Ordinary morality does not imply atheism

Many theist as well as many atheist philosophers have maintained that if God exists, then every instance of undeserved, unwanted suffering ultimately benefits the sufferer. Recently, several authors have argued that this implication of theism conflicts with ordinary morality. I show that these arguments all rest on a common mistake. Defenders of these arguments overlook the role of merely potential instances of suffering in determining our moral obligations toward suffering.

In order to defend this claim, I will proceed as follows. In ''The actual world formulation'' section, I set out one version of the above argument, and show how it falls prey to the mistake I have highlighted. The version of the argument set forth in ''The actual world formulation'' section, however, is improvable, and indeed in some places some of the authors referred to above have improved upon it. Thus, in ''The necessity formulation'' section, I set forth an improved version of the argument faithful to these authors. I show, nonetheless, that even this improved version suffers from the same problem, albeit less obviously. In ''Responses to objections'' section, I engage with two important objections to my argument. The upshot of the paper, if successful, is that the style of argument employed by these authors cannot show that ordinary morality implies atheism.

The Actual World Formulation
The first version of the argument that ordinary morality implies atheism focuses on the consequences of theism for the actual world. 3 It proposes, in particular, that: (1) If God exists, then all actual instances of undeserved, unwanted suffering produce a net benefit for the sufferer. 4 The defender of the argument then attempts to show that the claim that all actual instances of undeserved, unwanted suffering produce a net benefit for the suffererthe consequent of claim (1)-conflicts in one way or another with ordinary morality. Because it does, God's existence conflicts with ordinary morality. The authors who have defended this kind of argument all explain this conflict either in terms of a conflict with the obligations or the reasons that are constitutive of our ordinary, commonsense moral commitments. The conflict is that while the consequent of (1) implies that we do or don't have some moral obligation or reason for action, ordinary morality dictates otherwise. For example, it might be argued [as in (Jordan 2004, p. 174)] that while the consequent of (1) implies that we never have a moral obligation to prevent undeserved, unwanted suffering, our ordinary moral 3 The arguments in Jordan (2004, p. 174) and Maitzen (2013, p. 260) come closest to endorsing this version, rather than the version to be discussed in ''The necessity formulation'' section. The presentation in Maitzen (2013), however, is perhaps to be explained in part by the context in which it appears-in a book not aimed primarily at philosophers. The earlier (Maitzen 2009), published in a specialist journal in philosophy of religion, presents a version more akin to that discussed in ''The necessity formulation'' section. 4 Although they play no significant role in this paper, two features of claim (1) are worth highlighting for those interested in the broader context of the debate to which this paper is contributing. First, the language of ''producing a net benefit'' has a narrow meaning that is important in some contributions to this debate. By claiming that the suffering produces a net benefit, the consequent of (1) implies that the suffering was necessary for this benefit, and that the net value had the suffering not occurred would not have been as high as it was with the suffering and this benefit. Second, following Maitzen (2009Maitzen ( , 2013, an argument with the same structure as that in the text can get off the ground with a weaker first premise that takes into account the idea of retrospective consent appealed to by some theist philosophers, such as Alston (1996). The weaker premise would be: If God exists, then either (i) all actual instances of undeserved, unwanted suffering produce a net benefit for the sufferer or (ii) all subjects who undergo instances of undeserved, unwanted suffering will ultimately retrospectively consent to this suffering. commitments include a commitment to our sometimes having an obligation to prevent such suffering. Or, it might be argued [as in (Wielenberg MS)] that while the consequent of (1) implies that we have fact-relative reason to cause underserved, unwanted suffering, it would make life absurd if our ordinary moral commitments contained such reason.
For simplification, I will focus my discussion here on the impact of the consequent of (1) on our moral obligations, though what I will say in response to the argument formulated in this way applies just as well to versions of the argument that focus instead on moral reasons. The second claim of the argument can accordingly be presented as follows: (2) If all actual instances of undeserved, unwanted suffering produce a net benefit for the sufferer, then the moral obligations of human persons are not as ordinary morality dictates.
The remainder of the argument can be completed straightforwardly. It is affirmed next that (3) The moral obligations of human persons are as ordinary morality dictates. and so concluded that (4) God does not exist. (from 1, 2 and 3) With (1)-(4), we have an interesting and important moral argument for atheism. While there are various strategies one might employ to object to this argument, 5 I will focus on a strategy that has not yet received sustained attention from defenders of this kind of argument. 6 The strategy objects to claim (2) on the grounds that it is not only the actual instances of an action type that determine the moral status of actions of that type, but the merely potential instances as well. This point is an important one emphasized, for example, by rule consequentialist moral theories, though it is not restricted to them. 7 It is also emphasized, for example, by Kantianinspired moral theories that invoke universalizability-the idea, roughly, that action types are wrong if the actor could not coherently will that all persons in relevantly similar circumstances perform instances of these action types. For my illustrative purposes here, I will focus on the role of merely potential instances of action types in rule consequentialist moral theories.
On such theories, even if all actual instances of an action type have ultimately beneficial consequences, the deleterious consequences of merely potential instances of this action type can render the actual instances morally wrong. 8 For example, if the only instance of a particular variety of injustice in the actual world were one that happened to have ultimately beneficial consequences, it could still be that this injustice was morally wrong because, if human persons were generally to commit such injustices whenever opportunity arose, this would lead to disastrous consequences. One might put the point slightly differently by saying that, while in certain peculiar circumstances, actions of this type can lead to ultimately beneficial consequences, actions of this type nonetheless tend to have ultimately deleterious consequences, and this is why they have the moral status they do.
Abstracting away from rule consequentialist moral theories, the general point is this. The moral status of actions is determined not merely by facts about actual instances of those action types, but by facts about merely potential instances as well. We might express the relevant facts using subjunctive language or we might do so using dispositional language, and I will here remain neutral about whether such language is ultimately interchangeable. 9 If we use subjunctive language, then the relevant facts are facts about what would be the case if there were certain instances of these action types that do not actually exist. If we use dispositional language, the relevant facts are facts about the tendencies of acts of this type, including non-actual instances of such acts.
The lesson for assessing this first version of the argument that ordinary morality implies atheism is the following. Nothing conflicting with ordinary morality follows from the mere fact that all actual instances of undeserved, unwanted suffering produce a net benefit for the sufferer. It does not follow, for example, that there is no obligation to prevent such suffering. For, while it may be that all actual instances of undeserved, unwanted suffering promote a net benefit for the sufferer, nothing has thus far been said to rule out the possibility that if human persons were, as a general policy, to refrain from preventing such suffering whenever they had opportunity (and the suffering were to therefore occur), 10 their doing so would be disastrous. 11 Nothing has been said thus far to rule out the possibility that despite their consequences in the actual world, actions of failing to prevent undeserved, unwanted suffering tend, especially when multiplied, to be overall destructive. But these possibilities are directly relevant to assessing whether it is ever obligatory to prevent undeserved, unwanted suffering. And, indeed, prima facie, it seems that these possibilities are quite plausible. That is, prima facie, it is quite plausible that refraining from preventing undeserved, unwanted suffering tends to be overall destructive, and that if human persons were to as a general policy refrain from such prevention this would be disastrous. Claim (2), then, is highly questionable.

The necessity formulation
While the objection developed in ''The actual world formulation'' section may be convincing enough so far as it goes, the moral argument for atheism presented in that section can easily be improved. Moreover, it can be improved in a way that may at least at first glance appear to avoid the objection raised in ''The actual world formulation'' section. In this section, I consider this revised argument, which I take to better represent the best versions of the moral argument for atheism contained in the writings of the authors with whom I am engaging. I argue that, perhaps somewhat surprisingly, this improved version does not ultimately avoid the objection.
The improved version of the moral argument for atheism focuses on the consequences of theism not just for the actual world, but for all possible worlds. For, given that God's existence is necessary, what is true of all actual instances of undeserved, unwanted suffering is plausibly also true of all merely possible instances as well. Thus, we can replace (1) with (1*) If God exists, then all possible instances of undeserved, unwanted suffering produce a net benefit for the sufferer.
The consequent of (1*) will then replace the antecedent of (2), yielding: (2*) If all possible instances of undeserved, unwanted suffering produce a net benefit for the sufferer, then the moral obligations of human persons are not as ordinary morality dictates.
The remainder of the argument continues as before. With (1*), (2*), (3) and (4), we have a revised moral argument for atheism. Moreover, the revised argument appears to be a significantly improved revision. While there are some who would baulk at replacing (1) with (1*), (1*) will nonetheless be granted by a great many theists and non-theists alike who are prepared to grant (1). So, the cost of this replacement is minimal. Yet, the benefit of replacing (2) with (2*) may appear quite significant. After all, the objection of the previous section was that (2) illicitly assumes that it is only the actual instances of an action type that are relevant for the moral status of actions of that type, when merely potential instances are relevant as well. Yet, a defender of (2*) needn't concede that only actual instances are relevant. Indeed, she can even maintain, for example, that all possible failures to prevent undeserved, unwanted suffering are relevant for the moral status of actual failures to prevent such suffering. How then could she be accused of overlooking the role of merely potential failures to prevent suffering? I actually think there is a good answer to this question. One way to see the answer is to return to the kinds of facts that were cited in section one as relevant for assessing the moral status of actual instances of failing to prevent suffering. I offered there two proposals for how to understand these facts. One proposal used subjunctive conditionals. On this proposal, it is relevant for the moral status of actual failures to prevent undeserved, unwanted suffering what would be the case if human persons were to fail to prevent such suffering as a general policy whenever given opportunity. If failing to prevent such suffering as a general policy would be disastrous, then this speaks against the morality of actual failures to prevent such suffering. The other proposal used dispositional language. On this proposal, it is relevant for the moral status of actual failures to prevent undeserved, unwanted suffering whether such failures, or such failures when multiplied, tend toward disaster. If they do, this speaks against the morality of actual failures to prevent suffering.
The way in which the revised argument overlooks merely potential failures to prevent suffering can now be explained in two ways. First, claim (2*) will only be plausible if facts about all possible instances of undeserved, unwanted suffering are sufficient to determine all that is relevant to assessing the moral status of actual actions concerned with such suffering. But, the subjunctive facts cited in the previous paragraph, which are relevant for assessing the morality of actual failures to prevent undeserved, unwanted suffering, are quite plausibly not determined by the facts about all possible instances of such suffering. For, given (1*), it is plausible that there is no possible world in which human persons, as a general policy, fail to prevent undeserved, unwanted suffering. God simply never would create such a world, precisely because such a world would be morally disastrous. 12 Any world in which human persons as a general policy fail to prevent undeserved, unwanted suffering is an impossible world. So, subjunctive facts about how things would be if human persons as a general policy failed to prevent such suffering are not merely counterfactuals, they are counterpossibles. As such, their truth-values cannot be determined solely by looking at possible worlds and at what failures to prevent suffering are like there. Rather, their truth-values are determined by examining how things are in impossible worlds in which human persons do as a general policy fail to prevent undeserved, unwanted suffering. Given that the disastrousness of such worlds is precisely the reason why they are impossible, it is plausible for the theist to maintain that it would be disastrous if human persons were (counterpossibly) to fail to prevent undeserved, unwanted suffering in the way they do in such worlds. And because it would be, human persons do have a moral obligation to prevent such suffering in the actual world.
Much the same line of argument can be applied to the facts expressed using dispositional language. Given (1*), the question of whether failures to prevent undeserved, unwanted suffering tend toward disaster cannot be answered by looking to the manifestations of possible instances of such failures alone. All of these, given (1*), will produce net benefits for the sufferer. But, this may be the case precisely because these instances of failing to prevent undeserved, unwanted suffering are exceptions to the rule. They are permitted by God only because, given certain peculiarities about the sufferer and the person who fails to prevent her suffering, they produce a net benefit for the sufferer. But, they do not generally tend to produce such benefits. They tend instead to produce disaster. It is just that this tendency is covered up in every possible world because, in every possible world, God only permits instances of such failures that, because of the peculiar circumstances in which they take place, happen not to give rise to the disaster toward which they tend.
Each of these lines of argument against claim (2*) in the revised moral argument for atheism can be strengthened through appeal to recent philosophical developments. The first line of argument that appeals to counterpossible conditionals about what would be the case if human persons were to, as a general policy, fail to prevent undeserved, unwanted suffering, can be strengthened by appealing to recent work on counterpossible conditionals. It has become increasingly popular, especially within recent work in the Philosophy of Religion, to deny that all counterpossible conditionals are trivially true. One finds, for example, on both sides of the aisle of the debate about the compossibility of perfect goodness and omnipotence authors who maintain that there are non-trivially true as well as false counterpossible conditionals. 13 This bolsters the key idea appealed to here that there is a non-trivial fact of the matter about the disaster that would (counterpossibly) obtain were human persons to embark on a general policy of failing to prevent undeserved, unwanted suffering. And given that there is such a fact of the matter, there is good reason to deny (2*), according to the moral theories cited earlier in which potential instances of action types play an important role in determining the moral status of actual instances of those action types.
The second line of argument against (2*) can be bolstered through appealing to recent work defending the fundamentality of dispositions. This work challenges views which attempt to analyse dispositional ascriptions in terms of subjunctive conditionals. It is commonly pointed out in this work that dispositions can be masked so that they do not give rise to their constitutive manifestations. The tendency of a fragile vase to shatter when dropped, for example, can be masked if the vase is wrapped in a protective covering. 14 This bolsters the key idea appealed to in the second line of argument against (2*) that particular failures to prevent undeserved, unwanted suffering can have a tendency toward disaster even if they do not give rise to it in light of the peculiarities in which they take place. Yet, given the moral theories alluded to earlier, the fact that failing to prevent instances of undeserved, unwanted suffering tends toward disaster, or the fact that it does so when practiced as a general policy, provides reason to think that we do have an obligation not to fail in this way.
Thus, perhaps somewhat surprisingly, it turns out that even the improved, necessity formulation of the moral argument for atheism is threatened by much the same objection that threatens the weaker, actual world formulation. Some may quibble with my description of this problem as involving an oversight of the role of merely potential instances of undeserved, unwanted suffering. After all, the instances in view in this section are impossible instances, and so one might well maintain they are not potential at all. The point of the section remains, though, whether or not we eschew the convenient abbreviation ''potential instances of suffering.'' What matters is that the subjunctive and dispositional facts cited in this section are relevant for the moral status of our actual behaviours toward undeserved, unwanted suffering, and these facts are not determined by the fact that all possible instances of undeserved, unwanted suffering produce a net benefit for the sufferer. Thus, (2*) is in jeopardy. The support for atheism offered by the necessity formulation of the moral argument for atheism is consequently much weaker than its defenders have thought.

Responses to objections
In this final section, I anticipate and respond to the two best objections to the argument I have thus far presented of which I am aware.

The subjunctive and dispositional facts cannot be known
It is important for the defence of my argument that certain subjunctive or dispositional facts can be known by the theist. For example, the theist needs to be able to know that were all human persons to as a general policy fail to prevent undeserved, unwanted suffering, this would be disastrous. Or she needs to be able to know that failing to prevent undeserved, unwanted suffering tends toward disaster. For, it is on the basis of such facts that she claims that any world in which human persons were as a general policy to permit undeserved, unwanted suffering is an impossible world. And this is key to her resistance to the moral argument for atheism. Yet, it seems problematic to maintain that she can know these facts.
To see the problem, note the following asymmetry between the atheist and the theist. The atheist can maintain that we know the relevant facts, and that our knowledge of the relevant facts is inferred from our knowledge of the badness of particular, actual instances of failures to prevent undeserved, unwanted suffering. For instance, it is because we think that actual failures to prevent undeserved, unwanted suffering are disastrous that we are inclined to think that disaster would ensue if we as a general policy engaged in such failures. The theist, however, cannot accept this explanation of our knowledge of these facts. For, she denies that actual failures to prevent undeserved, unwanted suffering are disastrous. How, then, can she explain our knowledge of these facts, which is so important to her objection to the moral argument for atheism?
I propose the following response on behalf of the theist. Independent from the assumption of (1) or (1*), the theist has just as much reason as the atheist to affirm that actual failures to prevent undeserved, unwanted suffering are disastrous. When she comes to accept (1) or (1*) as a result of her theism, it surprises to her to find out that that actual failures to prevent undeserved, unwanted suffering ultimately benefit the sufferer. There must be something peculiar about these actual instances that explains how it is that they ultimately benefit the sufferer, she thinks. Indeed, to the extent that it is not plausible that there is something peculiar about them that explains how it is that they ultimately benefit the sufferer, her confidence in theism will diminish. If the world contained many more and various instances of these kinds-in particular, if it were to be the case that every human person upon every opportunity permitted all manner of suffering to her fellows-she would give up her theism altogether.
Theists in this kind of evidential position can plausibly maintain that they know the relevant subjunctive or dispositional facts needed to embrace my response to the moral argument for atheism. Independent of (1) or (1*), their evidence would support the claim that certain actual failures to prevent undeserved, unwanted suffering do not ultimately benefit the sufferer. (1) or (1*) provides a defeater for this. However, the conjunction of their former evidence with (1) or (1*) still supports the relevant claims from my objection: that failing to prevent undeserved, unwanted suffering tends toward disaster, and that were human beings as a general policy to fail to prevent undeserved, unwanted suffering, this would be disastrous.
We might express the point using the language of anomaly. The theist who accepts (1) or (1*) and wishes to coherently employ my response to the moral argument for atheism has the following cognitive features. Her commitment to theism and to (1) or (1*) can withstand certain anomalies, or apparent counterinstances-instances of undeserved, unwanted suffering that don't seem to her, independent of (1) or (1*), to benefit the sufferer. In particular, her commitment to theism can withstand apparent counterinstances comparable to those we find in the actual world. Yet, there is a threshold of such apparent counterinstances beyond which her commitment to theism will be retracted. And this threshold would be surpassed exceedingly if the world were such that human beings as a general policy failed to prevent undeserved, unwanted suffering whenever opportunity arose. Such a theist, I propose, can coherently maintain that she knows the relevant subjunctive or dispositional facts necessary for defending my objection to the moral argument for atheism, while denying that this knowledge is based on her knowledge that actual instances of undeserved, unwanted suffering are morally disastrous. 15 Of course, not every theist will be willing to embrace the perspective required here concerning the extent to which evil disconfirms theism. Some theists will prefer the (in my opinion, very strong) view that no matter what evils there are, they do not disconfirm God's existence. 16 Perhaps, for example, it is by virtue of God's omnipotence that for any evils whatsoever, God can guarantee that those evils are required for promoting outweighing goods for those who suffer them. 17,18 Such theists cannot embrace the response to the moral argument for atheism that I have developed in this paper; they must search for an alternative. However, for any theists who do not embrace such a view regarding the extent to which evil disconfirms theism-and there are plenty 19 -the response I have developed here remains a viable option. Thus, while the present objection teaches us that the response to the moral argument for atheism I have developed in this paper is not universally available for theists, it does not show that the response does not expose a significant shortcoming in this argument.

The response eliminates free will
A second important objection to my response to the moral argument for atheism is that it eliminates free will. This is because the response maintains that it is impossible for human persons as a general policy to fail to prevent undeserved, unwanted suffering. Yet, if it is impossible for human persons as a general policy to fail to prevent undeserved, unwanted suffering, then human persons are not free to as a general policy fail to prevent undeserved, unwanted suffering. And surely they are free to do so.
I think there are several worthwhile responses to this objection. One might question, first, why we should maintain that human persons are free, as a general policy, to fail to prevent undeserved, unwanted suffering. As some of the authors who have defended the arguments with which I am interacting in this paper have highlighted, many human beings find the prospect of inflicting or failing to prevent undeserved, unwanted suffering psychologically unbearable. 20 Yet, it is quite a fashionable view that a psychological make-up that makes certain kinds of actions unbearable can render one unfree with respect to performing those actions. A good parent, for example, is not free to torture her child for a nickel. 21 Second, one might argue that while it is not possible that every actual person embarks simultaneously on a programme of failing to prevent undeserved, unwanted suffering, it nonetheless is possible of each actual person that she fails to prevent such suffering. In other words, there are possible worlds where each of us fails to prevent such suffering, but no possible worlds where all of us do so routinely. Yet, it is enough to retain our freedom to fail to prevent such suffering that each of us in some possible world fails to prevent such suffering. At least, these 17 One anonymous reviewer suggests that this is how Adams's (1999) theodicy should be understood. This seems to me a controversial interpretation, however, as her explicit aim is limited to providing a theodicy for the sorts of evils we find in the actual world. 18 One problem with such a view is that it assigns to God control over modal facts-specifically, facts regarding what is required for what. The relationship between God and modal facts, however, is hotly debated. For a sample of the recent debate, see Gould (2014). 19 For a representative example of the approach to the evidential relationship between evil and God invoked here that itself uses the language of anomaly, see Doughery and Pruss (2014). 20 This idea plays an important role, particularly, for Wielenberg (MS). possibilities are enough to retain the extent of freedom to fail to prevent suffering that plausibly is enjoyed by us.
Third, and perhaps most controversially, one might maintain that the fact that it is impossible for all of us simultaneously to embark on a programme of failing to prevent undeserved, unwanted suffering is not a sufficient reason for thinking that we are not free to do this. In much the way that failures to prevent undeserved, unwanted suffering can have a tendency to give rise to disaster, even if there is no possible world in which this tendency is manifested in disaster, we human persons have a power or ability to simultaneously embark on a programme of failing to prevent undeserved, unwanted suffering, even if there is no possible world in which we exercise this power or ability. Yet, the relevant power or ability is enough to retain the freedom the objector seeks to maintain. In much the way that some have recently maintained that God is free to bring about instances of suffering that ultimately do not benefit the sufferer, despite the impossibility of God's doing so, 22 human persons are free to embark on the relevant disastrous programme, despite the impossibility of their doing so. The fact that the relevant actions are impossible for God is no sign of God's lack of control over Godself, but rather is a sign of God's control over Godself. Likewise, one might maintain that if it is impossible that all us human persons simultaneously of our own choosing to embark on a programme of systematically refraining from preventing suffering, this is a sign of at least some of us having significant control over ourselves-we simply wouldn't let ourselves do this. So, even if behaving this way is impossible, it may be that this does not rule out our being free to do it. There are, then, several initially promising responses to this second objection as well.

Conclusion
This paper has developed an objection to a kind of argument recently employed by several authors to show that theism conflicts with ordinary morality. All extant defences of this kind of argument make a common mistake in overlooking the role of merely potential instances of suffering in determining the nature of our moral obligations or reasons. This is not to say that no moral argument for atheism stressing the surprising consequences of theism for actual or possible instances of suffering can be defended. But arguments of the kind examined in this paper cannot, without the present oversight being addressed. 23 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. 22 See, in particular, (Byerly forthcoming). 23 Thanks to audiences at the Butler Society for Philosophy of Religion and the Glasgow Philosophy of Religion Seminar, both in 2016, for helpful feedback on drafts of this paper.