Response Overview

I thank Alberto Cavallarin and Hans Van Eyghen, Lluis Oviedo, and Konrad Szocik for their commentaries on my book Religion as Make-Believe: A Theory of Belief, Imagination, and Group Identity (2023). I found several points they introduced to be thought-provoking and worthy of further consideration.

At the same time, many of the points the commentaries present as critical turn out not to be, once we understand my views. So let me start by clarifying some basic features of the theory I develop.

One marvelous thing about us humans is that we can relate to ideas in a wide variety of ways. In other (more technical) terms, we can have different attitudes toward any given content.

The variation in question applies even to the most humdrum ideas: one can factually believe it will rain tomorrow; one can hypothesize it will rain tomorrow; one can imagine it will rain tomorrow (as a part of make-believe play); one can assume just to be safe it will rain tomorrow; and so on. The underlined terms indicate just a few of the many ways people can relate to and process any given content (in this case: that it will rain tomorrow). I call the dimension of variation I’m highlighting the attitude dimension.

Importantly, the cognitive flexibility it takes to be capable of distinct attitudes appears in the religious realm. Consider the following four psychological states:

1. Esther doubts Joseph Smith saw the angel Moroni.

2. Barry hopes Joseph Smith saw the angel Moroni.

3. John playfully imagines Joseph Smith saw the angel Moroni.

4. Sam factually believes Joseph Smith saw the angel Moroni.

5. Brigham religiously credes Joseph Smith saw the angel Moroni.¹

Despite all mentally representing the same idea (that Joseph Smith saw the angel Moroni), Esther, Barry, John, Sam, and Brigham relate to that idea in different ways. Much research on religious belief focuses heavily on hypothesized features of the contents of religious ideation. This is true of various disciplines, but it has also been largely true of cognitive science of religion, one of the main areas to which I hope my book contributes. For example, cognitive scientists of religion have hypothesized the following content features of religious belief: religious ideas involve supernatural agents; such agents are minimally counterintuitive; these minimally counterintuitive agents have socially relevant knowledge; etc.² But while issues of content are important, a clear understanding of religious psychology requires that we describe what variations in attitude amount to as well. (How, for example, do Barry’s and John’s mental states differ?)

Religion as Make-Believe highlights this human cognitive flexibility, theorizes the space of cognitive attitudes,³ and aims to illuminate how differences in cognitive attitudes play out in religious contexts (as well as other contexts where group identity is salient). With that in mind, the two main theses of Religion as Make-Believe are as follows (appearing first on page 15):

Distinct attitudes thesis: factual belief and religious credence both exist and are distinct cognitive attitudes (they are two different ways of processing ideas).

Imagination thesis: religious credence differs from factual belief in many of the same fundamental ways that fictional imagining does—by “fictional imagining,” I mean the cognitive attitude that underlies pretend play.

The basic idea is that humans often have very different attitudes that might loosely get called “beliefs” in different contexts, and researchers would do well to distinguish them. Roughly: Factual belief is a matter-of-fact way of relating to ideas in which those ideas just seem like knowledge to the subject⁴ (whether or not they really are); one typically factually believes many things about the layout of one’s neighborhood, for example. Religious credence, on the other hand, is a reverential, identity-constituting way of relating to ideas that (if my arguments are on track) has deep cognitive similarities to imagining, albeit imagining that defines group identity (Chapter 6) and activates sacred values (Chapter 7).

To be clear, while religious credence as an attitude type is important and widespread, it is far from the only attitude that can be taken toward what might be thought of as “religious” contents, as the examples of Barry, Esther, John, Sam, and Brigham show. I put the point like this in my first chapter: “a mental state’s content, though heuristic, is never a decisive indicator of its attitude type” (Van Leeuwen 2023, 20). Mixing and matching of different attitudes with different contents (religious and otherwise) is both possible and common. And as I point out at the end of Chapter 3 (“Religious Credence Is Not Factual Belief”):

for any religious doctrine or story, it is likely that humans at large hold a range of attitudes toward it, since content and attitude vary independently, but one cognitive attitude that is both widespread and strikingly similar to fictional imagining is religious credence, which is far different from factual belief. (Van Leeuwen 2023, 97)

Since I introduce the construct of religious credence as an attitude notion and that is not definitionally meant to indicate content, it may help to think of the adverb religiously—which is a way that one might do, or relate to, a great many different things. After all, a running theme of my book is this: anything can be sacralized. I claim it is a virtue of my approach that it not only theorizes the differences between religious credence and factual belief but also does so in a way that does justice to the human cognitive flexibility I have been highlighting here.

In what follows, I first address a major misconception of my position that runs through all three commentaries. Clearing this up, fortunately, reveals that there is far less the commentators and I disagree about than it might have seemed. After that, I discuss the commentaries individually to address some of the points of disagreement that remain or appear to remain.

Here is the major misconception. All three commentaries rest their criticisms on the idea that I am saying that all religious beliefs are religious credences (in the sense I define) that are held for the sake of group identity. Cavallarin and Van Eyghen (2024, XXX) write: “The book leaves the reader with the impression that Van Leeuwen’s claims apply to all religious states, all the time.” Oviedo (2024, XXX) has a section of his piece pointing out that religious beliefs have more functions than those I suggest—“social functions that go much further than the simple Durkheimian paradigm of social identity formation”—and calls my view “reductive” for not addressing the other ones. That criticism assumes I have a totalizing view that rules out other functions than the ones I discuss. And Szocik (2024, XXX) claims I assume “that religious beliefs are the domain of only secondary cognitive attitudes” and criticizes me for holding that religious beliefs are “always . . . on a par with imagination and conjecture.”

The commentators’ terms “all,” “reductive,” and “always” present views that I myself explicitly reject in the book. As I just made clear, one of the main points of my work is to help explain the cognitive flexibility that enables humans to hold various attitudes toward any given content—religious or otherwise. While I am clear that I think religious credence, as a kind of psychological state, exists and is widespread, I am equally clear that it is an empirical question how widespread it is. From the start of Chapter 3:

[W]e shouldn’t be surprised if different groups of people or even individuals held different attitudes toward their respective religious and other supernatural ideas.

My stance is this: many (and probably most) people around the world have two-map cognitive structures for processing their religious ideas: a factual belief layer and a religious credence layer. But empirical exploration is required when it comes to any particular religious community to work out what attitude(s) people in that community have toward their stories and doctrines. Neither a Two-Map Theory nor a One-Map Theory should be the default stance; rather, we should adopt whichever particular theory best explains the relevant data and then expand our explanatory scope from there as our evidence base grows. (Van Leeuwen 2023, 67)

So, it is clear that I think religious credence is widespread and has great explanatory potential. But the empirical approach I advocate rules out adopting a totalizing view of any sort in early stages of investigation. And even if I were right that most religious minds deploy the sort of two-map cognitive structure I describe (factual belief and religious credence), that still allows for such minds to be doing many other things as well.

Here is a way to visualize my view. Let the class of particular mental states with (in some sense) religious contents be represented by the area inside the solid border. Let the class of particular mental states that involve the attitude of religious credence be represented by the area inside the dotted line. And let the class of particular mental states that involve the attitude factual belief (in the sense I characterize) be represented by the area inside the dashed line. This diagram, then, roughly captures the logic of my position.

Of course, we could add more areas for other attitudes (doubting, hoping, etc.) and for other content types (scientific, political, everyday, etc.). The important methodological point is that it is an extremely complex empirical question how much each attitude area overlaps with any given content area (and to what extent different areas shade into one another).

So, while I maintain that religious credence and its differences from factual belief are important, my views—far from being totalizing—are designed to capture human cognitive flexibility and to facilitate the formulation of clearer and more nuanced hypotheses concerning the fascinating space of cognitive attitudes—religious and otherwise.

To Cavallarin and Van Eyghen: There Are Plenty of Non-Hinge Religious Beliefs

Once the misconception just addressed is dispelled—my view is not actually a totalizing one—little disagreement between me and Cavallarin and Van Eyghen remains. I can allow that, among the myriad “beliefs” in the heads of religious practitioners around the world, some may well be hinge beliefs, as they claim. And Cavallarin and Van Eyghen already grant that religious credence is (at least in some ways) an illuminating construct.

But there are some points of sincere disagreement. To start, let us be clear that my notion of religious credence is not the same as the notion of hinge belief, even though Cavallarin and Van Eyghen attempt to recast it that way. After that, my objection to their position is this: a great many religious “beliefs” held by religious people around the world are implausible as candidates for being hinge beliefs. Thus, while I grant there may be some religious hinge beliefs, a great many religious “beliefs” are not. If that is true, then my notion of religious credence (not conflated with the notion of hinges) will still give extensive explanatory purchase in its own right and ought not to be recast as they propose.

A typical example of a hinge belief would be someone’s accepting (let that be a neutral term here) that the external world exists. Another would be someone’s accepting that the world existed before she was born. A third might be G. E. Moore’s example of accepting that I have two hands. The character of these hinges is threefold: first, one doesn’t have any more basic evidence from which the hinge propositions could be demonstrated; second, although hinges are in the first sense without evidence, one nevertheless lacks a serious framework from which it makes sense to doubt the hinge propositions (doubting them, as it were, seems silly); third, hinge beliefs are basic in that they provide a framework in which evidence for other beliefs even makes sense (e.g., my evidence that I’ll need more bricks to finish a wall makes sense in the framework of my hinge beliefs that the external world exists and that my visual capacities can detect it, etc.).

Ludwig Wittgenstein (1969) captures the relevant notion in an illuminating way in On Certainty paragraph 247: “What would it be like to doubt now whether I have two hands? . . . What would I believe if I didn’t believe that? So far I have no system at all within which that doubt might exist.” This suggests the following test for hinge beliefs:

System for Doubt Test: in order for p to be a hinge proposition for someone, that person must lack a framework or system of ideas they find plausible from which p may be doubted (in that it includes sense-making alternatives to p).

Otherwise put, if a person has a plausible cognitive framework that enables them to doubt p, then, even if they believe that p, that belief is not a hinge belief. (Example: I fully believe my front door is currently locked, but that is not a hinge belief, since I have a coherent system of thought within which doubt whether p might exist.) Let us now apply this test to a range of propositions that many religious people in some sense “believe”:

God exists

Demons exist

Prayer for healing works

The first chapter of Genesis is literally true

Jonah lived in the belly of a whale for three days

Jesus of Nazareth caused Lazarus to rise from the dead

The book of Revelation is literally true

I will grant that the first proposition, God exists, makes a plausible candidate for being a hinge proposition for some people. Whether it is so would depend on how the rest of their cognitive life is structured, and perhaps many people accept the idea of God’s existence in the same way that they accept the external world’s existence.

But it is important to note that, in point of psychological fact, many professed “believers” in God’s existence admit that they often doubt and sometimes have difficulty not doubting God’s existence, as Tanya Luhrmann (2012, ch. 9) and many others document. Not only does doubting not seem silly; they have a hard time dispelling doubt. So, for those believers, the proposition God exists fails the System for Doubt Test and hence does not qualify as a hinge proposition for them. And it is not hard to see what the relevant alternative cognitive systems might be within which the doubt can exist: there are plenty of systematic secular frameworks for thinking about what the world is like that do not include a deity; insofar as someone who is devout has cognized such frameworks and finds them plausible, they have a framework from which to doubt God exists, which makes that a non-hinge proposition for them.

Furthermore, the remaining propositions in the list I just gave can easily and sensibly be doubted even by most people who profess belief in them. “Are there really demons?” can sensibly be asked by pretty much anyone, so demons exist fails to qualify as a hinge proposition for the vast majority of people, no matter how religious. The same argument applies even more so to other items on the list. The proposition about Jonah, for example, lacks all three features of hinges and clearly fails the System for Doubt Test. That is not, of course, to say there is anything wrong with the Jonah story, only that it is not in any way a story made up of hinges. Many people have religious credence in it nonetheless, so many people have credences that are not hinges.

To speak generally, hinges tend to be beliefs concerning matters that are extremely basic, ontologically or epistemically (there is an external world; perception reveals objects). Many religious beliefs, by way of contrast, concern very specific doctrines and stories—often with florid details like those about Jonah—that accordingly do not make sense to regard as hinge beliefs.

Let me point to one other cognitive phenomenon that is relevant in this connection. In A Diagram for Fire, anthropologist Jon Bialecki (2017) writes the following concerning members of the Vineyard Church and their tendency to frame the same event sometimes in religious terms (prophecy) and other times in secular terms (intuition) [“diagram” is his technical term for a comprehensive epistemic framework]:

Vineyard believers must live in a secular world infused with countless other religious possibilities, including the possibility of there being no religion and no transcendence at all. . . . It is not surprising at all that when a more openly charismatic diagram decoheres, the next stable state that it collapses into should be a set of immanent relations in which the miraculous and God are not immediate forces. (Bialecki 2017, 169, emphasis added)

Bialecki calls this tendency to have two frameworks that one switches between “double coding.” I discuss the phenomenon extensively in Chapters 3 and 4 of my book. The relevance to the present discussion is this: insofar as Vineyard members operate with a “double code”—one secular, one religious—they have at least a coherent system of ideas from which to doubt their religious beliefs. Those beliefs, as important as they may be to those who hold them, thus cannot be hinges for them according to the System for Doubt Test.

None of this, again, is to say that the notion of hinge belief has no place in the general study of religious belief. But I suspect it will end up taking up far less of the canvas than Cavallarin and Van Eyghen think. My hypothesis is that much of the canvass that hinges do not take up will be taken up by religious credences (in my intended sense).

To Oviedo: Maps Aplenty

Oviedo’s commentary contains a number of interesting proposals that I find welcome. Once we jettison the false impression that my views are “reductive,” it is clear that I have room for these proposals, some of which bolster the usefulness of my conceptual framework. Importantly, insofar as my approach can help inject more clarity into points Oviedo is making, his intended criticisms in fact end up advertising the theoretical utility of my ideas. I illustrate this point in relation to two of his attempted critiques.

First, Oviedo, in a section entitled “Too Many or Too Few Maps?”, writes: “A first problem arises with the description of this double cognitive map. In my opinion, there are too few.” Later he writes:

[W]e use a different cognitive map when we try to negotiate a loan in a bank; when we do neurological research; when we campaign for a social or political cause; when we try to convince someone of our love for him or her; or when we immerse ourselves in a concert or visit an art exhibition. (Oviedo 2024, xxx)

This is in some sense obviously right. But, importantly, this list Oviedo gives conflates two senses of “map.” One sense is just something like a conceptual framework within a certain content domain. That would be a content sense of the word “map.” Another sense—the one I develop—is a distinct layer of processing that can apply to a given set of ideas. That would be the attitude sense of the word map that I develop in the book.

The list Oviedo gives highlights distinct “maps” in the first (content) sense; this exhibits no tension whatsoever with anything I say, since of course I grant there are various content domains and conceptual frameworks. But does that mean that there also different attitude maps corresponding to items on Oviedo’s list—distinct attitudes or manners of processing? Well, the existence of different content maps does not imply a corresponding number of different attitude maps, since attitude and content are independent. Be that as it may, my view is indeed structured to be able to describe many different cognitive attitudes—people, again, can relate to ideas in all sorts of ways. My framework is meant to characterize the space of cognitive attitudes, as I emphasized above.

Why, then, do I use the phrase “two-map cognitive structure” so often if I agree that there are far more than two cognitive attitudes? As my discussions in the Prologue and in Chapter 2 make clear, my phrase “two-map cognitive structure” describes how factual beliefs and a given secondary cognitive attitude can both be implicated in guiding the same action at a given time, without collapsing into one another: there are two parallel layers of cognitive guidance (e.g., make-believe play is guided both by factual belief and by playful imagining at the same time, a two-map cognitive structure). So, the phrase, pace Oviedo, does not imply that there are only two cognitive attitudes that humans are capable of!

Importantly, my framework is designed to be able to characterize any cognitive attitude, even ones I don’t explicitly address in the book. It is possible, for example, that some neuroscientists could have a special way of entertaining hypotheses that deserves to be thought of as an idiosyncratic secondary cognitive attitude in its own right. Far from ruling it out, the theory I present in Chapter 2 could be used to describe such a possibility. As I take care to point out, the attitudes I discuss (like factual belief and religious credence) are attractor positions—not monoliths—in a much larger space of cognitive possibilities.

Thus, what Oviedo thinks of as a criticism (“too few” maps) is in fact a friendly suggestion for extensions of my theory—extensions that can be stated more crisply using the framework I propose.

Second, when it comes to the functions of religious beliefs, Oviedo calls my theory “reductive” for focusing on group identity. But we have already seen that my view leaves open that there can be other functions. Emphasizing an aspect of Luhrmann’s work that highlights the “therapeutic” aspects of religious belief, Oviedo writes:

These experiences [of healing] have been much more studied in recent years in a growing body of scientific literature on religion and health, under the labels of “religious coping,” “religion and resilience,” “religion and well-being,” or “religion and flourishing” . . . What does all this research add to our understanding of religious “credences”?

Many studies show that only more intense and shared forms of religion achieve these healing properties. Simply believing in God is of little effect unless it is translated into attitudes nourished by celebration, prayer and other engaging practices. (Oviedo 2024, xxx)

This is all plausible. But importantly, far from being at odds with it, frameworks for understanding religious belief that emphasize group identity—such as mine or Émile Durkheim’s ([1912] 2008)—can help explain these phenomena. It is not the dry cognitive representation of God that does the most therapeutic work, it is what one does in building community with others that has the greatest therapeutic effects. For understanding this phenomenon in greater empirical depth, I recommend the work of another neo-Durkheimian, Dimitris Xygalatas (2022), who summarizes his research program in his book Ritual: How Seemingly Senseless Acts Make Life Worth Living. That work gives empirical validation to Durkheim’s notion of collective effervescence.

Importantly, however, none of that commits me to the position that constituting group identities is the only thing religious credences do for people. Yet it is likely that their other non-Groupish functions, whatever they are, also pressure them to be different from factual beliefs. I write the following in Chapter 6:

Being Groupish is unlikely to be the only pressure on religious credence to differ from factual belief. Religious credences may also have non-Groupish imaginative functions in the lives of individuals, like enabling them to have certain personal experiences that they might not have had otherwise, as theorists from William James to Tanya Luhrmann have emphasized. One may simply find many aspects of life more meaningful when one gives them an imagined supernatural gloss. But this is compatible with the perspective of this chapter: such supernatural glosses are not likely to come from evidentially constrained factual beliefs, so the imaginative role that religious credences play in “personal religion” (to use James’s phrase) most likely also pressures them to have properties that constitute them as secondary cognitive attitudes as opposed to factual beliefs. (Van Leeuwen 2023, 172)⁵

In sum, Oviedo’s main criticisms turn into interesting complications that my views, properly understood, can put into sharper relief.⁶

To Szocik: Cognitive Flexibility Is the Point

Szocik’s discussion of my book is the most polemical of the three commentaries. It also contains the most distortions—distortions on which the criticisms rest. So let me start by clarifying three basic points about Religion as Make-Believe for the present reader.

1. My book presents a descriptive theory of some significant aspects of religious psychology. As such, is it is not a normative appraisal or criticism.

2. Though it focuses on the attitude dimension of psychological states, my book contains ample discussion of the various ways mental state contents are relevant to psychological processing (religious and otherwise).

3. My book is designed to capture the fact that people can have various attitudes concerning religious ideas (or any ideas). Capturing this cognitive flexibility is one of my main aims.

With those points in mind, we can better evaluate Szocik’s criticisms.

In his title, Szocik calls my book “a failed reactivation of Enlightenment criticism of religion.” But as just stated (point 1), my book is a descriptive theory of certain aspects of people’s psychology. It is not a normative criticism of anything. It is true that in the Epilogue I venture some very brief criticisms of certain self-deceptions that my work makes possible to characterize. But none of that is a criticism of religion in general; I make it clear that that form of self-deception occurs outside religion as well. Relatedly, elsewhere in the commentary Szocik casts what I am doing as a “debunking” argument. In no place in the book do I make a debunking argument, as that term is generally understood.⁷ So Szocik misconceives the basic purpose of my book from the start. Also, when it comes to “Enlightenment criticism of religion,” it is opaque which Enlightenment figures he has in mind; he does not say. And though it is entirely interesting, Enlightenment criticism of religion is not even a topic in my book.

Also in the title, Szocik writes, “content, not cognitive attitude, makes the difference between factual beliefs and religious credence.” This is a false dichotomy (and that I think it is a false dichotomy follows from point 2 above): For any contentful mental state, both its general manner of processing (attitude) and its content⁸ influence its downstream effects. Furthermore, mental states that have religious contents can involve varying cognitive attitudes, as Szocik himself explicitly says later in his piece (see the following quotation), so it cannot be even for him that content is the only difference maker.

This brings us to Szocik’s most serious distortion (the one that confuses point 3).

The methodological error of Van Leeuwen’s philosophy is the assumption that religious beliefs are the domain of only secondary cognitive attitudes. . . . Perhaps this is the biggest metatheoretical error committed by Van Leeuwen. It would be appropriate to correct this error by proposing the following assertion in place of the aforementioned thesis. People, both believers and non-believers, can adopt different cognitive attitudes to different types of beliefs. (Szocik 2024, xxx, emphasis added)

I emphasized the last sentence, because (in addition to being at odds with Szocik’s own title) it is essentially something that I myself say in multiple places, since (as I have been saying throughout this response) one of the most important points of my work is to emphasize how attitude and content vary independently. Getting readers to be clear on that independent variation is largely the point of Chapter 1. And on the topic of religious “belief” in particular, I wrote this toward the end of Chapter 3 (also quoted in the general discussion above):

for any religious doctrine or story, it is likely that humans at large hold a range of attitudes toward it, since content and attitude vary independently, but one cognitive attitude that is both widespread and strikingly similar to fictional imagining is religious credence, which is far different from factual belief. (Van Leeuwen 2023, 97, emphasis added)

Just compare the italicized lines in the two preceding quotations; they are basically different formulations of the same point.

So, to review, the “biggest metatheoretical error” Szocik can attribute to me is one that my framework is designed not only to avoid but to correct. And his suggested way to “correct” this error involves his proposing an “assertion” that is a notational variant of a point that I make explicitly in numerous places in the book as a whole.

I can only say that I am gratified to realize that Szocik, at the end of the day, in fact agrees with one of the major points of my book.⁹

Conclusion: A Welcome Discussion

Once again, I thank Alberto Cavallarin and Hans van Eyghen, Lluis Oviedo, and Konrad Szocik for their commentaries. I hope I have been able to dispel the misconceptions on which their main criticisms rested and thereby to have clarified my views. More than anything, I am glad we share an appreciation of this important topic.