Some options

Atheism comes in many flavours. The most common is simple atheism:

A more complex flavour is evidential atheism:

Doxastic atheism, in contrast, states the following:

Someone accepting either SA or DA can consistently (if oddly) say: I believe that God does not exist, but I have no view regarding the status of our overall available evidence regarding God's existence and thus no view regarding EA. A person could endorse simple or doxastic atheism, then, without being an evidential atheist. Evidential atheists, however, are logically required to recommend belief that simple atheism is true, even if some of them fail actually to believe that God does not exist. The history of philosophy abundantly represents advocates of SA, EA, and DA. (For a recent case for evidential atheism, involving a claim to ‘show’ that no epistemic reasons for belief in God are ‘available,’ see Martin (Reference Martin1990), 38; cf. ibid., 11, 30, 33; for some additional proponents of atheism, see the historical discussions of Buckley (Reference Buckley1987) and Hyman (Reference Hyman and Martin2007).)

Simple theism entails the falsity of simple atheism; it states:

In addition, evidential theism states:

ET entails that we should believe the opposite of what evidential atheism states we should believe.

Doxastic theism, in contrast, states:

Someone accepting either ST or DT can consistently say: I believe that God exists, but I have no view regarding the status of our overall available evidence regarding God's existence and thus no view regarding ET. Accordingly, a person could endorse simple or doxastic theism without being an evidential theist; some fideists appear to fit into this category. Evidential theists, however, must recommend that one believe that simple theism is true, even if some of them fail to believe that God exists.

If reality is just material bodies (large or small) in motion, then simple atheism is true, because God would not be just material bodies (large or small) in motion. This suggests a possible quick case for simple atheism, but a problem arises for this case, regardless of the actual truth-value of SA: we apparently lack decisive evidence for holding that reality is just material bodies in motion. At least, this is a topic of ongoing controversy among philosophers and others.

Another possible quick argument for simple atheism runs as follows: if God exists, the evil in the actual world would not exist; the actual world's evil does exist; so, God does not exist. Here, again, the case would not be decisive, because we have no decisive reason to think that God would not allow the evil in the actual world. Arguably, as the free-will defence (from Plantinga (Reference Plantinga1977) and others) implies, God could create various kinds of beings with free wills, and they could be causally responsible, directly or indirectly, for the evil in the actual world. A problem, in any case, arises from our limited evidential resources concerning divine purposes. We humans are simply not in a position to know that God would not allow the evil in the actual world. Of course, God would be a moral tyrant in causing the evil in the actual world, but simple theism does not imply otherwise. However, we now can bracket any evidential problem of evil for the sake of a different, less appreciated evidential problem.

Potential surprise evidence

A serious problem concerns whether evidential atheism allows for due evidential modesty for humans in the face of potential surprise evidence of God's existence. The problem is particularly vivid in a possible universe where (unbeknown to some) God actually intervenes, if unpredictably, in human experience. (This problem, however, does not assume that God exists in the actual universe.) Accordingly, some people will ask, regarding EA, whether we ever reasonably can suppose that we have canvassed all available evidence in a way that calls for our believing that God does not exist. Such canvassing seems to be a tall order, particularly in connection with the issue of God's non-existence. As a result, it might seem doubtful that we can reasonably recommend our believing that God does not exist.

Part of the problem lies in the vagueness of our talk of ‘available evidence’. Although available evidence can vary among persons, we should not assume that our ordinary talk of ‘available evidence’ offers a precise specification for the ‘availability’ of evidence. Taking ‘available’ in a rather liberal sense, we can ask whether it is always the case that one's turning the next corner, so to speak, could yield ‘available’ surprising, salient evidence of God's existence that is undefeated. If currently undisclosed evidence just around the next corner is part of our overall available evidence, then the actual direction of our overall available evidence of God's existence will be potentially elusive with regard to atheism. In other words, the disclosure of previously undisclosed available evidence in favour of God's existence could defeat prior support for atheism that discounted this evidence, and we apparently have no basis to rule out such disclosure in our available evidence. (On the function of evidential defeaters in general, see Moser (Reference Moser1989) ; cf. Pollock & Cruz (Reference Pollock and Cruz1999).)

We should distinguish ‘canvassing all of our available evidence’ from ‘the direction of our overall available evidence’. Evidential atheism does not require our ‘canvassing all of our available evidence’ or our canvassing any evidence, for that matter. Instead, on the basis of the direction of our overall available evidence, it recommends that we should believe that God does not exist. Our overall available evidence can point in a specific direction without our canvassing this evidence to identify the specific direction. For instance, my available evidence can indicate the presence of a gnat in my grapefruit juice although I fail to identify this indication, perhaps as a result of a distraction. More generally, what our evidence indicates does not depend on what we identify our evidence to indicate. A contrary view would risk a level confusion between what our evidence indicates and what we identify our evidence to indicate.

The plot thickens if we seek to justify (that is, to give a justification for) evidential atheism and thus to go beyond a claim to EA's being true or to one's having evidence that justifies EA. In that case, we have to ask about confirming the direction of our overall available evidence regarding God's non-existence. An answer will be elusive if we face elusiveness in what our overall available evidence actually includes regarding God's non-existence. More specifically, given that we could confront available salient evidence for God's existence around the next corner, an attempt to justify EA apparently faces a problem. We cannot cogently, or persuasively, tell if our total available evidence lacks undefeated evidence of God's existence as long as our available evidence includes currently undisclosed available evidence. So far as we can cogently tell, the latter available evidence could include salient undefeated evidence of God's existence. (Here we can allow for the possibility of various ways in which God provides salient evidence of God's existence in human experience.)

If we were to exclude currently undisclosed evidence from our available evidence, we would implausibly collapse the notion of available evidence into the notion of actually possessed evidence. This would be unacceptable, particularly if we aimed to consider an evidential assessment that bears on our overall available evidence, including evidence we do not yet possess but readily could or will come to possess. We often seek to assess not just the disclosed evidence one actually possesses but also the broader evidence available to one, that is, the evidence one could come to possess without undue difficulty. A case for evidential atheism that concerns only disclosed evidence one possesses, and cannot withstand broader available evidence, would lack a kind of epistemic resilience that we value and often seek. Accordingly, evidential atheism would sacrifice epistemic resilience in retreating from available evidence to a more limited base of just disclosed evidence one possesses.

The problem at hand concerns not the truth of atheism or even one's having evidence that justifies atheism, but rather one's cogently justifying atheism. In reply, one might propose the following: whatever the conditions that justify simple atheism, if those conditions are satisfied by one's case, or argument, for the justification of simple atheism, then one has justified simple atheism. Suppose that some feature, F, is sufficient for the justification of simple atheism (in the absence of defeaters). Perhaps F is either some kind of doxastic coherence, some kind of experiential feature, or some combination of the previous two features. In addition, if one's case for the justification of simple atheism (beyond its being true) possesses F, then one may infer that one's case is justified, on the ground that F is sufficient for justification (in the absence of defeaters). This position properly allows for a distinction between the justification of (=the evidence that justifies) simple atheism and the justification of a case, or an argument, for simple atheism, given that simple atheism is not itself an argument.

We should grant that F's adequacy in the justification of (=the evidence that justifies) simple atheism can be paralleled by F's adequacy in the justification of a case for simple atheism. Accordingly, F could be a feature that confers justification not only on a simple proposition, such as SA, but also on an argument for that simple proposition. Even so, the justification of a case need not be a cogent justification, because it need not be a justification that avoids begging a key question under dispute.

A key disputed question facing EA is whether our currently undisclosed available evidence includes salient undefeated evidence of God's existence. Proponents of EA will beg a key question if they simply assume a negative answer to that question. In addition, they will do the same if they offer a simple inductive inference based on their currently disclosed evidence compatible with atheism. The latter inference will beg the key disputed question of whether our currently undisclosed available evidence regarding God's existence agrees with the direction of our currently disclosed evidence with regard to God's existence.

We plausibly can distinguish between what is justified relative to F regarding atheism and what is justified relative to our total available evidence regarding atheism. The latter evidence would encompass any available defeaters of evidence for atheism (including defeaters in currently undisclosed available evidence) that would be neglected by the more restricted evidence consisting of F. Accordingly, we can ask whether one can cogently justify the non-existence claim of atheism – that is, the claim that God does not exist – relative to our undisclosed available evidence and our disclosed evidence.

The problem is not in cogently justifying a non-existence claim in general. Instead, the problem is in cogently justifying a non-existence claim relative to undisclosed available evidence in a particular kind of context: namely, a context where many otherwise reasonable people report their having experiential evidence for the opposing claim that God exists. The latter context, so far as we can tell, offers the evidentially live potential that undisclosed available evidence includes an undefeated defeater for any disclosed evidence for atheism. In addition, the context includes otherwise reasonable people who plausibly will raise the question of whether an atheist's undisclosed available evidence includes an undefeated defeater for disclosed evidence for atheism. As a result, the serious matter of begging a key question naturally arises. The problem at hand, then, is not a general problem regarding the cogent justifying of a claim that something does not exist. The current problem for cogently justifying atheism offers nothing against cogently justifying a claim that unicorns, for instance, do not exist.

We can clarify the problem by comparing a case for evidential atheism and a case for evidential agnosticism. Such agnosticism states:

A common motivation for agnostics is to avoid error or at least to minimize the risk of error. If relevant evidence about God's existence is highly mixed, then in answering either yes or no to the question of whether God exists, one seriously risks falling into error. The better alternative, according to agnostics, is to refrain from answering either yes or no (that is, to withhold judgement), because this can save one from error. (For a popular variation on evidential agnosticism, see Russell (Reference Russell, Greenspan and Andersson1999).)

Clearly, one pays a price in adopting EG: one then will miss out on an opportunity to acquire a truth. Either it is true that God exists or it is true that God does not exist. Agnostics forgo acquiring a truth in this area of reality, while holding that evidential atheists go too far in the negative direction and evidential theists go too far in the positive direction (at least relative to agnostics' evidence). In contrast, evidential atheism entails that evidential theism and evidential agnosticism make the wrong recommendation on the basis of our evidence. It implies that our overall available evidence counts decisively against simple theism.

Suppose, as appears to be so, that we cannot cogently tell if our total available evidence lacks undefeated evidence of God's existence, given that our available evidence includes currently undisclosed available evidence. As far as we can cogently tell, our undisclosed available evidence actually could include salient undefeated evidence of God's existence. Although this lesson raises a problem for a cogent case for EA, a case for EG can accommodate it, because EG recommends simply withholding judgement that God exists on the basis of our overall evidence.

More specifically, a case for EG can avoid begging a key question begged by a case for EA: the question of whether our undisclosed available evidence includes salient undefeated evidence of God's existence. A case for EG can leave this question wide open, while affirming that we lack the evidential resources to give either a cogent negative answer or a cogent positive answer. In this respect, EG is evidentially preferable to EA. Regarding undisclosed available evidence, evidential agnostics can offer the following advice: withhold judgement until the relevant evidence is actually disclosed. This is in keeping with the statement of EG regarding simply our overall evidence rather than our overall available evidence. Accordingly, one could argue that, from an evidential point of view, advocates of EA should forsake their atheism for the more resilient position of evidential agnosticism. EA, then, is not epistemically stable relative to its being cogently justified in competition with EG.

One might be inclined to counter now that advocates of EA should retreat from the broader base of our overall available evidence to the narrower base of our overall disclosed evidence, in keeping with EG. I have suggested, however, that this would remove a certain kind of epistemic resilience that we value and often seek: namely, the ability of a position to withstand our broader available evidence, beyond the disclosed evidence we possess. In any case, atheists, such as Martin (Reference Martin1990) Martin (Reference Martin1990), often invoke our overall ‘available’ evidence in a manner that does not limit their position to our actually disclosed evidence relative to our overall available evidence. In favouring such a limit, atheists would leave us with a position that has a relatively timid evidential scope, particularly regarding undisclosed available evidence. We might call such a position restricted evidential atheism, to distinguish it from the bolder version represented by EA. Such restricted atheism amounts to agnosticism, rather than atheism, regarding our undisclosed available evidence. In this respect, it involves a retreat from the evidential atheism of EA (and, for instance, of Martin (Reference Martin1990) ).

Evidence and attitudes

Let's highlight the problem of cogently justifying EA in connection with two variations, neutral evidential atheism and positive evidential atheism:

An advocate of PEA could share the following views of Thomas Nagel: ‘I want atheism to be true. I hope there is no God. I don't want there to be a God; I don't want the universe to be like that’ (1997, 130). We also could distinguish a version of evidential atheism that includes our deeming it bad that God does not exist, but NEA and PEA will serve our purpose now.

Unlike PEA, NEA does not include a judgement in favour of the goodness of God's non-existence. Likewise, NEA does not include a judgement in favour of the badness of God's non-existence. In this regard, it recommends neutrality. We apparently have no way to show that such neutrality is impossible for humans. On the contrary, it seems to be a live option, even if many people actually lack such neutrality.

Advocates of PEA face a serious but widely neglected problem in any attempt to give a cogent justification of their evidential agnosticism. The problem stems from the live prospect of intentional divine elusiveness if a God worthy of worship exists. Such elusiveness would include at least the following: God typically would hide God's existence from people ill-disposed towards it, in order not to antagonize these people in a way that diminishes their ultimate receptivity towards God's character and purposes. As a result, we should expect evidence of God's existence typically to be hidden from advocates of PEA; so, their lacking such evidence is not by itself the basis of a case for atheism. Advocates of PEA should expect their disclosed evidence of God's existence to be potentially misleading as a result of God's being purposive and selective (if God exists) in disclosing salient available evidence of God's existence.

In virtue of being worthy of worship and thus morally perfect, the God in question would seek from humans more than their intellectual assent to the proposition that God exists. God would also seek a receptive, agreeable human attitude towards a divine moral character worthy of worship. When people are not yet ready to adopt such an attitude, God reasonably could hide divine evidence from them, so as not to repel them in their responses to God. Such divine hiding is acknowledged by the major monotheistic traditions, and it figures in some contemporary work on theistic epistemology (see, for instance, Moser (Reference Moser2008), (Reference Moser2010) ).

The previous consideration raises a problem for any attempt to give a cogent justification of evidential atheism in conjunction with PEA. The problem concerns not a lack of overall available evidence, but rather a human attitude towards God's existence that potentially interferes with God's purposive disclosing of evidence for God's existence. This problem extends the general problem for EA identified previously to any variation on EA that includes PEA.

One could argue that the same extended problem applies to NEA, but I shall not digress to this matter. The needed argument would rest on the view that God would have an attitude of withdrawal towards human neutrality about the goodness of God's existence; this would be similar to God's attitude towards the human response of PEA. This view would gain plausibility from a case for the following position: for redemptive purposes, God would seek a human attitude of co-operative receptivity towards God's existence and character, and thus would typically hide from people who are not ready to adopt such an agreeable attitude. This approach to NEA is not implausible, but it is independent of the argument of this article.

Consequences for theism

We have seen that, for better or worse, the role of potential surprise available evidence pushes evidential atheists towards evidential agnosticism as an epistemically better option. We should ask if theists share the same fate from potential surprise evidence.

Let's distinguish logical from evidential exclusion of defeaters. It seems clear that, given any disclosed evidence for God's existence, we cannot logically exclude potential surprise defeaters of this evidence in one's overall available evidence. At least, I can find no way to do so, and I find no hope in any ontological argument (on which, see Moser (Reference Moser2010), ch. 3; cf. Oppy (Reference Oppy1995) ). For instance, the disclosing of one's hitherto undisclosed available evidence logically could bring nothing but an unending, uninterrupted onslaught of pointless and excruciating suffering for all concerned. We may plausibly think of the realization of such a logically possible dark prospect as yielding an undefeated defeater for any previously disclosed evidence for God's existence. I, for one, think of it as doing so, because it strongly calls into question the reality of a God who cares for people in virtue of divine moral perfection. Accordingly, I regard the claim that God exists to be logically contingent and falsifiable, but not actually falsified or false.

The mere logical prospect of the defeater in question does not yield an actual defeater of any experiential evidence for theism. (A mere logical possibility does not a defeater make, at least when the contingent evidence regarding a contingent claim is at stake.) The prospect of the defeater would need to be realized in one's experience for an actual defeater (of the sort imagined) to arise. As a result, this prospect does not push theists towards evidential agnosticism. Theists still can have their experiential evidence for theism, and this evidence still can be unaccompanied by actual undefeated defeaters. It follows, so far as this case goes, that one logically can have evidence that justifies theism.

What of the cogent justifying of evidential theism (ET)? Does this meet the same troubled fate as EA? The answer depends on whether undisclosed evidence available to theists could be an obstacle to a cogent justification of ET. Here we move from logical to evidential exclusion of defeaters. The key question becomes: can theists cogently tell if their total available evidence lacks an undefeated defeater of their evidence for God's existence?

The answer will depend on what kind of evidence of God's existence theists actually have. If one's evidence for theism amounts to evidence for a kind of deism that does not offer (evidence of) testimonial evidence from God concerning one's future, then the fate of EA appears to be a genuine threat for ET as well. In that case, one apparently will lack the evidence needed for evidential exclusion of the defeater in question. That is, this defeater will be an evidentially live option relative to one's evidence for deism.

The claims of traditional monotheism differ significantly from deism. Such monotheism, whether in Judaism, Christianity, or Islam, offers purported evidence of God's promise not to abandon God's people to ultimate futility. One can argue that this future-involving evidence offers an evidential (but not a logical) exclusion of the defeater arising from the aforementioned dark prospect. At least, many monotheists would argue thus, and their case may recruit abductive considerations relative to their experience (on which, see Wiebe (Reference Wiebe2004), ch. 3; cf. Moser (Reference Moser2008), 63–69). Aside from the details of their arguments, they do have purported evidential resources to undermine the evidential threat of such a defeater, despite the logical possibility of the dark prospect in question. In this respect, traditional monotheism can make use of purported evidential resources unavailable to atheism. (Clearly, however, this article cannot entertain all of the alleged defeaters facing evidence for traditional monotheism; nor therefore can it develop a full cogent case for such monotheism.)

Of course, atheists cannot consistently make an appeal to future-involving evidence purportedly from a divine knower, but traditional monotheists can, and often do. As a result, the kind of potential surprise evidence that undermines a cogent justification of evidential atheism will not automatically undermine a cogent justification of traditional monotheism. If monotheists can vouchsafe a favourable promise from God, they have an opportunity to undermine the dark defeater in question. (The exact development of this opportunity would take us far beyond the scope of this article.) It follows further that the demand for a cogent justification relative to our overall available evidence need not push evidential monotheism towards agnosticism in the challenging way it pushes evidential atheism.