’t adopted on the basis of a single experience, but only after participating in multiple religious rituals over time. Along these lines, Yamane (2000) argues that we shouldn’t think of conversion in terms of a single religious experience, but in terms of a broader conversion narrative that extends over time and encompasses multiple experiences.

So, it seems there’s often a delay between the time people first start participating in religious rituals, which are aimed at inducing religious experiences, and the time they first start believing. This suggests that they aren’t forming beliefs in a way that’s akin to perceptual belief formation, which would automatically occur at the time one has an experience.⁶

This delay between experience and belief formation doesn’t only appear in cases of religious conversion. Next, consider cases in which those who are already religious believers form beliefs about, for example, how to act and live their lives. Luhrmann (2012) provides in depth illustrations of the role of religious experience in such belief formation through her ethnographic research on a Christian evangelical community called The Vineyard.

Luhrmann describes subjects who, over time, practice becoming deeply absorbed in prayer and attuning themselves to God’s voice and presence. These subjects start out reporting that they experience the thoughts of God being inserted into their own internal trains of thought, through, for example, inner speech and mental images. Over time, they train themselves to focus as much as possible on these thoughts; with practice, they become more and more sharp, clear, and vivid. Among those who hone this skill enough, a few eventually report actually hearing God’s voice, feeling God’s physical touch, or seeing phenomena such as a divine light, in a way that’s as vivid as a perceptual experience; they typically don’t report this as common, but as occurring a few times throughout their lives. It can occur in the context of praying about a decision on where to live, where to go to school, whether to take a new job, and the like.

However, these subjects don’t simply have these experiences and then form beliefs on the basis of them right away—for example, while having an experience as of God’s voice telling them to take the new job, they don’t automatically come to believe God is telling them to take the job. Instead, Luhrmann describes a temporally extended process of “discernment”: after receiving an experience during prayer, these subjects often spend some time reflecting on whether the experience really did come from God, or whether it could’ve been generated by their own minds. They reflect on, for example: whether it was consistent with other religious experiences they’ve had; whether it was consistent with what they know about God and his character; and whether the experience brought a sense of peace. They also discuss their experiences with other community members, trying to discern whether they cohere well with the religious experiences of others.

So, it seems that these subjects again have some delay between their religious experiences and the beliefs they form in response, during which they reflect on and discuss their experiences. Of course, this is just data from one, particular religious community that Luhrmann studied. However, it’s not difficult to see how it could generalize to many other communities.

In general, it seems natural that, when one seems to experience God, the typical reaction wouldn’t be to simply accept the experience at face value. Instead, it seems like one would be aware of the salient possibility that the experience was generated by one’s own mind, which would prompt reflection. Luhrmann argues that this tendency to reflect is one mark that distinguishes mentally healthy individuals from psychotic ones: people with psychosis are often unable to question whether an anomalous experience was veridical or hallucinatory, while we’d expect most people to interrogate whether a highly unusual experience was genuine. All of this is especially true if one is considering making a monumental life decision because of a religious experience: if you’re deciding whether to move across the country or start a new career, you’d try to make sure that what you seemed to experience was real, not generated by your own mind.

Now, it may be that, as one has multiple religious experiences over one’s time as a believer, these experiences will eventually start to seem less anomalous. For example, if one becomes well-practiced at engaging in deep, meditative prayer, it may become easier to induce a state of mind in which religious experiences occur. Plausibly, as one has such experiences more regularly, one may no longer require a period of reflection to interpret them. Specifically, it may be that, after going through processes of discerning the meaning of religious experiences on multiple occasions, one will internalize the interpretative framework of one’s religious community, such that one can now automatically deploy this background knowledge while the experience is occurring. The Perceptual Account might then be adequate for describing such cases. In other words, it may be that, the more experienced a religious practitioner becomes at having and discerning the meaning of religious experiences, the more often the Perceptual Account applies to their processes of belief formation.⁷

This would be consistent with my arguments in this subsection, given that I haven’t claimed the Perceptual Account never gets things right. Instead, I’ve offered reasons to think religious experiences often don’t cause religious beliefs in the way the Perceptual Account describes—i.e., they often don’t simply cause beliefs in the automatic way that ordinary perceptual experiences do. Instead, there’s often a delay between an initial religious experience and the time one forms a belief, during which one reflects back on the experience.

This suggests that we need an account other than the Perceptual Account to adequately theorize these cases. I turn to this in the next subsection, introducing the Memory/Reflection Account.

3.2 Memory and reflection in religious belief formation

It should be clear why relevant cases in some way involve memory. These are cases in which a subject’s religious experience(s) play(s) a causal role in bringing about their belief, but in which there’s a delay between the original experience(s) and the belief formation. It therefore can’t be that the belief is caused by the experience in the automatic way perceptual beliefs are formed. Instead, it seems these subjects are later remembering their experiences, and that these memories have a role to play in their belief formation.

However, the empirical evidence suggests they also aren’t forming straightforward episodic memory beliefs. Beliefs based on episodic memories are typically thought to be immediate in a way that’s much like perceptual beliefs: much as one has a perceptual experience and simply takes it at face value, one has an episodic memory and simply takes it at face value. There’s reason to think that religious belief formation is typically more reflective than this.

Luhrmann’s (2012) subjects at The Vineyard offer a clear example of this. Luhrmann describes these subjects as going through an extended process of discernment before accepting an experience as genuinely from God. This involves reflecting on various characteristics of the experience, discussing it with others, and the like. These subjects therefore aren’t merely remembering an experience and automatically accepting it at face value. Instead, they’re reflecting on their memory before forming a belief. It also seems natural to think this would generalize to other religious communities beyond the one Luhrmann describes. As I argued in §3.1, this kind of reflective discernment seems like a natural, rational response to an unusual kind of experience.

Whitehouse (2001) also surveys evidence that, when religious subjects interpret the meanings of religious rituals, the ones that stick out in memory as anomalous and distinct from typical routines become the subject of deep, personal reflection. He contrasts more anomalous rituals with those that are performed as part of regular routines, during which subjects often simply “go through the motions.” Routinized rituals are interpreted in an unreflective way, with subjects simply adopting “official” exegesis provided by religious authorities. In contrast, when a ritual sticks out from the norm, subjects are much less likely to rely on religious authorities, instead engaging in their own, personal reflection. Since vivid, perception-like religious experiences typically aren’t part of regular routines, we should expect them to attract this sort of deep reflection (see also Xygalatas & Maňo, 2022, who found that one is more likely to engage in personal reflection about rituals which are more intense, arousing, and sensory).

All of this evidence fits well with the ways social scientists describe the reflective nature of religious belief more generally. Dawson (1990) argues that religious conversion often isn’t a passive process, but one in which subjects actively monitor and reflect on the way their beliefs and religious identities are formed. Similarly, Boyer (2013) argues that religious beliefs involve reflective, “system 2” reasoning, as opposed to intuitive and automatic “system 1” processing.

So, it seems that religious belief formation often involves processes of both remembering religious experiences and reflecting on those memories. Such beliefs therefore aren’t simple episodic memory beliefs, in the sense of being based directly on episodic memories. Instead, they’re based on a combination of episodic memory and explicit reflection about whether to take those memories at face value, reflection which (as we’ll see below) could take into account various different kinds of factors and background knowledge. So, rather than the source of these beliefs being episodic memory in a straightforward sense, these beliefs instead have their source in reflective processes that take episodic memories as one of their inputs. This reflection involves deciding whether to accept the contents of an apparent experience of God at face value.

This gives us an initial answer to the Psychological Question (although, in the next section, I’ll explore more specific ways of cashing out how the memorial and reflective processes involved might work). The next section turns to considering the Reliability Question under the Memory/Reflection Account.⁸

4 The Memory/Reflection Account and the Reliability Question

Under the Memory/Reflection Account, we can break down the Reliability Question into two parts. First, are the processes involved in producing memories of religious experiences generally reliable? Second, are the processes of reflecting on the resulting memories generally reliable? If the answer to both is “yes,” then it seems religious beliefs formed via a combination of memory and reflection are formed via generally reliable processes.

§4.1 will consider the Reliability Question with respect to reconstructing religious experiences in episodic memory, while §4.2 will consider the reliability of the process of reflecting on these memories. In §4.3, I’ll then conclude with some broader implications about making progress on the Reliability Question under the Memory/Reflection Account.

4.1 The reliability of the memory construction component

A starting assumption in this subsection is that episodic memory is a generally reliable belief forming process. Across a wide range of contexts, we rely on episodic memory to guide our present actions (my memories of our previous interactions influence how I talk to you in the present; my memory of last week’s meal influences what I order at the same restaurant today; etc.). It’s therefore doubtful that our episodic memory system would be selected for if it didn’t typically produce true beliefs (for relevant discussions, see Boyle, 2019, 2022; Michaelian, 2016). This isn’t to claim that episodic memory is perfectly reliable, since the conditions under which it can go wrong are well-documented (see, e.g., De Brigard, 2014). It’s simply to claim that, on the whole, it tends to produce a high proportion of true beliefs.⁹

We should also assume by default that any apparent episodic memory is an instance of genuine, reliably functioning episodic remembering, unless we have some reason to think otherwise. This is just an instance of a more general assumption that we apply to all of our cognitive faculties. For example, we typically assume that our perceptual experiences are instances of reliable, well-functioning perception, except in circumstances where we have reason to think otherwise (e.g., if another person’s testimony gives you reason to think you’re hallucinating).

With this in mind, our default assumption should be that religious believers’ apparent memories of religious experiences are constructed via reliable episodic memory processes, unless we have good reason to doubt this. In other words, when one seems to reconstruct a prior religious experience before reflecting on it, our default assumption should be that this reconstruction occurs via reliably functioning episodic memory processes.¹⁰ If we want to deny this, we need some positive reason for doing so.

There’s one recent thread of empirical work that might seem to put pressure on this default assumption about memories of religious experience. I’ll next appeal to this empirical work to raise an objection to the assumption, then respond to that objection.

Several psychologists of religion have argued that many apparent memories of religious experiences aren’t very accurate to the actual experience one seems to remember (Schjoedt et al., 2013; van Mulukom, 2017). To explain this, they argue that religious rituals, during which religious experiences occur, have a particular combination of features: they evoke high levels of emotional arousal, while at the same time involving an imperative to suppress outward, behavioural signs of one’s emotions. Such emotional suppression is very cognitively taxing. The result, these authors argue, is that many details of a religious experience fail to be retained, meaning one isn’t able to draw on that information when one later remembers it. Instead, the details of the experience must be filled in based on various factors: testimony received after the fact from other members of one’s religious community; testimony from religious authorities about what the experience was supposed to have been like; and the way one wants to remember the experience, given its implications for one’s religious identity and self-conception.

Intuitively, this sort of evidence might seem to suggest that apparent memories of religious experiences are really constructed via unreliable processes. How might we more precisely cash out this objection?

One possibility is to claim that religious subjects’ apparent memories are typically confabulations. Confabulation involves constructing a representation of a past event in one’s imagination and mistaking it for a memory. This occurs when one fails to draw on stored information about an actual past event, instead piecing together a representation purely from memories of other events, from background beliefs, and the like. The resulting apparent memory might bear no resemblance to any actual past event; or, if it does, this resemblance would be purely by accident. As a malfunction of one’s memory system, it makes sense to think of confabulating as an unreliable process that’s distinct from genuine remembering (Michaelian, 2021; Robins, 2020). (There’s philosophical disagreement about this, however, with some arguing we can’t draw a sharp line between genuine remembering and confabulating (e.g., De Brigard, 2014); I’ll return to the topic of individuating memory and confabulation below.)

However, we’d be wrong to interpret these empirical results as suggesting that religious experiences are typically confabulations. These empirical researchers don’t argue that the details of religious rituals aren’t retained at all, or that they’re altogether forgotten after the fact. Instead, they argue that very few details of religious experiences are retained. So, these authors shouldn’t be read as arguing that apparent memories of religious experiences are confabulated after the fact. Instead, the constructive process likely mixes together scant information that’s retained from the event with other kinds of information.

It seems better, then, to interpret these empirical researchers as arguing that apparent memories of religious experiences are typically misrememberings. Unlike confabulating, misremembering does draw on information about a remembered event. However, it involves representing the event inaccurately. Various factors could explain this inaccuracy. For example, it could be that one retained details from the original experience in a spotty, incomplete way, such that one’s memory must be filled in based on background beliefs about the world or one’s desire for the event to have occurred a certain way. Or, it could be that false testimony about an event, received after the fact, was incorporated into one’s memory. This sounds closer to the way Schjoedt et al. (2013) and van Mulukom (2017) describe what occurs during apparent memories of religious experiences.

Philosophers disagree about whether misremembering cases involve genuine episodic remembering or distinct processes (for relevant discussion, see De Brigard, 2014; Michaelian, 2021; Munro, 2021; Robins, 2020). So, if it were true that apparent memories of religious experiences are typically misrememberings, that would raise an issue about how to individuate processes similar to the issue for perception and hallucination discussed in §2. This, in turn, would mean that we’d come up against some deep philosophical disagreements when trying to make progress on the Reliability Question under the Memory/Reflection Account, disagreements that again seem to stall progress on the Reliability Question.

However, the situation isn’t so dire for making progress on the Reliability Question. The reason is that the relevant empirical evidence likely won’t generalize very widely to many cases in which subjects construct memories of religious experiences.

Schjoedt et al. (2013) and van Mulukom (2017) base their arguments largely on one study by Xygalatas et al. (2013), which looks at one, particular religious ritual context—a fire-walking ritual in a small Spanish town. This study did find clear evidence that the experience was a high arousal one in which participants were under social pressure to suppress their emotions and appear calm. And, as a result, it does seem that participants remembered very few accurate details about the experience immediately after, then later recalled false details with high confidence—i.e., they do seem to end up misremembering the original experience.

However, it seems that many contexts in which religious experiences occur simply aren’t going to involve the same kind of pressure to suppress intense emotions. For one thing, it doesn’t seem like this would apply to experiences that occur during private prayer when someone is alone. Such an experience might not be as high arousal, and, even if it was, there will be less reason to suppress one’s emotions if one is in private. There are also group religious rituals where people seem encouraged to express their emotions rather than suppress them. Consider, for example, charismatic Christian contexts where participants fall to the floor, writhe around, scream, speak in tongues, bark like dogs, and the like.

So, Schjoedt et al. (2013) and van Mulukom (2017) may be correct that the evidence suggests there are some cases in which apparent memories of religious experiences are misrememberings. However, they rely on an explanation that won’t generalize widely enough to be considered typical.

With this in mind, then, our default assumption—that apparent memories of religious experiences are instances of genuine, reliably functioning episodic memory—still stands, at least outside of certain special cases. This assumption is of course defeasible, in that future empirical work could undermine it. However, unless some such undermining research crops up, we should answer the Reliability Question in the affirmative when it comes to the memory component of the Memory/Reflection Account, at least for typical cases.

One might still worry that my arguments in this subsection are in tension with what I said about perception and hallucination under the Perceptual Account. In §2, I argued that, to decide whether the process the Perceptual Account describes is reliable, we’d first have to settle the difficult question of whether successful perception and hallucination are distinct cognitive processes or instances of the same type. However, in this subsection, I’ve claimed that we can assume reliably functioning episodic memory processes are involved in bringing to mind prior religious experiences. But why not treat this as parallel to the perceptual case—i.e., why not think that, to decide whether the (apparent) memory component of the Memory/Reflection Account is reliable, we first have to settle the difficult background question of whether or not memory is the same process as confabulation and/or misremembering? Doesn’t the same sort of difficult individuation problem arise for memory that came up for perception?

I agree that this individuation question for memory is philosophically very difficult, much like the issue of how to individuate reliably functioning perception versus hallucination. However, there’s an important dis-analogy between them when it comes to the religious case, such that we needn’t settle this issue in order to assume that the memory component of the Memory/Reflection Account is reliable.

The difference is that, when it comes to the Memory/Reflection account, we’re warranted in maintaining the default assumption that reliable episodic memory processes are involved; when it comes to the Perceptual Account, however, we’re not warranted in assuming subjects are successfully perceiving rather than hallucinating. That’s because, in the perceptual case, we have some positive reason to take seriously the possibility that subjects might be hallucinating: because religious experiences are anomalous, we can’t simply assume that they’re successfully perceiving God. When the very question at issue is whether a reliable process is involved, the fact that religious experiences are abnormal makes hallucination a salient possibility. So, we can’t just proceed under the assumption, which would normally be in place by default for other perceptual experiences, that reliably functioning perception is involved rather than hallucination.

We don’t have the same sort of reason for questioning the default assumption that religious subjects are employing reliable episodic memory processes. The fact that the original religious experience was anomalous doesn’t give us any reason to think subjects are failing to employ the usual episodic memory processes when they later reconstruct that experience. In other words, whether or not the original experience was an instance of hallucination or reliably functioning perception, subjects can do equally well at later reconstructing the experience—we have no reason to think the episodic memory process is abnormal just because the contents of the experience it’s reconstructing were abnormal.

So, when it comes to the memory component of the Memory/Reflection Account, we can maintain the default assumption that subjects are employing reliably functioning episodic memory processes. This makes it different from the Perceptual Account, where we’re warranted in questioning the default assumption that subjects are employing reliably functioning perception.

4.2 The reliability of the reflective component

In this subsection, I draw on empirical evidence to consider three ways that one might reflect on memories of religious experiences. These aren’t meant to be mutually exclusive—different subjects could engage in different processes, while individuals could rely on multiple different processes. I also don’t claim to survey every possibility. Instead, I aim to conceptualize a few possibilities for how these processes might look, though future research may uncover others. Doing so will make it more concrete why I claimed above that the resulting beliefs aren’t simply based on episodic memories alone; instead, they result from more complex reflective processes about how to interpret those memories and whether to accept the original experience as veridical.

4.2.1 Coherence with one’s other beliefs, past experiences, self-knowledge

We often think of our episodic memories in terms of how they fit into a broader “autobiographical narrative,” which includes their coherence with other beliefs, past experiences, self-knowledge, and the like (cf. Debus, 2018). We might think about memories of significant past events not just as isolated episodes, but as parts of a coherent, overarching narrative; similarly, we might think about memories of our own past actions in terms of how they cohere with our broader self-conceptions. van Mulukom (2017) argues that this should be especially true of memories of religious experiences, since they tend to be vivid memories of significant life events. Similarly, Yamane (2000) surveys sociological research arguing that religious believers don’t typically have a singular “conversion experience” or “conversion event”; instead, the significance of a religious experience is interpreted via incorporation into a broader conversion narrative.

These sorts of coherences seem like they could factor into one’s reflections on whether to believe the contents of a remembered experience. That’s because, if you’re skeptical about the contents of some past experience, it seems like good evidence that it was veridical if it coheres well your background beliefs, other past experiences, and self-knowledge.

To see this, first imagine a non-religious case. Suppose you vividly remember an anomalous experience from your childhood, but that you’re unsure whether the experience was veridical or a dream—suppose, for example, that you remember seeing a miniature donkey in the living room of your childhood home. You then reflect on the memory to decide whether there ever really was a donkey in your childhood home, or whether you’re merely remembering a strange dream you had as a child. How might you go about engaging in such reflection?

One obvious way to do so is to reflect on whether your remembered experience exhibits the right kind of coherence with your other beliefs, memories, and self-knowledge. Suppose you realize you have beliefs that help to explain why a miniature donkey would’ve been in your house—for example, you believe that your eccentric uncle kept a miniature donkey as a house pet. You also realize that you have other memories related to this event—for example, you remember your parents telling you your uncle is going to bring his pet donkey over, as well as laughing about his eccentricities together after he left. All of these sorts of factors seem natural to include in your reflection about whether to believe or doubt that your experience of the donkey was veridical.¹¹

Luhrmann’s (2012) study of religious subjects suggests they often do something similar when reflecting about whether a past religious experience was veridical. Her subjects describe reflecting on whether an experience coheres well with other religious experiences they’ve had, both in terms of the experiences’ content and phenomenological qualities. They also describe reflecting on whether their experience fits well with their beliefs about God and his character—for example, on whether they have independent reason to think that God would want them to make the life decision he seemed to be encouraging. And they describe reflecting on whether a religious experience fits well with other aspects of their autobiographical memories, such as how they felt during and after the experience—for example, whether it resulted in a sense of peace.

Beyond the particular examples Luhrmann describes, it’s easy to imagine other forms this sort of reflection might take, since one might have various different kinds of background beliefs, past experiences, and self-knowledge that cohere well with the idea that God directly spoke to them on a particular occasion. One might have a strong background belief to the effect that, under the right circumstances, it’s possible for supernatural entities to communicate with people (as van Mulukom & Lang, 2021 note, the prevalence of such beliefs varies between cultures). One might also have a memory of a trusted friend or religious authority telling them that, if they participate in a certain ritual, it could enable them to get directly in touch with God. And one might take oneself to be the kind of person who is generally spiritual and open to these sorts of experiences. One could reflect and observe that, given all these other background factors, it doesn’t seem so far-fetched to think that God really was speaking to them during the experience they remember.

So, this sort of reflection seems perfectly natural. Furthermore, it seems that this sort of pattern of reasoning is a generally reliable one. Think again about the case of reflecting on your memory of the donkey. If it turned out that the content of this memory was completely isolated from any of your other memories or beliefs, that would seem like an indicator that it might’ve been a dream, since dreams are often discontinuous with our other memories and beliefs about the world. Given this, the converse also seems true: a high degree of coherence makes it more probable that your experience was veridical.

So, the religious person who reflects on these sorts of coherences seems to be instantiating a type of reasoning that’s typically a reliable guide to whether a past experience was veridical. Of course, this doesn’t entail that the conclusions they reach are accurate. It’s just to say that they’re employing a style of reasoning that’s in general truth conducive.

4.2.2 Corroboration with others’ experiences

When reflecting on memories of religious experiences, one might attempt to corroborate the contents of one’s experiences against that of others’ experiences. Specifically, to see whether a remembered experience was veridical, one might check whether other people had religious experiences whose contents cohere well with one’s own experience.

This is another method of reflection Luhrmann (2012) describes as occurring amongst her subjects from The Vineyard. These subjects’ religious experiences often occur while they’re alone, such as during moments of private prayer. In order to discern whether these experiences really involved God getting in touch with them, they discuss their experiences with others in the community—for example, whether the messages God seemed to be sending to people were the same, or at least whether they cohere well with one another. Corroborating the contents of experiences is especially important for subjects who have asked other people to pray for them about some important matter, such as asking for God’s help making an important life decision. If I’m unsure about the message I seemed to get from God, I might seek reassurance that a friend praying about the same matter received a similar message.

Beyond the specific community Luhrmann studied, it’s plausible that this would be a relatively widespread practice amongst religious subjects, at least in communities that place a shared value on openly discussing their religious experiences. That’s because, in general, if you’re unsure about whether some past experience was veridical, it’s natural to inquire about whether others had similar experiences. Take again the case where you’re doubting whether your experience of the donkey in your living room was veridical or something you merely dreamed. A natural way to try to verify that your experience was veridical is to see whether your family members also remember seeing a donkey in your living room. Analogously, if one is worried about whether an apparent experience of God was veridical, it’s natural to see if others experienced similar things.

As with the previous subsection, this also seems to instantiate a generally reliable way of reasoning. Typically, multiple people from the same community don’t experience the same dream or hallucination. So, the probability that a past experience was a dream or hallucination decreases as more people corroborate that they experienced the same sort of content.

Of course, this way of reasoning could be misleading in special cases where multiple people are likely to experience the same sort of dream or hallucination. Perhaps religious experiences are like this. Religious rituals are often conducive to achieving altered states of consciousness in which hallucinations can occur, and the contents of hallucinations can be shaped by cultural background beliefs and expectations (van Elk & Aleman, 2017; larøi2014culture?). The result may be that members of a religious community who share similar rituals and beliefs end up having similar hallucinations. However, this doesn’t negate the fact that, in general, the pattern of reasoning described in this subsection is typically a reliable one. This is consistent with thinking that, in religious contexts, following it could result in a false belief.

4.2.3 Biased interpretations of religious experiences

In the previous two subsections, we saw how reflections on memories of past religious experiences could instantiate generally reliable reasoning patters. I now turn to one possibility for how unreliable processes could be involved: certain cognitive biases might infect the ways people interpret their memories of religious experiences. Specifically, there’s some evidence that this reflection can be influenced by one’s desires or what one wants to believe, through processes such as confirmation bias and motivated reasoning.

van Mulukom (2017) suggests that reflecting on memories of religious experiences may be a process of constructing an interpretation that coheres well with one’s existing religious identity and sense of self—in other words, that one may interpret a religious experience based on what one already believes, rather than based on impartial reflection (see also van Mulukom & Lang, 2021). Some support for this is provided by Luhrmann’s (2020) finding that religious subjects often engage in experience-inducing religious rituals in part because they have deep desires for God to be real and to vindicate their existing religious identities. In other words, rather than always spontaneously undergoing a religious experience and then simply basing a belief on (a memory of) the experience, they purposely try to induce these experiences to confirm what they already (want to) believe.

If Luhrmann is right that religious experiences sometimes come about this way, then it’s not hard to see why reflection on past experiences could be biased by existing beliefs and identities. If a desire to come to a conclusion that fits with one’s existing religious self-conception is already salient and driving one’s behaviour, then it’s not a stretch to see how it would bias one’s reasoning. Perhaps, for example, one could be biased to focus only on details of a memory that confirm things one wants to believe, while discounting aspects of the memory that fail to do so.

Biases of this sort make our reasoning less truth-conducive, since it’s not truth-conducive to allow what one wishes were true to influence one’s beliefs. So, this sort of bias would be an example of an unreliable process.

Of course, such biases typically don’t totally take over one’s reasoning processes. Instead, they more subtly modulate our reasoning to drive it off course and make it less reliable to a degree (cf. Epley & Gilovich, 2016; Kunda, 1990). So, if such biases are indeed operative in reflection on memories of religious experience, it’s likely that they’re simply one process that combines with other kinds of reflection one might engage in.

4.3 Making progress on the Reliability Question

In this section, I first argued (§4.1) that the memory component of the Memory/Reflection Account typically involves reliably functioning episodic memory. Then (§4.2), I drew on empirical evidence to explore three different kinds of reasoning that could be involved in the reflective component. While the first two were instantiations of generally reliable reasoning patterns, the third pointed to biases that may make reflection unreliable.

Now, many of the issues explored in this section thus far ultimately depend on further empirical investigation. For one thing, my argument that the process of remembering religious experiences is typically reliable is open to being defeated by further empirical evidence that some sort of unreliable process is widespread. Furthermore, more empirical research is required to conclusively say whether the reflective processes I described are the ones typically involved in reflecting on memories of religious experiences, as well as whether there are others commonly involved that I haven’t considered.¹²

However, I want to close this section by suggesting that, at least at this stage of inquiry into the Memory/Reflection Account, we should be optimistic that we can make better progress on the Reliability Question than we could under the Perceptual Account alone. That’s because, to explore whether the processes involved in remembering and reflecting on religious experiences were reliable, we didn’t have to settle the same kinds of large, philosophical background questions that prevented us from making progress under the Perceptual Account.

For one thing, we didn’t need to settle any deep problems about individuating cognitive processes. In attempting to answer the Reliability Question, I simply focused on subjects who are reconstructing past religious experiences and then reflecting on them. This focus doesn’t force us to settle deep disagreements about whether perception and hallucination are instances of the same process, because the process of reconstructing a past religious experience and then reflecting on it would be the same kind of process whether or not the original experience was veridical or a hallucination. And, as long as no new reasons emerge for doubting that these subjects are employing reliably functioning episodic memory processes, we don’t have to settle deep disagreements about how to individuate episodic memory, confabulation, and/or misremembering.

Making progress on the Reliability Question also didn’t require us to settle the question of whether God really exists to be perceived. That’s because, again, describing the processes at work in the Memory/Reflection account doesn’t require us to know whether a subject’s original religious experience was veridical or not. Of course, we would have to settle the question of whether God exists to know whether beliefs resulting from a subject’s memory and reflection are true; however, it’s possible to conclude that the belief forming processes involved are generally reliable even if we don’t know whether the resulting beliefs are true.

So, it seems that we’re better able to make progress on the Reliability Question under the Memory/Reflection Account than we could under the Perceptual Account. Again, this isn’t to argue that we should outright reject the Perceptual Account in favour of the Memory/Reflection Account, since the Perceptual Account may still accurately characterize some cases of religious belief formation. Instead, my point is that, when we followed philosophers of religion in being fixated on the Perceptual Account, it prevented us from making progress on questions about the reliability of processes involved in religious belief formation. This changes once we enrich our conception of these processes to include the Memory/Reflection Account, too.

I haven’t shown that we can make progress on the Reliability Question for every case of religious belief formation. Since the Perceptual Account still accurately describes some cases, the same difficulties remain for answering the Reliability Question regarding those cases. However, we can at least be optimistic about making progress on this question for the cases which are accurately described by the Memory/Reflection Account. This is more than we could say when the Perceptual Account was all we had.

5 Conclusion

Both philosophers and scientists studying religion agree that religious experiences play some role in bringing about religious beliefs. It’s perhaps most natural to assume that they cause beliefs in the same way ordinary perceptual beliefs do, i.e., in an automatic, in-the-moment way. However, I argued that this Perceptual Account is importantly incomplete: while it may be that religious experiences sometimes cause beliefs this way, empirical evidence suggests religious belief formation is often a reflective process that’s extended in time beyond the original experience. I therefore proposed that we should supplement the Perceptual Account with the Memory/Reflection account, according to which religious beliefs are formed by reflecting on episodic memories of religions experiences.

Adopting the Memory/Reflection Account is theoretically fruitful, because doing so allows us to make progress on the Reliability Question, the question of whether religious beliefs are formed via generally reliable cognitive processes or reasoning patterns. Under the Perceptual Account, it’s very difficult to make progress on this question without settling some large, deeply entrenched philosophical disagreements lurking in the background. However, such disagreements don’t prevent us from making progress under the Memory/Reflection Account. Since there are open empirical questions about the exact processes involved in reflecting on memories of religious experiences, there’s room to continue exploring the Reliability Question in future research.

Acknowledgments

For helpful discussion and comments, thank you to Nicolas Crozatier, Felipe De Brigard, Shao-Pu Kang, Amy Kind, Chris McCaroll, Kourken Michaelian, Kengo Miyazono, Sarah Robins, Lu Teng, two anonymous reviewers for Philosophy and the Mind Sciences, and audiences at the Centre for Philosophy of Memory Public Seminar and the Successful and Unsuccessful Remembering and Imagining workshop.

References

---

1. Note that I don’t mean to imply that every religious belief is formed in response to a religious experience. Clearly, some religious beliefs are formed via processes such as testimony. By focusing on the Psychological Question in this paper, though, I bracket such cases and focus only on beliefs formed in response to religious experiences. I can’t hope to give an answer to the Psychological Question that captures every case of such belief, since the relationship between religious experience and belief may look different for different people. Still, we can investigate some of the most common ways this relationship looks. My arguments should be understood along these lines.↩︎

2. I take it that this sort of question should be of interest to epistemologists, since reliability often plays a central role in theories of justification and/or knowledge: some equate reliability with justification, but even non-reliabilists typically agree that reliability is necessary for knowledge.↩︎

3. Plantinga argues that, while exercising this “sensus divinitatis” sometimes involves experiences with the presentational phenomenology Alston describes, it doesn’t always involve perceiving God or experiencing his presence. Instead, on Plantinga’s view, this sense can operate in various ways, some of which aren’t perceptual. For example, it could work by producing feelings of guilt or gratitude towards God that make us aware of God’s existence in a more indirect way.↩︎

4. To clarify: I don’t mean to claim in this section that proponents of the Perceptual Account will be unable to make progress on any epistemologically interesting questions about religious experience. I leave it open whether, for example, we should define epistemic justification in terms of reliability or in a more internalist way that appeals merely to how things subjectively seem in one’s perceptual experiences (cf. Huemer, 2006; Pryor, 2000). If we adopt the latter understanding, we may still be in a position to make progress on the question of whether religious experiences can justify religious beliefs, since in that case what matters is whether they have the right kind of phenomenology rather than whether they’re reliable. Still, as per §1, I think the Reliability Question is independently interesting (including for non-reliabilist epistemologists—see fn. 2).↩︎

5. For one line of empirical motivation for thinking perceptual beliefs are formed in a way that’s automatic and directly in response to experiences, see Block (2023). It may be that some degree of explicit reflection on whether to accept an experience at face value is consistent with a resulting belief being genuinely perceptual. However, even if we grant this, this reflection would have to occur while one is having the experience; otherwise, the belief would be based on a recollection of the experience rather than the experience itself. When I talk about reflecting on past experiences in what follows, I have in mind cases where subjects reflect on memories of experiences that are no longer occurrent.↩︎

6. Of course, this might be because it’s not the religious experiences at all that cause these subjects to convert—perhaps there’s some other factor that, over time, eventually convinces them or wins them over? It’s likely that many different factors contribute to religious conversion: people might convert, for example, because they’ve found a religious group that gives them a sense of belonging and community, that makes them feel special, or that fills some other such hole in their lives. However, we should keep in mind Dawson’s (2006) criticism of scholars who focus too much on social and psychological motivations of this sort, thereby neglecting the role of religious experience. Dawson argues that such explanations only go so far, because, at the end of the day, only a few non-religious people end up converting, even though many of us have the same sorts of motivations to find community, to feel special, and the like. Furthermore, if you ask people why they converted, many firmly cite personal, firsthand experiences to explain their beliefs, experiences they take to have yielded some form of enlightenment or insight.

   In any case, it’s difficult to make sweeping generalizations about why people undergo religious conversion (or form the other kinds of religious beliefs to which I next turn in the paper). It’s likely that the processes involved are complex and vary between individuals, as with any complex domain of belief. It seems hard to deny, though, that at least some form their beliefs in part because of religious experiences they’ve had, given that, as Dawson notes, at least some religious believers claim this. So, even if some do form beliefs because of other social or psychological factors, my focus should be read as restricted to those for whom religious experiences are involved. It’s a further empirical matter exactly how widespread this involvement is.↩︎

7. At the same time, there’s likely variation between different religious communities about how often this sort of internalization occurs. For Luhrmann’s (2012) members of The Vineyard, for example, a period of reflection and discernment after a religious experience seems to be an important sociocultural practice for all members, rather than one that becomes less frequent over time. Some religious communities also place more of an emphasis on inducing religious experiences than others—consider, for example, the divide between “charismatic” Catholics who regularly engage in practices such as speaking in tongues and more traditional Catholics who largely shy away from such practices (cf. Schmalz, 2020). Religious experiences may always seem anomalous among groups like the latter, for whom they may always prompt a period of reflection to discern their meaning.↩︎

8. As I’ve described it, the Memory/Reflection Account involves reconstructing a prior experience that felt like an experience of God, then reflecting on whether it was veridical. However, subjects may sometimes engage in processes of memory and reflection that look slightly different than this: namely, they could remember a prior experience which they haven’t yet decided was a genuine religious experience, then reflect on whether or not it was a genuine religious experience. This could be because the experience was originally vague in some way, such that it was unclear whether it felt like an experience of God; or, it could be because the experience initially felt ordinary, but something one learned later triggered reflection on whether it was in fact an experience of the supernatural (see Barrett, 2000, p. 32, for discussion of such a case). My arguments in what follows are adaptable to cases like this. However, I’ll gloss over them for the sake of brevity. Thank you to Sarah Robins and Shao-Pu Kang for suggesting I consider such cases.↩︎

9. This assumption isn’t affected if we adopt the view, defended recently by several philosophers and psychologists, that memory and imagination are operations of the same cognitive capacity, and thus are instances of the same kind of cognitive process (Addis, 2018; Michaelian, 2016). On such a view, memory would simply be one function of a generally reliable capacity for imaginatively simulating past, future, and counterfactual events.↩︎

10. This is weaker than claiming that religious believers always construct their prior religious experiences accurately. Again, it’s possible for a reliable process to produce inaccurate results sometimes. It’s also not to claim that the original religious experience was an instance of reliably functioning perception; instead, it’s just to claim that one is reconstructing that experience via reliable memory processes, although that original experience itself could’ve been a hallucination.↩︎

11. Of course, it seems that most typical subjects wouldn’t explicitly conceptualize relations between these factors under the label “coherence relations,” since that seems like technical terminology ordinary subjects don’t employ in everyday life. Still, there’s an intuitive sense in which most people are capable of reflecting on whether the content of an experience fits well with other beliefs, memories, and self knowledge. So, one can at least be implicitly sensitive to coherence relations between a memory and these other factors.↩︎

12. An anonymous reviewer points out another potential source of skepticism about what I’ve argued: perhaps we can’t make general claims about the reliability of different patterns of reflective reasoning, given that these might exhibit varying degrees of reliability from person to person. Ultimately, I think this is yet another area that requires more empirical investigation, regarding whether the specific patterns of reflection I described in this section exhibit such individual differences. Even if they do, it may still be possible to make claims about their reliability in general, as long as that variation isn’t too drastic. Alternatively, finding variation in the reliability of how individual religious people form their beliefs would still allow us to come up with claims about the rates at which populations of religious believers employ reliable processes, rather than making claims about how reliable specific processes are in general.↩︎