A Bayesian Exploration of C.S. Lewis’s ‘Argument from Desire’

Abstract

C.S. Lewis’s ‘Argument from Desire’ is best summed up by his famous line, ‘If I find in myself a desire which no experience in this world can satisfy, the most probable explanation is that I was made for another world’. In short, unfulfilled ‘seemingly transcendent desires’ point to fulfilment in another realm. Lewis’s argument is fraught with disagreement, and subsequently, questions remain as to its efficacy as a theistic argument. In this essay, I will take a novel approach by using Bayes’ theorem to assess the success of Lewis’s argument in support of theism over and against naturalism. As I will argue, a strength of this approach is that it allows me to largely bypass a significant area of disagreement within the literature, namely, whether these desires are best seen as natural or artificial. Additionally, I will argue for why these desires are the sort of good God would be interested in giving to persons. Then, I will proceed to outline what I take to be the best naturalistic accounts of these desires and outline some of the challenges these accounts face. Finally, I will conclude with plugging these results into Bayes’ theorem.

Introduction

As an argument for the existence of God, C.S. Lewis’s ‘Argument from Desire’ stands in a category of its own. While it shares the common trait that other arguments for the existence of God share, namely, significant disagreement, what makes Lewis’s argument unique is the polarization of disagreement over his purported argument. Indeed, even by referencing the argument, I have already taken a stand in the debate. On one hand, Lewis commentator Peter Kreeft comments that outside of Anselm’s ontological argument, Lewis’s ‘Argument from Desire’ is the most interesting argument ‘in the history of human thought’ (1980: 201). On the complete other hand, Arend Smilde notes that Lewis never used the phrase ‘Argument from Desire’ in his entire corpus, pointing out that the phrase was first coined by John Beversluis in 1985,¹ twenty-two years after Lewis’s death (2014: 69). Combined with the fact that the Cambridge Companion to C.S. Lewis does not mention anything about an ‘Argument from Desire’, Smilde argues it is legitimate to question if Lewis ever intended his ‘argument’ as such (2014: 81).

With recognition that the ‘argument’ is not one, in the formal sense of the word, a sort of abductive ‘Argument from Desire’ can be constructed on the basis of Lewis’s remarks in at least four of his major works: The Pilgrim’s Regress, Mere Christianity, The Weight of Glory (a sermon), and Surprised by Joy (Lewis’s autobiography). In its arguably most concise form in Mere Christianity, the ‘Argument from Desire’ can be summed up in Lewis’s quote, ‘If I find in myself a desire which no experience in this world can satisfy, the most probable explanation is that I was made for another world’ (1997: 118). Clearly this is not a syllogistic argument arguing from premise to conclusion, but more so an inference to the best explanation from a given phenomenon of desire. Now certainly, when looking at Lewis’s corpus as a whole, one could try and construct a more formal ‘Argument from Desire’ into syllogistic form. One such example of a possible deductive approach to Lewis’s argument is given by Robert Holyer:

1. Since all natural desires are desires for an object that satisfies them, ‘Joy’ is a desire for some object.

2. ‘Joy’ is not a desire for any finite object, since no finite object satisfies it.

3. Since all desires have existents that satisfy them, ‘Joy’ too has an existent which satisfies it. Therefore, (combined with 2) an infinite object exists. (1988: 68)

However, given that Lewis never himself formulated a syllogistic ‘Argument from Desire’, and further, given the widespread disagreement over the premises involved in such a syllogistic format,² it seems a more plausible way through the disagreement over Lewis’s argument is to assess the merits of Lewis’s argument using Bayes’ theorem. Regardless of whether Lewis even intended his argument as such, an argument has emerged in the subsequent literature and therefore can be assessed as such. By using Bayes’ theorem to assess Lewis’s argument, I can assess whether Lewis’s argument is a ‘good’³ argument for theism, whilst side-stepping the inevitable conflict that arises from syllogistic formats of the argument.

Preliminaries

Before proceeding, it is necessary to clarify that whilst side-stepping the disagreement that results from syllogistic forms of Lewis’s argument, like Holyer’s above, using Bayes’ theorem will doubtlessly generate its own disagreement, specifically in the assignment of prior probabilities I use in assessing Lewis’s argument. In agreement with Richard Swinburne, in using Bayes’ theorem, ‘it is impossible to assign anything like exact numerical values to the probabilities involved’ (2004: 341). Inevitably, there is an element of subjectivity in the assigning of probability values. Whilst I will do my best to construct the probability values based upon the relevant arguments in support, I recognize that the value assignments could differ depending on the person involved. With that said, I will purposely strive to be conservative in my estimations. Prima facie, it may seem to many that the existence of these seemingly transcendent desires that Lewis references would be more likely on theism than naturalism.⁴ From this, one could easily over-state the case for theism. And consequently, given the phenomena of this specific type of desire, one could easily under-state the naturalistic account. Therefore, I want to clarify at the outset, that I will strive to conservatively estimate my probabilistic valuations, which may doubtless frustrate those who think I have under-stated either case.

Further, I recognize that using Bayes’ theorem as a means of assessing the strength of Lewis’s argument as a positive argument for theism is treating Lewis’s argument in a way differently than both Lewis and his commentators have treated the argument. Lewis’s argument is not that the experience of what I will call ‘seemingly transcendent desires’ raises the probability of theism; in contrast, Lewis’s argument is that the experience of ‘seemingly transcendent desires’ makes it likely that a transcendent object of fulfilment, what Lewis took to be Heaven, exists. So Lewis’s argument is making the much stronger claim that the experience of “seemingly transcendent desires” gives us good reason to think the transcendent realm exists. However, as Lewis’s argument seems to be implicitly dealing with likelihoods, namely, that “seemingly transcendent desires” make it likely that the transcendent realm exists, I see no reason why Bayes’ theorem cannot be used to assess how strong this likelihood is. Clearly Lewis thinks it to be quite strong. This essay will aim to assess just how strong.

In assessing to what extent Lewis’s argument is evidence for theism, I will, following Ted Poston, ‘take it for granted that we can make comparative claims about the probability of various competing hypotheses and the probability of some items of evidence on those hypotheses’ (2018: 373). With E representing the seemingly transcendent desires Lewis speaks of, and H representing theism, then Lewis’s argument will be positive evidence for theism, if E raises $\mathrm{Pr}(H|E)$, and consequently, will be positive evidence for naturalism if E lowers $\mathrm{Pr}(H|E$). This method of probabilistic reasoning is otherwise known as ‘likelihoods’, where the likelihood is the probability that some evidence ‘e’ is true given some hypothesis ‘h’. One common example of this sort of likelihood probability is flipping a coin. On a fair coin, the probability that the coin lands heads, or lands tails, is equally 0.5.⁵

Thus, in order to assess the strength of Lewis’s ‘Argument from Desire’ in support of theism, I will leave aside any other evidential considerations that could impact the prior probabilities of both theism and naturalism.⁶ Instead, I will stipulate that the prior probability of theism and naturalism is equal—50:50. To be clear, as this essay is assessing the extent to which Lewis’s ‘Argument from Desire’ raises the probability of theism as opposed to naturalism, these prior probability assignments are done simply for the sake of convenience. Further, and, continuing to use Poston’s example, for the purposes of the following Bayesian assessment, I will assume the useful falsehood that theism and naturalism are mutually exclusive and exhaustive (2018: 374).

The Phenomenology of Desire

As this essay seeks to assess the likelihood of a specific phenomenon of desire, what I have so far called ‘seemingly transcendent desires’, on theism and naturalism, it is necessary to try and articulate what exactly I mean when referring to these desires. As will be shown, a concise definition of these desires that Lewis called ‘Joy’ is hard to come by. With that said, in attempting to summarize Lewis’s various descriptions of the experience of these desires, I will treat ‘seemingly transcendent desires’ as continuously unfulfilled desires which appear to have a transcendent referent.⁷

As I’ve just alluded to, Lewis’s word for this desire, which ‘nothing in this world can satisfy’, was ‘Joy’ (2012: 12), which is not to be confused with other ordinary English uses of the word, i.e. happiness or good fortune. According to Lewis, ‘Joy’ is an unsatisfied desire that is, itself, more desirable than any other satisfaction. Per Alister McGrath, ‘Joy’ is perhaps philologically closest to the German word, sehnsucht, which German Romantic writers used to describe a deep yearning for a seemingly unreachable state of life, or, a poignant desire for something agonizingly elusive (2014: 106).

This theme of ‘Joy’ being agonizingly elusive can be clearly seen in Lewis’s biography. Despite having continually sought satisfaction for this desire, Lewis was continually let down in his quest; and it was this experience of continual unfulfilment of his desire that led Lewis to conclude, not that satisfaction for this desire did not exist but that fulfilment of the desire in question was to be found in ‘another world’, what Lewis took to be Heaven.⁸ As might be obvious, such a conclusion was pivotal in his own eventual conversion to Christianity.

As one might expect, Lewis’s experience fits well within a theistic worldview, as the theme of desiring Heaven and/or God, or, perhaps more minimally, ‘seemingly transcendent desires’ can be seen in Augustine, Aquinas, and Pascal, to name a few (Stump, 2010: 440). However, it would be wrong to suppose from this, that the experience of ‘seemingly transcendent desires’ is a specifically theistic experience, or, more broadly, an experience for those who ascribe to religious practice of some form. In fact, several of the foremost atheists in the past two centuries appear to have had similar experiences to Lewis’s. Consider, for example, Bertrand Russell, who, in a letter to his mistress, said,

the centre of me is always and eternally a terrible pain…a searching for something beyond what the world contains, something transfigured and infinite - the beatific vision, God - I do not find it, I do not think it is to be found - but the love of it is my life…it is the actual spring of life within me. (2001: 79)

Or take for instance, Friedrich Nietzsche, who, as John Cottingham has argued, seems to have written about this experience of ‘seemingly transcendent desires,’ in Zarathustra:

Deep is her woe –

Joy – deeper than heart’s agony;

Woe says: ‘Be gone!’

But all joy wants eternity –

Wants deep, deep eternity. (Cottingham, 2019: 379-380; Nietzsche, 1954: §12)

Others like Camus and Sartre could be mentioned,⁹ but suffice to say that the experience of ‘seemingly transcendent desires’ cannot be written off as an exclusively theistic experience, as clearly some of the most well-known atheists of the last two centuries have seemingly experienced them. Given, then, the prevalence of these sorts of desires within these theistic and atheistic writers, and, given that a significant majority of the world’s population could be classified under one of these two categories, it would seem that this experience of ‘seemingly transcendent desires’ is, in reality, a fairly common human experience. Obviously, one cannot know for certain just how widespread these desires are, but for the purposes of this paper, all I aim to establish is that the experience of these desires is widespread. Given the precedent already noted, it seems such a conclusion is not without warrant.

Now naturally, this then leads to the much-contested issue within Lewis scholarship as to what sort of desire, these ‘seemingly transcendent desires’ are. Are these desires innate/natural or secondary/conditioned? With regard to natural desires, John Haldane notes three significant markers:

1) Spontaneity of occurrence; 2) prevalence to the extent of normal universality; and, 3) common linguistic identification of types of desire and/or of their satisfaction, and/or of their deprivation. The desire for food and sex occur without cultivation; are prevalent to the extent that their absence generally invites explanation; and natural languages have words for these desires and/or for their fulfilment and/or their frustration.

In contrast to these markers of natural desires, artificial desires are (1) acquired or inculcated, (2) liable to be non-universal and culturally variant; and (3) not ubiquitously linguistically represented (2007: 49).

The reason that this question of whether these ‘seemingly transcendent desires’ are natural or artificial is so central, is that much of Lewis’s justification for his “Argument from Desire” seems to be based upon the assumption that the desires in question are natural.¹⁰ Since Lewis believed all natural desires have corresponding objects of fulfilment, Lewis inferred that his ‘seemingly transcendent desires’, being natural desires, must have an object of fulfilment. This inference, in particular, from natural desires to a corresponding object of fulfilment, has generated significant disagreement in the subsequent literature.¹¹

To be clear, one of the strengths of using Bayes’ theorem to assess the efficacy of Lewis’s argument is that, as far as I can see, the answer to this question of whether ‘seemingly transcendent desires’ are natural or secondary has no bearing on the outcome of a Bayesian comparison. Consider: ultimately all one is doing in employing Bayes’ theorem is assessing whether evidence, E, in this case ‘seemingly transcendent desires’, raises the probability for hypothesis, H, which in this case, is theism. Prima facie, it is difficult, for me at least, to see how these desires being natural, instead of artificial, at all changes the probabilities involved. If ‘seemingly transcendent desires’ were something God desired to instil in created persons, then it seems like He would be equally capable of producing those desires naturally or artificially. The important question for a Bayesian exploration is how likely these sorts of desires are, rather than what type of desires they are. Mutatis mutandis, the same applies to naturalism. Whether ‘seemingly transcendent desires’ are natural desires, albeit confused ones,¹² or they are artificial, it seems the central question is whether there is a good naturalistic explanation for the origination of, and the evolutionary benefit¹³ of, ‘seemingly transcendent desires’. Either way, it seems the contentious debate between natural and artificial desires can be avoided for my purposes.

With that said, and, with a nod to the broader scholarship within which this essay takes place, it seems unavoidable to take a stand on the issue. Therefore, I will briefly argue for why I take ‘seemingly transcendent desires’ to be natural desires, with the caveat already noted that whether these desires are natural or artificial has no bearing on the outcome of my Bayesian exploration.

A Brief Excursus: ‘Seemingly Transcendent Desires’ as Natural Desires

It is beyond the scope of this essay to try and engage all of the various authors and arguments against ‘seemingly transcendent desires’ as natural, so I will simply focus on one recent contribution to the literature by David McPherson. McPherson argues that the sorts of desires that theists typically associate as desires for Heaven and/or God, what I’ve been calling ‘seemingly transcendent desires’, are not natural, but secondary desires, as they do not ‘come about without some acquired abilities, dispositions, and sensibilities, which depend upon an upbringing or formation in a particular culture or form of life’ (2019: 393).

As McPherson affirms, secondary desires ‘depend upon an upbringing or formation in a particular culture or form of life,’ it would seem safe to conclude that non-secondary desires, natural desires, do not depend upon an upbringing in a particular culture or form of life. Though I am not sure I agree with McPherson’s argument that desiring a ‘complex concept’ like God depends upon a specific cultural upbringing (2019: 391), I am happy to grant it in this instance, as it does not demonstrate that the ‘seemingly transcendent desires’ which are the focus of this essay are also secondary desires. Just because specific examples of ‘seemingly transcendent desires’, i.e. desires for ‘God’, may be culturally dependent, it does not follow that the phenomena of ‘seemingly transcendent desires’ in the broadest construal are themselves culturally dependent. In fact, given the presence of these sorts of desires listed above in the works of prominent atheist thinkers, and the wide range of theistic belief and practice, it seems just the opposite: ‘seemingly transcendent desires’ in the broadest construal are culturally transcendent. Therefore, whereas desires for God specifically conceived may be culturally dependent, and people might disagree as to the content of the referent of these desires, the desires simpliciter seem to be culturally transcendent. Which is one reason why, contra McPherson, I think these desires are best treated as natural desires.

Now surely one might rightly ask the question, ‘If “seemingly transcendent desires” are natural, how come not everyone experiences them?’ It seems to follow based off commonly held intuitions, that in order for a desire to be natural, it must be universal, i.e. everyone has them. However, I think, following Haldane, it is necessary to qualify what we mean by universality. One can assert that natural desires are prevalent to the extent of normal universality, viz., their absence generally invites explanation, without having to hold that absolutely everyone, everywhere experiences these desires. To illustrate why, consider sexual desire. It goes without saying that sexual desire is a natural desire. But is it universal in the sense that everyone, everywhere experiences it? No. Recent research from Anthony Bogaert suggests that 1% of the of the global population is asexual, meaning that roughly seventy-six million people experience a lack of sexual desire (2015). The significance of this statistic for my argument is that at least one natural desire, sexual desire, is transcultural, and is prevalent to the extent of normal universality, despite it not being universal in the sense that it applies to everyone, everywhere, as clearly there are millions who do not experience sexual desires. But does that in any way undermine our viewing sexual desire as a natural desire? Presumably not. From this then, I see no reason to conclude from the lack of ‘universality’ (in the sense of the word as applying to everyone, everywhere) of ‘seemingly transcendent desires’ that they cannot be considered natural desires. As such, arguments against ‘seemingly transcendent desires’ being natural, on the basis that not everyone experiences them seem to not carry as much weight. As long as ‘seemingly transcendent desires’ are widespread to the extent that their absence generally invites an explanation, which I take to be the case, then they appear to fit the criteria for universality that applies to other natural desires. Absent of an argument which demonstrates that the universality of ‘seemingly transcendent desires’ is relevantly dissimilar to the universality of other natural desires, like sexuality, I see no reason to preclude ‘seemingly transcendent desires’ from being natural desires.

Theism and ‘Seemingly Transcendent Desires’

For reasons I will explain and reiterating my attempt to be conservative in my estimates, I take the $\mathrm{Pr}(E|H)$, where E is ‘seemingly transcendent desires’, and H is theism to be 0.80, or, 80% likely that seemingly transcendent desires would exist given the truth of theism. My justification is as follows.

As has been noted above, persons having ‘seemingly transcendent desires’, what theists typically understand as desires for Heaven and/or God, is a theme seen throughout the history of theology, ranging from Augustine, to Aquinas, to Pascal, to Lewis. Therefore, I do not view it as particularly ground-breaking, or controversial, to assert that the existence of ‘seemingly transcendent desires’ is more likely than not given the truth of theism, not only because it has intuitive appeal, i.e. people have transcendent desires because the transcendent realm exists, and it contains the summum bonum of human life, but also because many of the brightest minds throughout the centuries have thought similarly. To be clear, I am not asserting that these theists are known for thinking that the probability of transcendent desire is high given the truth of theism. None, to my knowledge, asserted anything of the sort. But I am asserting that clearly all of these theologians believed theism to be true, and that ‘seemingly transcendent desires’ made sense within the theistic worldview they held. And as for Lewis in particular, these desires made so much sense on the theistic picture, that they were a central reason for his conversion to theism. In sum then, though these writers did not write about the presence of ‘seemingly transcendent desires’ in probabilistic terms, they all affirmed the presence of these desires given the truth of theism. Thus, following these theists, in setting $\mathrm{Pr}(E|H)$ at 0.8, I am asserting that if theism is true, these desires can be expected as more likely than not.

However, justification is still needed for why this might be the case. As space constraints do not allow for me to cover each of these authors’ arguments in full, I will simply present an argument that could be used to support this thesis; namely, if theism is true, what we take to be ‘seemingly transcendent desires’ are more likely than not to exist. Thus, even if one were to disagree with the specific argument I present, it does seem that these authors’ combined support, plus the intuitive appeal of the presence of these desires in a theistic worldview, presents strong support for the notion that if theism is true, these desires are more likely than not.

Given the existence of God, and the typical omni-properties associated with God, particularly His being omnibenevolent, it seems intuitive to think that the sort of relationship God would desire to have with created persons would be one of love. Leaving to a side the question of why God has not chosen to more fully reveal Himself, it would seem, given ‘an even modest doctrine of divine transcendence’,¹⁴ that created persons are in need of some sort of revelatory clues to God’s existence. After all, it would seem that a precondition for having any sort of personal relationship with anyone is one has either interacted with the other person in question, or, more minimally, one has some knowledge about the possible existence of said person.

So assuming that God has overriding reasons for remaining hidden,¹⁵ it would seem that God has various options available to Him in order to obtain a relationship of love with created persons. One option would be to imbue all persons with a general knowledge of His existence in order to ensure that everyone would have some knowledge of His existence, as that would at least ensure that all persons have the opportunity of participating in a loving relationship with Him. However, it would seem that knowledge of God’s existence simpliciter is not sufficient for obtaining the sort of relationship God is after. After all, just knowing God exists does not guarantee He desires a relationship with us, and even if it did, it is quite conceivable that a majority of persons would choose to reject a relationship with God if all they knew about Him was merely that He existed.

Furthermore, it would seem that an all-loving God would be interested in maximizing the amount of loving relationships He could have with persons. As such, it would seem that God has a reason to give us more than knowledge of His existence simpliciter, as it seems He would desire to ‘stack the deck’ in His favour. What I mean by this is that not only is a loving relationship between Himself and human persons the good that God is after, it would seem to be the case that this sort of relationship would be, if not the highest good of created persons, at least one of them. Therefore, God, as an all-loving, omnibenevolent God, has an overriding reason to try and make the outcome of a loving relationship between Him and a maximal amount of human persons more likely than not. It is what He desires, and furthermore, it is a great good for us. And ‘stacking the deck’ in His favour allows Him to try and obtain a specific outcome while still allowing for the free will of created persons, which is also a loving thing to do.

Having suggested the need for God to do more than very generally reveal His existence, in order to obtain a loving relationship with humans, it would seem that God would be interested in imbuing persons with some sort of innate capacity to not just know of His existence, but an innate capacity to know what He is like; something akin to a modified sensus divinitatis.¹⁶ To be clear, all I am arguing is that I think there is warrant for thinking God would imbue persons with something like the sensus divinitatis, at minimum, some sort of widespread human capacity that facilitates knowledge of the transcendent realm and what that transcendent realm might be like.

The markers of ‘seemingly transcendent desires’, as I have argued, are that they are widespread, transcultural, and, as desires, they suggest the transcendent realm, which is their seeming referent, is desirable. And in terms of facilitating knowledge, it does seem to be the case that knowledge can be extracted from desires; at the very least, desires can communicate that an object of fulfilment for said desires exists. Thus, God, by giving innate desires for Himself, or, for the transcendent realm more broadly, could allow people to know of the existence of the transcendent realm; and, in light of the fact that persons typically pursue that which they desire, could also stack the deck in favour of the end which He desires: persons pursuing and obtaining a loving relationship with Himself. Therefore, it seems that ‘seemingly natural desires’, could fit the bill for the type of innate capacity God would be interested in giving to persons.

While I do think ‘seemingly transcendent desires’ make sense on a theistic picture, I am happy to cap the likelihood at 0.8 due to what I see as the major objection: ‘If God gave us these “seemingly transcendent desires” in order to help facilitate a loving relationship with Him, why is it that these desires are so often confused?’ Though a majority of the world is either a theist or an atheist, I cannot overlook the rather obvious fact that clearly there is disagreement between the two sides as to what the desires might be pointing to, not to mention the confusion within differing theistic traditions based upon their differing conceptions of God. So though it seems God has a good reason to give us these sorts of desires, it is also the case that these desires, in reality, seem to lead persons to very different conclusions about the object of their desires. Consequently, while ‘seemingly transcendent desires’ may lead a vast majority of persons into a loving relationship with God, one might wonder whether God might prefer a loving relationship that also entails right understanding of the person with whom one is in relationship. Now that is not to say that theistic traditions do not offer up various explanations as to why these ‘seemingly transcendent desires’ do not always result in the same understanding of God,¹⁷ but surely adding additional explanations in order to make sense of one phenomena should result in a decrease in the likelihood of that phenomena as well. So while there is clear historical precedent for thinking ‘seemingly transcendent desires’ to be likely on a theistic picture, and one can make a plausible case for why these desires would be the sort of good God would be interested in actualizing, I do think the diversity of religious experience, despite the available ways to account for it within specific theistic traditions, does limit the likelihood. Therefore, I am setting the $\mathrm{Pr}(E|H)$, where E is ‘seemingly transcendent desires’ and H is theism, at 0.8.

Naturalism and ‘Seemingly Transcendent Desires’

As I take it that my estimates for $\mathrm{Pr}(E|\sim H)$, where E is ‘seemingly transcendent desires’, and $\sim H$ is naturalism, will be the more controversial of my estimates, I will first lay out my reasoning, and then conclude my reasoning with the probabilistic assignments.

To begin, it is worth noting the difficulty of such a task is inevitably linked to the possibility of spandrels. By a spandrel, I mean a trait that is a by-product of, rather than a direct product of, adaptive selection. It is entirely possible that ‘seemingly transcendent desires’ are spandrels—they are not necessarily advantageous in and of themselves, but due to them not being particularly harmful, they are a trait that is retained. The difficulty spandrels present in assigning the likelihood of a naturalistic account is that no matter how many naturalistic accounts one can conjure, one can always postulate ‘seemingly transcendent desires’ being spandrels. I grant such a possibility. However, as I am interested in assessing the likelihood of ‘seemingly transcendent desires’ given naturalism, I do not think spandrels are necessarily that problematic. Absent of accounting for why these sorts of spandrels are likely, which, given that a spandrel by its very nature is a conjunction of unlikely events,¹⁸ seems an arduous task, seemingly the best we could say by making appeals to spandrels is they make the likelihood 0.5. While this prima facie appears to be an overly generous estimation, I am happy to grant it, because, as I will argue, 0.5 is about as good as it gets for the more robust naturalistic accounts. Therefore, the spandrel objection, viewed as generously as possible, does not seem to improve the likelihood more than other naturalistic accounts.

Instead of appealing to spandrels, it seems like the more plausible way forward for naturalistic accounts of ‘seemingly transcendent desires’ is an account that can account for both the origination of this type of desire, as well as an evolutionary benefit these desires would confer, which could then account for the widespread prevalence of these desires. With acknowledgement that there are a number of naturalistic accounts that could be posited, in this section, I will focus on the two that I take to be most plausible.

One option, in a Freudian-esque spirit, would be to argue that ‘seemingly transcendent desires’ are the result of biological desires for security and protection being projected out into the cosmos.¹⁹ If this were the case, one could theoretically account for the origination of, as well as widespread nature of, this type of desire. Though I think this to be one of the two best naturalistic accounts available, I do think it stands in need of substantial support. Firstly, one could follow Lewis himself and counter that the desire for security and protection being projected outward could just as easily account for the experience of those who lack ‘seemingly transcendent desires’.²⁰ While it does seem possible that one’s desire for security and protection could manifest in belief, and perhaps, desire, for a transcendent realm, it seems equally possible that such desires could manifest in disbelief and lack of desire for a transcendent realm, specifically for those who desire security and protection from a divine being who is capable of punishing persons who disobey Him. So, it seems such an approach could plausibly account for both the presence and non-presence of ‘seemingly transcendent desires’.

The problem, as I see it, is that it is unclear why, on this Freudian approach, ‘seemingly transcendent desires’ are so widespread. If it is the case that the projection of our desires for security can account for the presence, and absence, of ‘seemingly transcendent desires’, why is it the case that a majority of persons seemingly experience ‘seemingly transcendent desires’? In other words, why think that a desire for security and protection being projected outward into the cosmos is more likely than persons desiring security and protection and thereby not projecting outward into the cosmos? As I’ve argued, it seems to be the case that a significant majority of persons experience ‘seemingly transcendent desires’. The Freudian account might be able to account for the widespread prevalence of these desires, but it also seems to just as plausibly explain their absence. Absent of an explanation for why the desire for security and protection is more likely than not to be projected outward into the cosmos, thereby resulting in the widespread prevalence of ‘seemingly transcendent desires’, the Freudian account does not seem to give a good reason for thinking ‘seemingly transcendent desires’ will be the majority experience. Beyond, that is, simply saying that these desires are the by-product of what is an evolutionarily advantageous desire, namely, desires for security and protection. I am not disputing the rather obvious fact that a desire for security and protection would be evolutionarily advantageous. What I am disputing is the claim that the desire for security and protection would necessitate in projecting outward into the cosmos. Seemingly one could have this evolutionarily advantageous desire without that desire ever resulting in a projection outward. As such, I see no reason to think such a projection more likely than not, and therefore see no reason to assign the Freudian-esque account’s likelihood any higher than 0.5. That is to say, I do not think the likelihood of ‘seemingly transcendent desires’ occurring on this account is any higher than 50%.

A second plausible naturalistic account is that posited by Erik Wielenberg.²¹ As has been argued, a key characteristic of ‘seemingly transcendent desires’ is that they continually elude satisfaction. On Wielenberg’s account, this dis-satisfaction is itself evolutionarily beneficial, as it could lead an individual to outcompete others for natural resources, be innovative, and be adept at problem solving. Consider two persons: person A and person B. Person A experiences satisfaction of his desires which leads to him being sedentary. On the other hand, person B is continually unsatisfied, with the result being that he is continually active. Per Wielenberg’s logic, person B will have the evolutionary advantage because his dissatisfaction will lead to him being more proactive in figuring out creative solutions for how to survive. Of course, whether or not satisfaction leads to being sedentary, as in the case of person A, is another matter altogether.

That aside, one can see the plausibility of Wielenberg’s logic. If Wielenberg is correct in saying that dissatisfaction leads to an evolutionary advantage, then ‘seemingly transcendent desires’ which continually lead to this result would be a likely candidate for natural selection. And in fairness to Wielenberg’s intuition that dissatisfaction leads to an evolutionary advantage, there are specific instances in which dissatisfaction would very clearly lead to an evolutionary advantage. Take sexual desire for example. Let us assume that ‘seemingly transcendent desires’ are nothing more than a confused natural desire for sex. If this were the case, one could easily imagine a scenario where someone’s dissatisfaction resulted in them seeking satisfaction via having more sex. As increased sex increases the likelihood of reproduction, the evolutionary advantage of the continual unfulfilment of these desires is obvious.

From this then, it would seem Wielenberg’s account can make sense of the widespread, trans-cultural nature of ‘seemingly transcendent desires.’ On such a view, these desires and their corresponding lack of fulfilment would have been advantageous for early humans, and, thanks to natural selection, have successively passed on through subsequent generations. And though Wielenberg is not explicit about the origin of such desires, which has led some to question the persuasiveness of his account,²² as I’ve alluded to, it seems his view could easily accommodate for the origination of ‘seemingly transcendent desires’. Theoretically, ‘seemingly transcendent desires’ could be nothing more than unsatisfied natural desires like food, water, sleep, or even sex. As these desires would result in dissatisfaction, which is, in itself, evolutionarily advantageous, according to Wielenberg’s account, one could expect this experience of ‘seemingly transcendent desires’, which are continually unfulfilled, to be passed on. For that reason, even if subsequent persons experienced biological satisfaction of their natural desires, one might still expect this feeling of dissatisfaction from ‘seemingly transcendent desires’ to remain present.

On the whole, one can clearly see how ‘seemingly transcendent desires’ would confer an evolutionary benefit on Wielenberg’s account. Further, Wielenberg’s account can seem to make sense of the origination of, and the trans-cultural nature of, ‘seemingly transcendent desires’. With that said, I do think there are several serious challenges that can be mounted against Wielenberg’s naturalistic account.

Firstly, it seems far from obvious that the conditions Wielenberg describes as disadvantageous, namely, satisfaction, and an existence with overall less struggle, would, in fact, be so. On the contrary, it seems like such conditions may be evolutionarily advantageous. It seems quite plausible to imagine how a person who experienced satisfaction of their desires and was not in a constant struggle for resources would have an advantage when it comes to reproduction, as such conditions seem ideal for raising a family. And if this were the case, then one’s experience of the fulfilment of one’s natural desires would promote reproduction, which quite obviously increases the likelihood of such traits being naturally selected.

Secondly, and more existentially, it seems intuitive that persons are interested in surviving in so far as they believe their quality of life will be worthwhile in the future (Goetz, 2018: 175). Yes, it does seem plausible that dissatisfaction, or, the continued lack of fulfilment of one’s desires, could be calibrated by natural selection such that it would spur on competitive advantages, and, on the whole, be generally beneficial.²³ However, is it more plausible than the other possible outcome of dissatisfaction, namely, an increased likelihood of depression or suicidal ideation? To me, it seems quite intuitive to think that the experience of continual dissatisfaction could just as easily result in either of these latter outcomes. And the empirical data seems to suggest that this intuition is correct. As for depression, per Constance Hammen, ‘there is a robust and causal connection between stressful life events and major depressive episodes’ (2005: 293). While I do not think the unfulfilment of one’s desires is, in itself, necessarily stressful, I do think that the person who Wielenberg thinks has the evolutionary advantage, who, due to their unfulfilled desires, is constantly competing for resources, is very likely be in a state of chronic stress. And given the connection between chronic stress and depression, it is necessary to further note the connection between depression and suicide. Per Hong Jin Jeon, ‘suicide attempts [in South Korea] showed a significant association with mental disorders, especially major depressive episodes’ (2011: 370). Considering that the locale of Jeon’s research is South Korea, which has the highest rate of suicide amongst OECD member countries, his findings seem to be particularly relevant for current considerations. Combining these sources together, one can clearly trace how chronic stress, under the form of continually unfulfilled desires, increases one’s likelihood for depression, and thereby, also increases one’s likelihood for suicide.

Even if we were to grant that the unfulfilled person in Wielenberg’s example may not experience any of the things I’ve just mentioned—chronic stress, depression, and suicide—it is hard to see why these evolutionarily disadvantageous conditions are any less plausible than the competitive advantages Wielenberg postulates.

Taken together, these objections lead me to conclude that there is no reason to think that the individual in Wielenberg’s scenario is more likely than not to have an evolutionary advantage as a result of their chronically unfulfilled ‘seemingly transcendent desires’. Which means, in the absence of evidence as to why unfulfilled desires are more likely to be advantageous, instead of them being disadvantageous, on Wielenberg’s account, I think it best to leave the $\mathrm{Pr}(E|\sim H)$ at 0.5.

Results

In light of everything that has been said, I will now plug the likelihoods into Bayes’ theorem to assess whether Lewis’s ‘Argument from Desire’ is evidence for theism. With E representing ‘seemingly transcendent desires’, H representing theism, and ~ H representing naturalism, Bayes’ theorem will look as follows:

$$\mathrm{Pr}\left(\mathrm{H}|\mathrm{E}\right)=\frac{\mathrm{Pr}\left(H\right) \times \mathrm{Pr}\left(E|H\right)}{\mathrm{Pr}\left(E\right)}\to \mathrm{Pr}\left(\mathrm{H}|\mathrm{E}\right)=\frac{\mathrm{Pr}\left(H\right) \times \mathrm{Pr}(E|H)}{\left[\mathrm{Pr}(E|H) \times \mathrm{Pr}\left(H\right)\right]+[\mathrm{Pr}(E|\sim H) \times \mathrm{Pr}(\sim H)]}$$

Given the simplifying assumption that theism and naturalism are mutually exclusive and exhaustive, and in order to assess whether the evidence in question, ‘seemingly transcendent desires’, raises the probability for theism, thereby counting as positive evidence for it, $\mathrm{Pr}\left(H\right)$, theism, is set at 0.5, and subsequently, $\mathrm{Pr}\left(\sim H\right),$ naturalism, is also set at 0.5. $\mathrm{Pr}(E|H)$ is how likely ‘seemingly transcendent desires’ are on theism, which I have argued should be set at 0.8. Consequently, $\mathrm{Pr}(E|\sim H)$, which is how likely ‘seemingly transcendent desires’ are given naturalism, I have argued should be set at 0.5.

Now plugging in our priors, the equation and solution looks as follows:

$$\mathrm{Pr}\left(\mathrm{H}|\mathrm{E}\right)=\frac{(0.5) \times (0.8)}{\left[\left(0.8\right)\mathrm{x}(0.5)\right]+[\left(0.5\right)\mathrm{x} \left(0.5\right)]}=\frac{0.4}{\left(0.4\right)+(0.25)}\approx 0.615$$

Therefore, the probability of theism, given the phenomena of ‘seemingly transcendent desires’ increases from 50 to ≈ 61.5%.

To conclude, in the absence of other evidences, C.S. Lewis’s ‘Argument from Desire’, which postulates the likelihood of a transcendent realm based upon the experience of ‘seemingly transcendent desires’, should count as positive evidence for theism. The benefit, then, of using Bayes’ theorem to assess Lewis’s argument is one can wade through the widespread disagreement surrounding Lewis’s argument and use probabilistic reasoning to conclude that Lewis seems to have been on to something when he reasoned from his experience of ‘seemingly transcendent desires’ to the existence of another world. Though Lewis and I have approached the ‘Argument from Desire’ in different ways, we can both agree that ‘seemingly transcendent desires’ give some reason to think that a transcendent object of fulfilment exists. And given the conservative nature of my probabilistic assignments, I concede that the outcome of my use of Bayes’ theorem may be a more modest approximation than others, including Lewis, might think reasonable.