Does absence make atheistic belief grow stronger?

Abstract

Discussion of the role which religious experience can play in warranting theistic belief has received a great deal of attention within contemporary philosophy of religion. By contrast, the relationship between experience and atheistic belief has received relatively little focus. Our aim in this paper is to begin to remedy that neglect. In particular, we focus on the hitherto under-discussed question of whether experiences of God’s absence can provide positive epistemic status for a belief in God’s nonexistence. We argue that there is good reason to accept an epistemic parity between experiences of God’s presence and experiences of God’s absence

The question of what role, if any, religious experience can play in grounding theistic belief has received a great deal of attention over the last 50 years, with a number of prominent philosophers, such as Swinburne (1979), Alston (1991), Plantinga (2000), and Stephen Evans (2010), arguing that religious experience can provide significant warrant for belief in God. By contrast, the relationship between what we might term ‘areligious experience’ and atheistic belief has received relatively little focus. In this paper, we aim to remedy this neglect by considering whether a particular kind of areligious experience—experiences of God’s absence—can provide warrant for atheistic belief. We will refer to any attempt to ground warranted atheistic belief in experiences of God’s absence as an ‘appeal to absence’. In “Martin’s challenge”, we outline the hitherto most influential discussion of such appeals. In “Evaluating Martin’s challenge”, we evaluate it and find it wanting. Based on these failings, in “A new appeal to absence”, we construct an appeal to absence which parallels two recent attempts to ground warranted atheistic belief in religious experiences. In “Evaluating the new approach”, we argue that this account legitimises the claim that experiences of God’s absence can provide a significant degree of warrant for atheistic belief.

Martin’s challenge

The most influential extant discussion of appeals to absence is found in the work of Michael Martin (1986, 1990). Martin poses the following question (1986, pp. 82–83):

Since experiences of God are good grounds for the existence of God, are not experiences of the absence of God good grounds for the non-existence of God? After all many people have tried to experience God and have failed.

Before attempting to answer Martin’s question, though, it’s worth pausing to address some potential sources of confusion.

Martin’s conditional

Importantly, Martin himself does not actually endorse the appeal to absence he makes here; his question is purely rhetorical. In Martin’s (1990, p. 174) view, neither religious, nor areligious experience is capable of grounding the corresponding belief, since, in order for these experiences to do so, we would first need to know the conditions under which an experience of the presence or absence of God would most likely be veridical. (In speaking in this way, it’s clear that Martin is not using ‘experience’ as a success term. We will follow his usage throughout this paper.) Yet, Martin maintains that if

God exists, we have no idea under what circumstances He would be experienced and under what circumstances he would not be. In a similar way, if God does not exist, we have no plausible theory of when people would have an experience of God and when they would not. (ibid.)

Martin’s argument, then, is not presented as a reason to think that areligious experience has epistemic value; rather, it’s presented as a means of countering attempts made by Swinburne (1979) and others to show that religious experience provides good grounds for endorsing theism over atheism. As such, his central contention is best characterised in the form of a conditional: if religious experiences can provide good grounds for theistic belief, then experiences of God’s absence can provide good grounds for atheistic belief.

The significance of Martin’s conditional

Confusingly, Martin presents this conditional as a means to counter arguments ‘based on experience of the presence of God’ (1986, pp. 82–83). However, just why we should take it as such is not clear. Of course, if we take ‘good grounds’ here to be factive then it would be impossible to have good grounds for both theistic and atheistic belief. However, many proponents of theistic appeals to religious experience don’t endorse anything this strong. Gutting (1982, pp. 149–150), for example, claims that ‘we should think of an individual of-God experience as providing significant but not sufficient evidence for God’s existence, needing to be included in a cumulative body of diverse evidence that can warrant the claim that God exists’. And Swinburne (1979, pp. 260–271) himself maintains that the grounds religious experiences provide for theistic belief are both defeasible and dependent on a number of background conditions. So, in the absence of a reason to think otherwise, we deny that there is any obvious tension in accepting both that religious experience provides good grounds for some individuals to believe in God and that areligious experience provides good grounds for some individuals to disbelieve in God.¹ Indeed, it’s not even clear that—given certain weak readings of ‘good grounds’—a single individual couldn’t have good grounds for endorsing both theistic and atheistic belief. After all, we often find ourselves in the position of having good grounds for believing a particular proposition, while also having good grounds for believing its negation.

Given this, our sole focus will be on the question of whether we should accept Martin’s conditional claim. We will not consider what implications this has for the overall epistemic status of either theistic or atheistic belief. In order to do so, we will take the antecedent of the conditional for granted and assume that religious experience can (given the right conditions) provide some kind of positive epistemic standing for theistic belief. We will talk below as if a belief has ‘positive epistemic standing’ iff it possesses some not insignificant degree of warrant, and we will follow Plantinga (2000, p. 325) in using ‘warrant’ to mean ‘the property or quantity, enough of which is what distinguishes knowledge from mere true belief’. However, we do not mean to commit ourselves to a precise view of exactly what kind of positive epistemic status (a)religious experience can provide. For our purposes, the crucial question is not what kind of epistemic standing (a)religious experience might provide, but whether accepting the antecedent of Martin’s conditional commits us to accepting the consequent. We will argue, in “A new appeal to absence” below, that it does, and that, in a number of important respects, appeals to religious experience are on all fours with appeals to absence. However, contra Martin, we do not take this as providing any challenge to the epistemic value of religious experience.

Absence of experience and experience of absence

Another possible source of confusion is that Martin’s challenge conflates two very different potential sources of warrant for atheistic belief. For, while Martin begins by focusing on experiences of the absence of God, he very quickly moves to discussing the absence of experiences of God. In doing so, Martin runs together two different lines of argument: the first concluding that experiences of a certain kind (experiences of God’s absence) can provide support for atheistic belief; the second that the failure of certain individuals to have experiences of God can provide such support. Whether arguments of the latter kind provide any support for atheism is a fascinating issue in its own right, but one which has already been discussed at length elsewhere as part of the so called ‘problem of divine hiddenness’.² As such, we won’t pursue it here. Our attention will be focussed exclusively on experiences of absence, rather than absences of experience.

Now, it’s possible that these issues are not as distinct as we claim; perhaps Martin was intending to suggest that whenever one attempts to experience God and fails, one thereby has an experience of God’s absence. If so, this is unconvincing. The general claim that a failure to experience something yields an experience of the absence of that thing is plainly false (we may, e.g., be unable to see the book we are searching for, while at the same time being unsure that the book is, or even appears to be, absent), and in the absence of any cogent argument to the contrary, we see no special reason to think that the theistic case would be any different.

Some further assumptions

Any appeal to experiences of absences clearly requires that we are literally able to experience absences of certain kinds. Further, if such experiences are to be capable of conferring warrant of the relevant kind, then this will likely require (as we discuss further below) that causation by absence is possible.³ We think that there are excellent reasons to accept both of these theses—as outlined in, for example, Sorensen (2008) and Farennikova (2013)—but we will not rehearse these arguments here. Instead, we will merely take it as an assumption that there can be both causation by absence, as well as genuine experiences of absences. That said, it does not follow from these general claims that we are able to experience the absence of God, less still that we could experience it in the manner required to legitimise appeals to absence. Clearly, though, any successful appeal to absence requires these further claims. Hence, we offer a defence of them in “Causal connections?”. For now though, we will merely take them as assumptions.

Evaluating Martin’s challenge

With these clarifications in place, we can now ask whether Martin’s challenge is successful. That is, we can ask whether Martin shows that—given that religious experience can yield warranted theistic belief—experiences of God’s absence are able to provide warrant for atheistic belief. Evaluating the success or otherwise of Martin’s challenge is no easy matter, not least because he never presents an explicit argument in favour of his conditional claim. However, we will argue below that the considerations Martin does present fail to support it.

Experience of absence

One obvious worry with Martin’s challenge is that the move from ‘I am experiencing the absence of x’ to ‘x does not exist’ is clearly a non-starter. Someone who denied that cows exist simply because their present experience was of cows being absent would manifestly be reasoning badly. As Kwan (2006, p. 650) phrases the point:

Let C be the claim that there is a cow and –C its negation. For C to be true, we just need one alleged perception of a cow to be veridical. However, for –C to be verified we need to examine all spatiotemporal locations and perceive no cow and all these infinite number of perceptions have to be veridical.

There’s a crucial asymmetry between the epistemic weight borne by experiences of something’s presence and that borne by experiences of its absence. Still though, it’s not clear that the failure of the general move from experiences of x’s absence to the non-existence of x is genuinely problematic for Martin. God is not the kind of entity which—like cows, cuckoo clocks, and the rest of the furniture of our everyday lives—can be absent from a particular location and yet present elsewhere. Divine omnipresence is, after all, a key part of theistic orthodoxy. Thus, if God exists, there is no place from which he is genuinely absent. Provided, then, that an experience of God’s absence from some location can genuinely warrant belief in God’s absence from that location, this appears to eo ipso provide warrant for the further conclusion that God does not exist.

It is not obvious, however, that the kinds of experience Martin is concerned with are the right ones to play this role. For, while there is certainly a sense in which divine omnipresence requires God’s not being absent anywhere, it’s far from clear that this is the sense in which those Martin discusses experience God as absent. Take, for example, Mawson’s (2005, p. 26) view according to which God is omnipresent in virtue of ‘knowing directly what is happening everywhere and being able to act directly everywhere’. To deny that a being is omnipresent in this sense would be to deny that there are (at least) some parts of the universe which that being is able to directly monitor or influence. Yet, there being parts of the universe which God cannot directly monitor or influence is an unlikely candidate for something that could be experienced in any straightforward manner.

We are not here endorsing the particular account of omnipresence Mawson provides. Rather, we are suggesting that if something like it is correct, then it’s difficult to see how our experiences could straightforwardly represent God as being absent in the relevant sense. And this is not just an idiosyncratic consequence of Mawson’s view. Many prominent accounts of omnipresence—e.g., those of Aquinas (1274/1978, p. 65), Hartshorne (1941) and Swinburne (1977, pp. 99–105)—also expound the notion with reference to God’s knowledge and/or his power to act.

Some may object that an obvious tu quoque lurks here. That God is able to both directly know and directly influence what happens in a particular location is also something we couldn’t have any straightforward experiential access to. This may be so, but it’s important to remember that the kinds of experience typically invoked as experiences of God’s presence are not ‘bare experiences’ of his being present in this sense. More typically, they are experiences of a richer nature, such as those described by Plantinga (1981, p. 46):

Upon reading the Bible, one may be impressed with a deep sense that God is speaking to him. Upon having done what I know is cheap, or wrong, or wicked I may feel guilty in God’s sight and form the belief God disapproves of what I’ve done. Upon confession and repentance, I may feel forgiven, forming the belief God forgives me for what I’ve done.

Similarly, when those in various religious traditions speak of experiencing God’s absence, they are typically speaking of an experience of God’s failing to act in these religiously significant ways, rather than an experience of God’s inability to act directly within a particular spatio-temporal location. Consider, for example, experiences of God’s absence as described by religious mystics such as Saint John of the Cross and Thérèse of Lisieux, or by the psalmist when he opines ‘I say to God my Rock, “Why have you forgotten me? Why must I go about mourning, oppressed by the enemy?”’ (Psalm 42v9 NIV). It is much less clear that these richer experiences, both of God’s presence and of his absence, couldn’t be straightforwardly experienced. We might accept, then, that experiences of God’s absence in this richer sense provide good evidence for God’s being absent in this sense, in much the same way that richer experiences of his presence provide good evidence for his being present. However, the two cases are importantly disanalogous. For, while God’s being absent in this sense on some particular occasion is neutral with respect to God’s existence, God’s being present is not. God may exist and yet sometimes fail to act in these religiously significant ways; but if God does act in these ways—indeed, if he acts at all—he must exist.

There is, then, a crucial difference between particular experiences which represent God as acting in a religiously significant way and those which represent him as failing to act in such a way. It is not obvious, though, that these are the only possible kinds of (a)religious experience. Under one plausible conception of ‘experience’, someone might have an experience of something of a much more general nature; such as an experience of God’s being present in, or absent from, the world at large. God’s failing to act in a religiously significant way on some particular occasion is not, so far as we can see, incompatible with his existence. (Indeed, the religious mystics we mentioned above sometimes take it, rightly or wrongly, as positive evidence for his existence.) By contrast though, it is a key tenet of classical theism that God sometimes acts in these religiously significant ways. Hence, if there could be experiences of God’s never having acted in such ways—i.e., experiences which represent our world as being devoid of divine influence—then there could be areligious experiences the content of which is not neutral with respect to God’s existence. It is, of course, controversial whether such generalised areligious experiences are so much as possible, but it’s worth noting that some defenders of appeals to religious experience have already postulated experiences of God’s presence of this very general kind. Evans (2010, p. 60), for example, suggests that there are experiences of the universe as a whole as being ‘mysterious or puzzling, crying out for some explanation’, and that these may give rise to—or even be part of the content of—experiences of God’s presence in the world at large. In “Evaluating the new approach” we defend the possibility (and epistemic value) of areligious experiences of this more general kind.

Background knowledge

A further worry concerns Martin’s contention (1990, p. 174) that we lack what he calls ‘background knowledge’—that is, an understanding of the circumstances under which we would likely have experiences of God’s presence or of his absence. Martin provides no clear argument for this claim; but argument is surely needed since this claim is highly controversial. Various religious traditions make claims concerning the kinds of activities which an individual might undertake to abet such experiences. Such as, e.g., ‘[following] the way by which [those who believe] began: by behaving just as if they believed, taking holy water, having masses said, etc.’ (Pascal 1995, p. 156). Similarly, proponents of the argument from divine hiddenness often maintain that if a perfectly loving God exists then it ‘shouldn’t be the case that, in general, people never (or only very rarely) have experiences […] of the love or presence of God’ (Rea 2011, p. 268).⁴ Even Martin’s own (1986, p. 83) appeal to the case of people who try to experience God and fail suggests a tacit endorsement of the thought that, were God to exist, he would most likely be experienced by those seeking to experience his presence, rather than those actively seeking to avoid it.

Martin therefore owes us some justification for the claim that we lack the relevant background knowledge. (Or, at the very least, for the claim that we possess such knowledge with respect to God’s presence only if we also have it concerning his absence.) Otherwise, it’s possible to maintain, contra Martin, that since we have an excellent working theory concerning the situations under which we would experience God’s presence, but no such theory with respect to God’s absence, religious experience carries significantly more epistemic weight than areligious experience.

Could Martin show this? It is difficult to say. Doubtless, there are various arguments that could be constructed in favour of the claim that we don’t possess (much) knowledge concerning the circumstances under which God would be experienced, but assessing these would require a paper in itself. Fortunately, we needn’t take a stance on these arguments. For, as we now show, the success or otherwise of appeals to absence doesn’t hang on the availability of background knowledge.

In support of his claim that such background knowledge is needed if religious or areligious experiences are to be capable of serving as sources of warrant, Martin asks us to consider the following analogy. ‘[I]f 50 people experienced the absence of dodos […] on the island of Mauritius, this would indeed be evidence of their absence [but] if 50 people reported seeing dodos in the middle of winter in Alaska, this would not be taken seriously, given our background knowledge.’ (1990, p. 174) However, this example fails to meet its intended function. The Mauritius case merely shows that a certain experience, taken together with relevant background knowledge, is sufficient for conferring warrant; it does not show that this combination is necessary. In the Alaskan case, on the other hand, the perceivers, and those who later receive their testimony, are not in the position of lacking background knowledge entirely. Rather, they are in the position of possessing a great deal of background knowledge (concerning the history of the dodo, their preferred climate and so forth), which makes the prior probability of the relevant dodo claims exceedingly low. Thus, while we may well agree with Martin’s judgements about these cases, and with his more general claim that background knowledge can play some role in determining whether our perceptual (and testimonial) beliefs are warranted, this does not lend any support to his additional claim that ‘[g]iven our lack of background knowledge […] no inferences can be made concerning [God’s] existence or lack of existence from our experiences of the presence or absence of God’ (ibid.). To support this further claim, Martin needs to show that individuals lack perceptual warrant in the absence of relevant background knowledge; not just in the presence of conflicting background knowledge. We will argue, however, that such a position is untenable.

Consider the first party of sixteenth century sailors—with no prior knowledge of dodos—arriving on the island of Mauritius. These sailors have, ex hypothesi, no background knowledge concerning dodos or the conditions for our perceptual experience of them, but it would seem wildly inappropriate to disregard their perceptual experiences (or their testimony concerning them) on this basis.⁵ Indeed, there is good reason to think that the requirement that putative observers be in possession of relevant background knowledge for their experiences to count as evidence precludes the very possibility of ever establishing such a background theory in the first place. As Hume (1748/1975, p. 153) has famously argued:

It is a question of fact, whether the perceptions of the senses be produced by external objects, resembling them: how shall this question be determined? By experience surely […] [b]ut here experience is, and must be entirely silent. The mind has never anything present to it but the perceptions, and cannot possibly reach any experience of their connexion with objects.

In a similar vein, Evans (2010, pp. 153–154) notes that, ‘[t]he view that we need empirical evidence of a regular connection between the sensation and the [extra-mental] object in order to rely upon the sensation as evidence for the physical object would make it impossible to have any knowledge of the [extra-mental] world at all’. Thus, we are seemingly faced with a choice: accept a thorough-going scepticism, or allow that we can have perceptual warrant even in the absence of relevant background knowledge. Although the former option is not without its proponents—including, on some interpretations at least, Hume himself—the latter has certainly proven most popular amongst contemporary epistemologists.⁶

We conclude, then, that there are very good reasons (at least for non-sceptics) to admit that we can have (quasi-)perceptual warrant in the absence of relevant background knowledge. Importantly, one could consistently endorse this claim while also allowing that no warranted inferences of the relevant kind can be made without background knowledge. It may well be that any argument appealing to (a)religious experience would have to appeal to some claims regarding background knowledge (we take no stance on this issue). Nonetheless, this doesn’t show that all appeals to experience of God’s presence or of his absence require such knowledge. Some of the most promising appeals to religious experience—including those of Plantinga (2000) and Evans (2010) which we discuss at length in “Casual connections” have been non-inferential in character.

We can see, then, that if there are no equally promising non-inferential appeals to absence, Martin’s case for a parity between appeals to experiences of God’s presence and experiences of God’s absence will fail. If, however, there are equally promising non-inferential appeals to absence available (and we argue in “Evaluating the new approach” that there are), then it may be that something like Martin’s parity claim is correct, but not for the reasons he cites.

What do these objections teach us?

Our main purpose in drawing out the shortcomings of Martin’s argument has been to highlight some of the features any successful appeal to absence must include. Firstly, the discussion in “Experience of absence” suggests that any experience of God’s absence capable of conferring positive epistemic status on atheistic belief should not merely represent God as absent from a particular location, but represent him as absent simpliciter; absent from the world at large. Further, the states of affairs these experiences represent must not be neutral with respect to (or worse, tell in favour of) God’s existence, and they must be of a kind we could plausibly have (quasi-)perceptual access to. Finally, as we discussed in “Background knowledge”, there seems to be good prima facie reason to think that—as with appeals to religious experience—the most successful appeals to absence will be non-inferential in nature. In the next section, we construct an appeal to absence which incorporates these desiderata.

A new appeal to absence

As we have seen, the most prominent extant attempt to employ atheistic experiences as part of an argument aimed at warranting atheistic belief—or, rather, aimed at showing that such beliefs would be warranted by areligious experiences, were theistic beliefs warranted by religious experience—is unsuccessful. This is not to say that no such argument could be constructed, however, and we don’t take ourselves to have demonstrated anything of the kind (though we are extremely sceptical about the prospects for doing so).

Non-inferential accounts

If we did accept this further claim, though, where would that leave us with respect to appeals to absence? One obvious response is that we now have reason to reject all such appeals and so—given our previous assumption concerning the epistemic value of religious experience—reason to reject Martin’s parity claim. This is not, however, the only response available. Instead, one might conclude that the problem is not with appeals to absence per se, but merely with those which try to warrant atheistic belief on the basis of inferences. A better tactic, we suggest, is to offer a non-inferential appeal to absence which parallels the various recent attempts to show that religious experience can give rise to warranted theistic belief in a non-inferential manner. To get a sense of how such an appeal would proceed, it will be useful to examine some of these attempts.

We have been assuming thus far that some kinds of religious experience can, in the right circumstances, provide warrant for theistic belief. We have, however, said very little about the kinds of religious experiences we are considering, or what precisely (or even approximately) constitutes the right circumstances. In this section, we look in detail at two related accounts of how religious experiences can provide non-inferentially warranted theistic belief; those of Plantinga (2000) and Evans (2010). We then construct a parallel account which appeals to areligious experience as a source of non-inferential warrant for atheistic belief. Although we focus on Evans’ and Plantinga’s accounts, we do not take the success of our own account to be dependent on the letter of either. Rather, we hope to show that anyone who accepts certain key aspects of these views should (modulo one important caveat to be introduced below) accept a similar story with respect to the epistemic value of areligious experience.

Evans’ theistic natural signs

Evans’ account is built around the concept of a ‘theistic natural sign’ (TNS)—which is modelled on Thomas Reid’s concept of a natural sign (2010, p. 26).⁷ TNSs, on Evans’ view, are experiences (or other features of the world) which have the following key features. First, they are caused by God (ibid., p. 35). Second, they bring God directly to our awareness and lead us to form beliefs about God in a non-inferential manner (ibid.). Third, God not only created these signs, he created them to be signs. Fourth, they are widely available to us. Evans maintains that TNSs of this kind provide defeasible non-inferential warrant for theistic beliefs (ibid., p. 26). Evans also claims that we possess a natural disposition to read TNSs correctly and to form theistic beliefs when we encounter them (ibid., p. 37). This explains why belief in God is fairly widespread. However, belief in God is not universal since our disposition to recognise these signs can be weakened in certain circumstances and the signs themselves are easily resistible (ibid., p. 38).

The examples of TNSs Evans considers include experiences of the world as exhibiting beneficial order (ibid., pp. 74–106); the experience of human beings as having intrinsic worth (ibid., pp. 142–147); and the sense of being morally accountable (ibid., pp. 139–142). They also include experiences of what he calls ‘cosmic wonder’. Such experiences, as mentioned in “Experience of absence”, represent ‘the world, or objects in the world, […] as mysterious or puzzling, crying out for some explanation’ (ibid., p. 60).

Plantinga’s A/C model

Alvin Plantinga’s now famous ‘A/C Model’ (so called because of its basis in claims made by Aquinas and Calvin) bears a number of important similarities to Evans’ account. Firstly, the A/C model postulates ‘a natural human tendency, a disposition […] to form beliefs about God under a variety of conditions and in a variety of situations’ (2000, p. 171). (Following Calvin, Plantinga calls this disposition a ‘sensus divinitatis’.) Secondly, Plantinga claims that certain signs—an awareness of being forgiven, or a beautiful spring morning, say—may trigger this disposition and thus occasion belief in God.⁸ Further, the deliverances of this faculty will (provided God exists) frequently be warranted. Thirdly, the relevant beliefs are formed in a non-inferential manner—they will be—in Plantinga’s terms—basic beliefs, and in cases where such beliefs are warranted, properly basic beliefs.

It isn’t that one beholds the night sky, notes that it is grand, and concludes that there must be such a person as God […] It is rather that upon the perception of the night sky […] these beliefs just arise within us. They are occasioned by the circumstances; they are not conclusions from them. (ibid., p. 175)

Fourthly, the sensus divinitatis is modifiable; it ‘has been damaged and corrupted by sin’, but, through ‘faith and the concomitant work of the Holy Spirit’, can be healed and made to function properly (ibid., p. 186).

As a final point of comparison with Evans’ account, the A/C model also involves some appeal to God’s intentions. On Plantinga’s (1993) view of warrant, any belief attaining warrant must result from some cognitive faculty successfully aimed at generating true beliefs. So, the sensus divinitatis, if capable of furnishing us with warranted theistic beliefs, must have been designed to do so. Hence, Plantinga famously contends that the de jure question of whether belief in God is warranted is not independent of the de facto question of whether God exists. If humankind was indeed created ‘with a natural tendency to see God’s hand in the world about us’ (2000, pp. 190–191), then, Plantinga maintains, theistic beliefs formed in the non-inferential manner he describes will—provided the other conditions he enumerates are also met—be warranted. If, however, this natural tendency is not the result of God’s intentional action—which, of course, it could not be if God does not exist—then the beliefs in question will lack warrant.

Evans and Plantinga do, however, differ in some important respects. Most relevantly for our purposes, while Evans wants there to be ‘something distinctive’ about those signs that can trigger our God-detecting faculty, Plantinga, following Calvin, seems quite happy for it to be triggered by a much more inclusive range of circumstances:

[I]t isn’t only the grandeur and majesty that counts; he [Calvin] would say the same for the subtle play of sunlight on a field in spring […] or aspen leaves shimmering and dancing in the breeze “There is no spot in the universe,” [Calvin] says, “wherein you cannot discern at least some sparks of his glory.” […] According to the model, therefore, there are many circumstances, and circumstances of many kinds, that call forth or occasion theistic belief. (ibid., p. 174).

Having outlined some key features of the Evans/Plantinga account, we will now go on to examine whether a parallel account could be offered in defence of atheistic belief.

Atheistic natural signs

Suppose we took an Evans/Plantinga style account to have a good degree of plausibility, and accepted that the signs they discuss may afford us with non-inferentially warranted theistic belief. Should we find it equally plausible that there could be ‘atheistic natural signs’ (ANSs)—signs that point to God’s non-existence and yield non-inferentially warranted atheistic belief? Before answering this question, it’s first sensible to ask what type of experience might function as an ANS. As mentioned above, experiences which represent something as absent from a particular location typically provide little reason to believe in the non-existence of that thing. And, while experiences of God’s absence are—due to his omnipresence—an exception to this general trend, there is reason to doubt that the kinds of experiences that represent God as absent from a particular location offer any support for atheism. The experiences we are looking for here will, therefore, need to be of a much more general kind since they must give rise to a warranted non-inferential belief in God’s absence from the world at large.

In order to avoid any worries concerning the sense of ‘absence’ in play here, we will focus on putative ANSs which give rise (non-inferentially) to a warranted belief that God is absent in virtue of his non-existence or, in other words, that atheism is true. What remains to be seen is whether there can be signs of this kind. We will consider three kinds of experience as possible examples of ANSs. First, in contrast to Evans’ experiences of the world as exhibiting beneficial order, there may be experiences of the world as being inhospitable to order. Second, there may be an experience of the world as being indifferent to our hopes and concerns. Finally, there may be a feeling of ‘absurdity’ as described by certain existentialist philosophers—a feeling, that is, that life, and existence more generally, has no meaning beyond that which we give it ourselves.⁹ What we need to show, then, is that such experiences are possible and that they can, given the right circumstances, yield warranted atheistic belief.

The first task does not strike us as especially difficult. Recall that ‘experience’ here is being used non-veridically; as such, we see no principled reason why someone who allows that we can, e.g., have experiences which represent the world as being governed by providential order should not grant the same with respect to experiences of the world’s lacking such order (especially since many people report having experiences of this kind). The difficulty lies, however, in establishing that such experiences can provide the kind of warrant necessary for them to serve as ANSs. Consider an individual who consistently experiences the universe as being inhospitable to order of any kind. Time and time again this individual sees how hard it is to impose order on the world, and how easily destruction and disorder arise. Such a person may be struck by the feeling that disorder is the natural order of things, and such an experience may trigger in them the belief that God does not exist. The thought here is not that the person experiences the world as inhospitable to order and then deduces from this that there is no God. (As Plantinga (2000, p. 175) points out with respect to theistic arguments which move from, e.g., the experience of the night sky as grand to the conclusion that God exists, such inferences would be manifestly fallacious.) Rather, the thought is that atheistic belief arises from the experience of the world as being inhospitable to order in a non-inferential manner; likewise for the other suggested ANSs. So what conditions must such signs meet to successfully warrant atheistic belief?

ANSs, much like TNSs, must be in some way caused by what they signify; in this case, by God’s non-existence. Now, it’s certainly safe to say that if the God of classical theism existed, the universe wouldn’t be inhospitable to order or indifferent to our hopes and concerns;¹⁰ nor would the world be lacking in a deep and inbuilt significance. Thus, there is very plausibly a counterfactual link between atheism’s being true and the relevant experiences being veridical.¹¹ Such a counterfactual relationship gives us some prima facie reason to allow that there is also a causal link (or, at least, that such a causal link would exist were God genuinely absent); whether this reason is merely prima facie will be the focus of “Casual connections?”.

ANSs must also characteristically give rise to the belief that God does not exist. Further, this feature must be characteristic in virtue of the signs triggering some disposition within us whose function it is to produce such beliefs. In “Distinctive signs?” we argue that our ANSs can meet these criteria. Now, it cannot, of course, be a necessary condition for something’s counting as an ANS—as Evans claims it is for something’s counting as a TNS—that it is intended to function so as to produce a certain kind of belief. However, we argue that this is not a plausible condition for a successful ANS (or TNS).

Evaluating the new approach

As already mentioned, it seems clear to us that we can have experiences of the kind described in the last section and that in an atheistic universe they would likely be veridical. This last point is important and it may well be that—as Plantinga contends—the epistemic status of theistic and atheistic belief ultimately depends on which of these beliefs is, as a matter of fact, true. We find this picture extremely plausible provided one accepts (as we have done for the sake of argument) the kind of externalist epistemic picture prominent in both Plantinga’s and Evans’ accounts. If this is the case, though, then our conditional claim—that if the Evans/Plantinga approach can warrant theistic belief, a parallel areligious experience account can be used to warrant atheistic belief—appears to be mistaken. To avoid such worries, we will slightly modify our conditional and introduce the caveat mentioned earlier. We will argue instead that if TNSs of the Evans/Plantinga kind can yield warranted theistic belief given the truth of theism, then parallel ANSs can yield warranted atheistic belief given the truth of atheism. It may well be, then, that contrary to the suggestion made in “The significance of Martin’s conditional”, a single individual could not have warrant for both theistic and atheistic belief. However, there will still be an important sense in which, if our account is successful, the two kinds of belief will be epistemically on par. So, how plausible is our modified parity claim?

Causal connections?

Evans himself dismisses the possibility of ANSs for a number of reasons, the first being that in an atheistic universe, there would be no relevant entity to cause ANSs:

To be a natural sign, an experience or feature of the world would have to be caused by the reality it signifies, and there must be a natural disposition to become aware of that reality and form a belief about it. If God does not exist, there could hardly be a causal relation between a non-existent being and the [putative ANS.] (2010, p. 93)

This is, however, misguided. If God does not exist, the signs we have discussed must in some respect be caused by a world from which God is absent or some part of that world (since that is all there is). And that the world is one from which God is absent is precisely what these signs could signify to us—in a manner we will explicate more fully below—when we encounter them. Thus, it is not true that ANSs could not be caused by the reality they signify if God does not exist. It is not required that ‘a non-existent being’ acts as a relata here; rather, it would be some genuine feature of the universe, its being one from which God is absent, which does the relevant causal work. Consider, for comparison, a standard case of apparent causation by absence; the absence of oxygen causing a match to fail to light. Defenders of causation by absence do not claim that some non-existent object (the oxygen in the room) causes the match to fail, but rather that some existing state of affairs does (viz., one that includes the oxygen’s being absent).

This, of course, depends upon our earlier assumption that causation by absence is possible. However, someone may object that, even granting the general possibility of causation by absence, there is still reason to deny that God’s absence could be causally efficacious. It’s hard to see, though, what principled reason there could be for believing this. It would not do, for example, for someone to advance the following reasoning: “if God is absent (by virtue of non-existence), then his absence is not a contingent matter. Hence, if a counterfactual analysis of causation is correct, the relevant counterfactuals involving God’s absence will—according to the standard Lewis/Stalnaker¹² semantics for counterfactuals—be merely trivially true (since their antecedents are necessarily false). Surely, though, such experiences being caused by God’s non-existence should (if this account is to work) be a substantive matter.” For, an obvious tu quoque awaits: God’s existence is, of course, also a non-contingent matter. Hence any counterfactual understanding of causal claims that involve God’s existence will likewise be trivially true at best.

Distinctive signs?

Evans’ second reason for dismissing the possibility of ANSs is that it does not ‘make sense to say that the intended function of [a putative ANS] is to produce unbelief.’ (2010, p. 93). His point seems to be that while, if God exists, our God-given, God-detecting dispositions will likely have been designed to occasion such beliefs in the presence of TNSs, the same will not apply with respect to ANSs. If God does not exist, then any disposition that might occasion atheistic belief in the presence of ANSs could not have been so designed. The last part of this claim is almost certainly correct since—modulo various far-fetched scenarios involving ancient aliens and the like—our ANSs cannot have been literally designed so as to function as signs.

Need we accept, though, that beliefs that arise in a non-inferential manner cannot be warranted unless the signs which trigger them are intended to act as signs? We think not. Evans’ general account of signs draws heavily from Thomas Reid’s account of direct perception, an account which makes no reference to ‘intended function’. According to Reid, we perceive by way of sensations, and such sensations give us direct (non-inferential) awareness of the objects of the external world (1785/2002: 209-11). All that matters, on Reid’s account, for a non-inferential perceptual belief’s counting as warranted is that the belief be caused by the sensation; that the sensation be caused by the object it signifies; and that we be ‘hard-wired’ in such a way as to move directly from the sign to the relevant perceptual judgement (Evans (2010, pp. 31–33). In short, a non-inferential perceptual belief is warranted if it is caused in the right sort of way by the right sort of mechanism. As already mentioned, accounts of this general sort—according to which that what matters to a belief’s counting as properly basic is just that it be formed in an appropriate way (and that it lacks undefeated defeaters)—have proven extremely popular with contemporary epistemologists. And it seems that Evans himself (2010, p. 170) is at least broadly sympathetic to this general account of perceptual belief. If, though, we accept (i) that one can represent ordinary objects of perception without one’s perceptual representation being in any sense intended to represent such objects, and (ii) that one can come to have (non-inferential) knowledge of the object’s existence and nature on the basis of such perceptions, then it looks like we cannot accept the general claim that to arrive at knowledge of what is represented the representation must be intended as a representation. Given this, we need some principled reason for accepting that signs in general, or ANSs in particular, must be intended to function as signs in order to either signify their object at all, or else to provide us with knowledge of what they signify.

Evans (2010, p. 35) himself invokes this requirement in order to allow that TNSs have a privileged status. The orthodox theistic view is that everything distinct from God is in some sense caused by God. Hence, any of these things could, in theory, point us to God. If there are to be TNSs in any distinctive sense, then, something must distinguish these from other features of the world. A similar line of argument could be offered in the atheistic case: any object encountered in a naturalistic universe would (in some sense) be a product of the naturalistic universe itself; hence, anything could serve as an ANS given atheism. However, such a strategy strikes us as misguided for two reasons. First, there appears to be no principled reason to set a limit on the range of objects capable of serving as TNSs or ANSs. Indeed, as discussed above, both Calvin and Plantinga seem attracted to a view whereby there is no object in the universe incapable of, in the right circumstances, serving as a TNS. Given this, we see nothing problematic in our account’s entailing that everything in an atheistic universe is a (potential) ANS. Second, even assuming we did want to place some restrictions on what could serve as an ANS, there are ways to do so which involve no appeal to intentionality. Various extant accounts of what it is for a sign to signify that something is the case deny both that the intention for that object to serve as a sign is a necessary condition of its being a sign, and that merely being caused by a certain state of affairs, S, is sufficient to signify that S obtains (for example, those presented by Cohen and Meskin (2006), and Skyrms (2010)).¹³

Unintentional signs?

In Plantinga’s case, the requirement for intentionality comes (as noted in “Plantinga’s A/C model”) as the result of more general features of his epistemology. On Plantinga’s view, any warranted belief must be formed on the basis of the operation of some properly functioning faculty successfully aimed at truth. What matters for Plantinga, then, is not that the signs themselves are designed to function as signs, but that we possess some faculty aimed at forming true beliefs about the object signified on the basis of the relevant signs. If this entails that the faculty was literally designed with the purpose of arriving at true beliefs then, as mentioned above, it should be clear that (given the truth of atheism) we do not possess such faculties. It is not obvious, however, that it entails anything of the kind. After all, a cognitive mechanism may (as we will explain in more detail below) be reliably aimed at truth, even though its reliability is not the result of any intentional design. Were this not the case, warranted belief would be impossible given atheism combined with a proper function epistemology. And, while Plantinga may happily accept such a view, most would not.¹⁴ Likewise, it needn’t be the case that functions—whether reliable or otherwise—are the product of intention.¹⁵ It may be that through ‘blind natural forces’, we have evolved to develop perceptual cognitive faculties that are successfully aimed at truth.¹⁶ For we have—as has already been ably argued by, for example, Christopher Stephens (2001)—good reason to think that in many cases evolution by natural selection favours those whose cognitive faculties are successfully aimed at truth. Even granting this, though, those hostile to our account may object that beliefs concerning atheism—in contrast to many of our other beliefs—have very little to do with survival. Thus, while we might willingly accept that we have developed a cognitive faculty that reliably gives rise to ordinary perceptual beliefs (since forming accurate perceptual beliefs about, say, our location relative to the nearest hungry predator would doubtless provide an evolutionary edge), it may be unwise to move from this belief to the further conclusion that we possess some faculty capable of reliably producing true beliefs concerning atheism.

We think, however, that this oversimplifies the situation. It is plausible that evolution would favour those creatures who see their environment for what it is—who see its brute reality in a not-too-overly-optimistic manner. And, conceivably, those who have attributed to their environment ‘blind natural forces’ as opposed to supernatural ones, would be at an advantage in terms of, for example, their ability to respond to natural disasters. Those who have attributed famine, say, to the work of a supernatural agent, may be inclined either to accept their situation as this agent’s will, or else to take various actions—involving prayers, sacrifices, and the like—in order to appease the agent’s displeasure; tactics which would, in a universe lacking such supernatural agents, likely compare unfavourably in terms of survival to alternative approaches (such as devoting all available resources to seeking further afield for new sources of food). Given this, it may well have been common in such a world for evolution to favour those who see their environment as naturalistic; and evolution may have furnished them with a ‘naturalism detecting faculty’—a disposition to see the world as ultimately impersonal.¹⁷

Now, doubtless some will object that this is merely a piece of ‘just so’ speculation about the possible origins of such a faculty. And whatever prima facie plausibility such speculations may have, they hardly establish that we have such a faculty (let alone that it developed in the way we suggested). Again, though, an appeal to parity is in order. Defenders of the TNS approach have been keen to suggest reasons why the God of classical theism, if he existed, would institute such signs—see, for example, Evans (2010, pp. 12–17) and Plantinga (2000, pp. 188–190)—but, again, this is hardly conclusive. Perhaps the God of classical theism would choose not to create any natural signs (or any faculty for detecting them) because he believed that special revelation or natural theology was sufficient to establish his existence; or perhaps he thought—paralleling some arguments offered in response to the challenge from divine hiddenness¹⁸—that such widespread indications of his presence (even if resistible) would still be unduly coercive. The point here is that, in offering a story that is epistemically possible—true for all we know—(given the truth of naturalism) the account we offer here is on no weaker ground than the accounts of religious experience offered by, e.g., Evans and Plantinga. For these also rely on claims that are merely true for all we know (given the truth of theism). Hence, the fact that there are epistemically possible scenarios in which our account is false need not undermine its plausibility (or, at least, need not undermine it any more than it undermines parallel theistic accounts, which is all we need to establish our parity claim).

Genuine absences?

A final worry is that the kinds of experiences we have considered are not really experiences of God’s absence as such, but of something else (the indifference of the universe, its tendency towards disorder etc.). Thus, our account does nothing to vindicate the prospects of appeals to absence. We have two lines of response to this kind of worry.

Firstly, we would not be especially concerned if our account was not, strictly speaking, an appeal to absence. Recall that our broader goals in this paper are to remedy the neglect of appeals to areligous experience, and to argue that there is no reason to think that such experiences couldn’t carry epistemic weight. Thus, since our account is of this more general kind, our underlying point that this neglect is unjustified still stands. Secondly, it seems clear that these experiences are experiences of God’s absence in the sense that the experiences which Evans, Plantinga et al. appeal to are experience of God’s presence. The experiences of the world as contingent, or as exhibiting beneficial order, which Evans describes, are not bare experiences of God’s being present, but experiences which point us towards God’s being present. Less still are the kinds of signs Plantinga describes—an experience of a beautiful spring morning, say—stark experiences of God’s presence. To the extent that we are inclined to count such as experiences as experiences of God’s presence, we should be inclined to allow that experiences of, e.g., the world’s having a propensity to disorder, count as experiences of God’s absence. Now there will, of course, be those who remain unconvinced with respect to either case. Those in such a position can, however, always fall back on our first line of defence concerning (a)religious experience more broadly. The crucial point is that, whichever line of defence we opt for, there will be parity between appeals to religious experiences of the kind Evans and Plantinga propose, and our own appeal to areligious experience.

Summary

The topic of areligious experience has received very little attention in the literature. Our aim in this paper has been to begin to remedy this neglect. The discussion has focussed on the hitherto under-discussed question of whether experiences of God’s absence could provide positive epistemic status for a belief in God’s non-existence. We have argued that there is good reason to accept an epistemic parity between experiences of God’s presence and experiences of God’s absence, and that, if the Evans/Plantinga approach can warrant theistic belief given the truth of theism, then a parallel account of areligious experience can warrant atheistic belief given the truth of atheism. At the very least, we hope to have opened up this unduly neglected topic for further critical reflection.¹⁹