For Christian philosophers, the putative existence of hell poses a particularly acute version of the problem of evil. Hell, if it exists, is a place of extreme suffering.² At the same time, most (though not all)³ philosophers of religion hold that an omnipotent, omnibenevolent, omniscient God could not permit gratuitous suffering. While the precise definition of “gratuitous suffering” is debated, for present purposes it will suffice to follow Rowe’s definition of gratuitous suffering as “instances of intense suffering which an omnipotent, omniscient being could have prevented without thereby losing some greater good or permitting some evil equally bad or worse” (Rowe, 1979, p. 336).⁴ In other words, God can only permit instances of intense suffering if he has a sufficiently good reason to do so – namely, if his not permitting these instances of suffering would either (i) prevent some countervailing good from obtaining or (ii) would result in some equally bad or worse evil obtaining.⁵ In practice, the distinction between (i) and (ii) is moot, since the avoidance of an equally bad or worse evil would (presumably) constitute a countervailing good for a given instance of suffering. The task for the Christian infernalist, then, is to identify the countervailing good that God explains God’s decision to permit⁶ the suffering of those consigned to hell. Otherwise, hell would appear to be a gratuitous evil, and so is incompatible with the existence of an omnipotent, omnibenevolent, omniscient God.

The traditional view identifies “justice” as the countervailing good that God seeks to gain by permitting the suffering of those consigned to hell (see Adams, 1975; 1993; Bukareff & Plug, 2013). On this view, God permits the suffering of those consigned to hell because this suffering is just punishment for their sins: by allowing sinners to suffer in hell, God is ensuring his justice, a good which he could not obtain were he simply to forgive those consigned to hell. Another view identifies creaturely freedom as the countervailing good that justifies the suffering of those in hell.⁷ On this view, God permits his creatures to freely choose to reject God and so to consign themselves to hell in order to preserve creaturely freedom.

All too often, however, discussions of the problem of hell have been conducted without regard for other areas of Christian doctrine. This is a mistake. Christian doctrine is fundamentally interconnected: eschatology is inextricably bound up with the doctrines of creation, sin, incarnation, atonement, justification, and grace, to name but a few. It is the latter that I wish to focus on in this paper. In particular, it is my contention that an Augustinian understanding of grace undermines both of the above responses to the problem of hell.⁸ For the purposes of this paper, I take the Augustinian view of grace as comprised, at root, by two key claims:⁹ (1)

Grace is unmerited: no human action, belief, or decision can merit the gift of God’s grace.

I take it that (1) is fairly uncontroversial: any remotely orthodox Christian is committed to the claim that grace is unmerited, lest they fall into the heresy of Pelagianism. (2) is more contentious: as the early modern debates regarding the resistibility of grace within both the Catholic and Reformed Churches attest (see Marschler, 2014, and Hampton, 2014), this claim never received universal acceptance, and while some (including, it seems, Augustine himself) have held that (2) is entailed by (1), this argument is open to challenge (Cross, 2005). Nonetheless, there are, I think, compelling theological arguments for accepting (2)) as true.¹⁰ In any case, (2) has been held by a large portion (perhaps the majority) of Western theologians from Augustine onwards, many of whom (including Augustine himself) were also infernalists. As such, it is worth considering whether this claim (together with (1)) is consistent with the claim that God permits the (eternal) suffering of sinners in hell.

Taken together, (1) and (2) fatally undermine responses to the problem of hell which appeal to either God’s justice or creaturely freedom as the countervailing goods for hell. The doctrine of unmerited grace (1) subverts arguments which seek to identify justice as the countervailing good for hell, since it suggests that God can give sinners better than they deserve (by saving them from hell) without violating his justice. Approaches which identify creaturely freedom as the countervailing good, on the other hand, are challenged by the Augustinian view of grace as efficacious (2), inasmuch as this suggests that God is able to secure human consent to grace (and so the salvation of the sinner) without impinging upon creaturely freedom. When considered in conjunction with an Augustinian theology of grace, then, the problem of hell becomes more intractable. For, when taken together, the twin claims that grace is

(1) unmerited and (2) efficacious pose a fundamental challenge to the two standard infernalist responses to the problem of hell in the literature.

In the first section “The Problem of Hell and Unmerited Grace”, I will first set out the problem of hell, understood as a species of the problem of evil. I will show how the claim that grace is unmerited fundamentally undermines attempts to solve this problem by appealing to divine justice as the countervailing good for hell. “The Problem of Hell and Efficacious Grace” introduces the Augustinian claim that grace is efficacious. I offer a brief defence of this claim before showing that it undermines free will responses to the problem of hell. In “Skeptical Infernalism”, I briefly sketch out an alternative response to the problem of hell, based on Augustine’s appeal to the mysteries of the divine will when explaining why God gives efficacious grace to some and not to others. I call this response skeptical infernalism, since it parallels the skeptical theist response to the problem of evil. In the fourth and final section of this paper I respond to two objections to skeptical infernalism.

The predominant conception of hell in the Western Christian tradition is of hell as the place of everlasting torment and unhappiness to which certain sufficiently bad persons are sent at the day of judgement as a punishment for their sins.¹¹ Marilyn McCord Adams (1993) has argued that this view of hell is incompatible with the existence of an omnipotent, omnibenevolent, and omniscient God. According to Adams, the following two statements are incompossible: (i)

God exists, and is essentially omnipotent, omniscient, and perfectly good.

Adams formulates the following argument for their incompossibility: 1.

If God existed and were omnipotent, God would be able to avoid (ii).

Responses to Adam’s argument typically target premise (3). As Adams herself notes, the truth of (3) is not obvious (Adams, 1993, p. 304). (3) is motivated by the fact that (ii) represents an instance of extreme suffering, and by the intuition that a perfectly good God would want to avoid instances of extreme suffering as a consequence of his goodness. Thus, an argument for (3) might look something like this: 3.1.

If God existed and were perfectly good, He would want to avoid any preventable instance of suffering.

Premise (3.1), however, is false. For, a perfectly good God might wish to permit an otherwise preventable instance of suffering if it is necessary in order to obtain some countervailing good (or if God’s intervention would prevent a countervailing good from obtaining), or if it is necessary in order to prevent some greater instance of suffering from obtaining. So, the argument for (3) must be modified as follows: 3.1*.

If God existed and were perfectly good, He would want to avoid any preventable instance of gratuitous suffering (i.e. suffering for which there is no countervailing good).

The question, then, is whether there is (or could be) any countervailing good which justifies God’s decision to permit the suffering of created persons in hell. The standard infernalist response to this problem argues that God’s justice functions as such a good. On this view, God chooses to permit the suffering of sinners in hell because this is necessary to achieve a greater good which outweighs this suffering, namely justice.

Understood this way, the problem of hell becomes a problem concerning God’s justice. In order for justice to be a countervailing good for hell, hell must itself be just. The question, then, on this understanding of the problem, concerns how God could justly condemn someone to eternal suffering in hell.

Advocates of the claim that hell is an expression of God’s justice must answer two particularly thorny objections.¹² The proportionality objection asks how God could justly consign human persons to hell, given the seemingly disproportionate nature of this punishment in relation to human sin.¹³ Or, to put it another way: how can acts committed by human persons (who are finite) merit infinite punishment? The vagueness objection, on the other hand, focusses on the problems posed by borderline cases.¹⁴ There must be some criterion by which God determines who deserves to go to heaven and who deserves to go to hell. Yet on any possible criterion there must be a cut-off point for heaven – a set boundary which divides those who get into heaven from those who get sent to hell. The problem is that, wherever we draw this line, there will always be one person who just makes the “cut-off” for heaven and another who just misses this “cut-off”, such that the person just over the line who goes to heaven is only marginally better than the person just under the line who goes to hell. Yet even though the one person is only marginally better than the other, the contrast between the reward this person gets and the punishment the other person receives is infinitely great. It seems unjust that two people with marginal differences in their moral characters should receive infinitely better or worse rewards from God.

For the purposes of this paper, I am not interested in whether these objections succeed or fail. (There is ample literature on this subject already).¹⁵ Even if these objections can be successfully responded to, there remains a further barrier to the claim that “justice” serves as a countervailing good for hell. For, the doctrine of unmerited grace fatally undermines attempts to present “justice” as the countervailing good for hell. That is to say, even if it can be shown that hell is just, this will not suffice to demonstrate that God’s justice is (or even could be) a countervailing good for hell. In order see how this is the case, a brief comment on the nature of the countervailing goods is needed.

Countervailing goods, I take it, serve an explanatory function. That is, they serve to explain how it could be that an (omnipotent, omniscient, perfectly good) God could permit the existence of a certain instance or instances of suffering. In order for a particular good (or goods) to serve this explanatory function, then, it must meet two criteria.¹⁶ First, the good(s) in question must be of sufficient value to outweigh the disvalue of the instance(s) of suffering it seeks to explain. Were the value generated by the good(s) less than the disvalue of the instance(s) of suffering, then the explanation as to why God permitted the instance(s) of suffering in question would be incomplete, since it would fail to explain how the good(s) in question can serve as sufficient justification for the instance(s) of suffering they are supposed to explain. Second, it must be the case that God cannot intervene to prevent the instance(s) of suffering without also preventing this good from obtaining. Again, this criterion must be met if the good(s) in question are to serve the explanatory function required of them. An explanation for why God permits a particular instance of suffering which appeals to a good God could have obtained without having to permit this suffering (and without having to permit suffering of equal or greater value) would not satisfy as an explanation, since such an explanation would fail to explain why God chose to attain the good this way, rather than some other way which would not require God to permit suffering (or would only require God to permit a lesser instance of suffering).

This latter point can be illustrated using a crude analogy. Say that Michael sees his friend Max about to cross the road into fast-moving traffic. Fearing for his friend’s safety, Michael shoots Max in the leg, thereby preventing him from stepping out in front of a fast-moving truck which would have surely killed him. Let us further say that Michael has other means at his disposal by which to prevent Max from stepping out into traffic: he could have called out to his friend to stop, perhaps, or perhaps he was sufficiently close to his friend that he could rugby-tackle him to the ground before he stood out into the road. Later, in the hospital, Max asks his friend why he shot him. Michael replies: “in order to save your life, by preventing you from stepping out in front of a fast-moving truck”. The good in question certainly outweighs the suffering: the good of still being alive outweighs the suffering caused by the pain Max feels in his leg, the inconvenience of the hospital stay etc. Yet Max would surely not be satisfied with this explanation. For, it does not explain why Michael chose to shoot him in the leg, as opposed to calling out to him to stop, or rugby tackling him to the ground. The explanation is incomplete, since Michael has not explained why he chose this course of action in order to save his friends life, as opposed to an alternative course of action which would have involved less suffering on Max’s part. It is not enough for the good to outweigh the instance of suffering it is meant to justify: in order for the explanation to be complete, we need an explanation as to why this good could not be achieved without permitting this instance or suffering (or some instance(s) of suffering equal or greater disvalue).

We can define the criteria for a countervailing good, then, as follows:

(Def.): For any Good, G, and any instance(s) of suffering, S, G is a countervailing good for S iff:

Thus, in order for God’s justice to serve as the countervailing good for hell, it must be the case that: (i) the value generated by God’s justice is greater than the disvalue generated by the (eternal) suffering of sinners in hell; (ii) if God were to intervene to prevent the suffering of sinners in hell, then the good of God’s justice would fail to obtain (or else some worse instance(s) of suffering would obtain). As Adams notes, this requires that justice be conceived of as follows: a person is just, if and only if (a) she is fair (i.e., treats like cases alike); (b) she treats no one worse than he deserves to be treated and (c) she treats no one better than he deserves to be treated (Adams, 1975, p. 434). (c) is required if God’s justice is to serve as the countervailing good for hell, since otherwise God could give save sinners from hell without violating his justice. If we remove (c) from our definition of justice, then God could avoid sending sinners to hell even if they “deserve” it without thereby ceasing to be just. If this were the case, then God’s could attain the good of justice without having to permit the suffering of anyone hell. If so, then God’s justice would not meet criterion (ii) of our definition of a countervailing good, and so could not serve as a countervailing good for hell.

The problem for the anti-univeralist who wishes to identify justice as the countervailing good for hell is that the above definition of justice contradicts orthodox Christian teaching on grace. On any orthodox understanding of grace, the gift of God’s grace is unmerited. In other words, by giving grace to some people, God gives those people that which they do not deserve. The doctrine of unmerited grace, then, requires ex definitio that God treat people better than they deserve. But if this is the case, then it cannot be true that God’s justice demands that God treat no one better than he deserves to be treated. As such, the orthodox Christian who accepts the doctrine of unmerited grace must reject (c) axiomatically.

Yet having rejected (c) from our definition of justice, justice can no longer serve as a countervailing good for hell. God can intervene and prevent sinners from going to hell – by giving unmerited grace – without thereby losing the good represented by his justice. God does not need to permit the suffering of people in hell in order to act justly, and so “justice” fails to meet criterion (ii) for being a countervailing good. To put it another way, the claim that justice is the countervailing good for hell comes at the cost of the doctrine of unmerited grace. For, if (c) is true, then God cannot give unmerited grace without ceasing to be just. This is too greater cost; and so orthodox Christians should reject the claim that justice serves as the countervailing good for hell.

This is not to say that hell cannot still be understood as an expression of God’s justice. It is one thing to say that hell is retributive – that the suffering inflicted upon those in hell constitutes a form a punishment – and another thing to say that retributive justice is the countervailing good that explains God’s decision to permit hell. Indeed, we might still suppose that being just is a necessary condition of God permitting persons to suffer in hell, even if it is not a sufficient condition. That is to say, it is plausible that God cannot permit anyone to suffer in hell unless they deserve it. Nonetheless, this by itself cannot serve as an explanation for God’s decision to permit this suffering, since God is able to give people better than they deserve without thereby losing the greater good of his justice. Even if hell is just, justice alone cannot serve as the countervailing good for hell.

An alternative response to the problem of hell identifies creaturely freedom as the countervailing good for hell. This response is typically associated with issuant views of hell, according to which those persons in hell are there because they choose to be there, or else as a natural consequence of their own free choices (for instance, as a natural consequence of their free choice to be separated from the presence of God).¹⁷ On this view, so the argument goes, God cannot intervene to prevent the suffering of people in hell without depriving those people of their creaturely freedom. Hence, creaturely freedom serves as the countervailing good for hell.

The free will response ceases to be viable, however, when viewed in light of an Augustinian theology of grace. As I noted above, the Augustinian is committed to two claims about the nature of divine grace: (1)

Grace is unmerited: no human action, belief, or decision can merit the gift of God’s grace.

While (1) undermines the argument that God’s justice serves as the countervailing good for hell, (2), I argue, renders the freedom response equally untenable. Before I make this argument, however, it is worth unpacking (2) in a little more detail. Augustine’s understanding of the efficacy of grace first gains clear articulation in his response to the second question in his letter Ad Simplicianum. Augustine’s account of the efficacy of grace emerges from his reading of Romans 9, and in particular from Paul’s use of the example of Jacob and Esau as an illustration of divine election. The question, for Augustine, is why did God call Jacob and not Esau, given that this calling occurred in the womb, and so prior to any works or faith or act of the will on their part, and given that Jacob and Esau are twins, and so admit of no natural difference between them? The answer, for Augustine, is summed up in two Pauline propositions: non ex operibus, sed ex vocante (“it is not from he who works but from he who calls”, Rom. 9.12); and non volentis neque currentis, sed miserentis est Dei (“it is not from he who wills nor from he who runs, but from God who has mercy”, Rom. 9.16). From the former proposition, Augustine concludes that God’s calling of Jacob is prior any meritorious act on Jacob’s part (Augustine, Ad Simpl. 1.2.6). Not only that, it is also prior to God’s foreknowledge of any meritorious act, since if God were to call Jacob on the basis of this foreknowledge, it would no longer be true that this calling came non ex operibus (Ad Simpl. 1.2.8). God’s calling must also precede faith or foreknowledge of faith, for otherwise Paul would have said that election comes “from he who has faith” as opposed to “from he who calls” (Ad Simpl. 1.2.7).

Yet if faith and good works are consequent upon God’s decision to give grace, as opposed to prior to it, then this means that God produces faith and good works in those whom he calls. This, for Augustine, is the fundamental significance behind the second proposition: non volentis neque currentis, sed miserentis est Dei. God’s mercy is prior not only to human effort, it is also prior to any act of the will. This priority means that even the individual’s willed response to God’s calling is, in some sense, a result of God’s grace. If God is unable to produce the response of the will in those whom he elects, then it would not be the case that grace comes non volentis…sed miserentis Dei, since there would be a sense in which their election was the product of the human act of the will as well as God’s mercy:

If he said, “it is not of him that willeth, nor of him that runneth, but of God that hath mercy,” simply because a man’s will is not sufficient for us to live justly and righteously unless we are aided by the mercy of God, he could have put it the other way round and said, “it is not of God that hath mercy, but of man that willeth,” because it is equally true that the mercy of God is not sufficient of itself, unless there be in addition the consent of our will. Clearly it is vain for us to will unless God have mercy. But I do not know how it could be said that it is vain for God to have mercy unless we willingly consent. If God has mercy, we also will, for the power to will is given with the mercy itself. (Ad Simpl. 1.2.12)¹⁸

God’s calling is entirely efficacious, in that God is able to call his elect in such a manner as to guarantee that they will respond to this calling. If this were not the case, then it would not true that grace comes “not from the one who wills…but from the mercy of God”:

For the effectiveness of God’s mercy cannot be in the power of man to frustrate, if he will have none of it. If God wills to have mercy on men, he can call them in a way that is suited to them, so that they will be moved to understand and to follow. It is true, therefore, that many are called but few are chosen. Those are chosen who are effectually (congruentur) called. Those who are not effectually called and do not obey their calling are not chosen, for although they were called they did not follow. Again it is true that “it is not of him who willeth, nor of him that runneth, but of God that hath mercy.” For although he calls many, he has mercy on those whom he calls in a way suited to them so that they may follow. But it is false to say that “it is not of God who hath mercy but of man who willeth and runneth,” because God has mercy in the way he knows will suit him, so that he will not refuse the call. (Ad Simpl. 1.2.13)

God is able to guarantee the response of those whom he elects by calling them “effectively” (congruentur) – that is to say, God calls his elect in a manner perfectly suited to them and thereby ensures that they recognise the calling and are moved to accept it.¹⁹ God’s calling, then, “is the effectual cause of the good will so that every one who is caused follows it” (Ad Simpl. 1.2.13). At the same time, it is important to recognise that Augustine regards the response of the elect to their calling as a genuinely free movement of the will.²⁰ For Augustine, grace’s sufficiency – that is to say, its efficacy – lies in the very fact that it is able to guarantee that act of the will by means of which the human person accepts and appropriates God’s grace, namely, the act of freely willed consent to God’s calling (see Katayanagi, 1990; Wetzel, 1992, pp. 197-206). God’s grace, in other words, does not override creaturely freedom; nor, however, can it be frustrated by creaturely freedom. Rather, God is able to give grace efficaciously precisely because he is able to give grace in such a way as to guarantee that those whom he wills to be saved will freely consent to freely consent to his call. God’s grace is sufficient, in other words, precisely because it is able to guarantee the correct act of the will on the part of the recipient.

There are deeper underlying theological concerns which undergird Augustine’s exegetical arguments in Ad Simplicianum. In particular, Augustine is keen, on the one hand, to avoid the implication that the elect could in any way merit – and so have reason to boast of – their salvation and, on the other, to uphold the absolute sovereignty of God’s will over creation. If human beings are able to resist God’s grace, then the elect would be able take pride in their not having resisted it. At the same time, if human beings are able to resist God’s grace, then God’s will to save someone could be eternally frustrated, which (Augustine thinks) is to unacceptably limit divine sovereignty. As Augustine puts in in Ad Simpl. 1.2.15: “Who would dare to affirm that the Omnipotent lacked a method of persuading even Esau to believe.” For Augustine, God’s will to save someone can never be frustrated; his mercy is never in vain.

I find Augustine’s account of grace’s efficacy – and the exegetical reasoning which supports it – theologically compelling. A full defence of this account, however, is beyond the scope of this paper. Rather, my interest is in the implications of this claim for the free will response to the problem of hell. Given an Augustinian view of the efficacy of grace, the free will response ceases to be tenable. For, on an Augustinian understanding of grace, God is able to intervene to prevent people from going to hell without thereby violating creaturely freedom, by giving them grace in an efficacious manner. That is to say, God is able to give grace in such a way as to guarantee its free acceptance by the recipient, thereby ensuring their salvation. If this is the case, then condition (ii) of our definition of a countervailing good is not met: God can intervene to prevent an individual from going to hell without thereby losing the good of creaturely freedom, and so creaturely freedom cannot serve as a countervailing good for hell.

The infernalist may be skeptical, however, as to the plausibility of the claim that God is able to guarantee the free acceptance of grace by the elect. At first blush, the very fact that God is able to guarantee acceptance would appear to rule out the possibility that this response could be in any way a product of the creature’s free will. The problem, for the Augustinian, is that efficacious grace appears to have the force of necessity: once God chooses to call someone effectually, to give them efficacious grace, it is necessary that said person accept the offer of grace and be saved. If, however, this acceptance is necessary, then it cannot be free (at least, in a libertarian sense) (see Baker, 2003; Rogers, 2004).

Of course, any argument about the compatibility of efficacious grace with creaturely freedom will depend, at least in part, on one’s definition of freedom. Augustine himself appears to have held to a broadly compatibilist theory of freedom, according to which freedom has more to do with the internal coherence of a person’s motivations, and the ability to act on these motivations than it does to do with the ability to choose between competing alternatives (Wetzel, 1992 pp.219-235). For the card-carrying Augustinian, then, one option is simply to abandon the libertarian conception of freedom as requiring choice between competing alternatives: if libertarian freedom is incompatible with an Augustinian theology of grace, so much the worse for libertarian freedom.²¹ While I am sympathetic to this move, I also want to suggest that Augustine’s account of the means by which God gives grace efficaciously is compatible with the claim that those who receive it have genuine metaphysical freedom to choose otherwise (even if God has acted in such a way as to guarantee that they do not, in fact, choose otherwise). In order to make this argument, however, I will first have to say a little bit more about Augustine’s conception of the means by which God gives grace efficaciously.

Central to Augustine’s understanding of the operation of efficacious grace is what Peter Brown calls the “psychology of delight” (Brown, 1967, pp. 154-155).²² In short, for Augustine God “moves the wills” of his elect by presenting before their minds something which will “delight” and “stir” it to movement (Ad Simpl. 1.2.22). This enables God to achieve the acceptance of grace by the elect by supplying to them such a powerful motivation for accepting God’s grace that they unfailingly do so. When God gives grace efficaciously to his elect, then, their acceptance of this grace follows not from metaphysical necessity but from psychological necessity. God’s giving of efficacious grace unfailingly achieves its end, but this is only because “knowledge of the good is fully and finally motivating” (Wetzel, 1992, p. 217).

Augustine’s thinking can helpfully be understood using a distinction William Wood makes in his analysis of Anselm’s account of the fall of the devil. Wood distinguishes between a particular choice or course of action being a) metaphysically possible for an agent and b) subjectively rational for an agent. In the context of the devil’s fall, Wood argues, it is not enough to explain how the fall of the devil is metaphysically possible by appealing to the devil’s possession of libertarian free will; one must also explain how it could be subjectively rational for the devil to choose to disobey God, even if he is free to do so. Wood illustrates this distinction with the following analogy:

[S]uppose a wealthy benefactor offers you the following “dilemma”. You must either accept a huge sum of money with no strings attached, or else throw yourself out of a twenty-story window. Even if you are free to jump out of the window, why would you? More to the point, if we heard a story about someone who chose the window, we would surely not be satisfied with an explanation that appealed only to the brute fact that his choice was free. Nor would we accept the vacuous explanation that “he just wanted the window more”. We would instead demand an explanation that presents this bizarre choice as intelligible and rational to the agent in question. (Wood, 2016, p. 226)

Applying this distinction to Augustine’s theology of grace, it is not the case that resistance to grace is metaphysically impossible – Augustine does not need to make so strong a claim for the purposes of his account – but, rather, that resistance to God’s grace can never be subjectively rational. For the person who has been called effectually, the offer of grace is so intrinsically motivating – and the alternative of rejecting this grace so intrinsically unmotivating – that the only subjectively rational option is to accept God’s offer of grace.²³ The person who has been called effectually is like the subject of the “dilemma” in Wood’s analogy: even granted the subject’s metaphysical freedom, the intrinsic disparity between the goods on offer mean that the subject has no subjectively rational motivation for choosing to throw themselves out of the window instead of accepting the large sum of money.²⁴

In fact, the disparity of goods on offer when God makes the offer of grace is infinitely greater than in the above analogy. On the one hand, the choice is eternal beatitude with God, the greatest (and so most intrinsically desirable) of all possible goods. On the other hand, the choice is eternal damnation. While it may be metaphysically possible for someone to choose the latter, it could never be subjectively rational. As such, anyone who receives this offer will always unfailingly choose the former (so long, that is, as their cognitive and willing faculties are functioning properly).²⁵ Yet this response to grace is nevertheless genuinely free inasmuch as it is metaphysically possible for the elect to choose an alternative course of action. Indeed, seen from this perspective, the anti-Augustinian who wishes to affirm the resistibility of grace faces a harder problem than the Augustinian, inasmuch as they need to be able to explain how resistance to God’s grace in this scenario is not only metaphysically possible, but also subjectively rational.²⁶

On the Augustinian understanding of grace, then, God is able to ensure the salvation of those whom he elects without violating their freedom. God can achieve this by calling them effectually – that is to say, by calling them in such a way that, while it remains metaphysically possible for them to choose to reject God’s grace, it is not subjectively rational to do so. If this is the case, however, then God is able to intervene to prevent sinners from suffering in hell without thereby losing the good of creaturely freedom. As such, creaturely freedom cannot serve as the countervailing good for hell.

Perhaps we have moved too quickly here, however. It could be argued that Augustine’s notion of efficacious grace only commits us to the claim that God is able to call some people effectually, not that he is able to call all people effectually. That is to say, it does not follow from the fact that God is able to give grace to some in such a way as to guarantee its acceptance on the part of the recipient that God is able to give grace to all in this way.

This, I take it, is the nerve of William Lane Craig’s argument against universalism (Craig, 1989). According to Craig, some persons who exist possess the property of “transworld damnation” – viz., the property of not freely responding to God’s grace (and so being lost) in every world feasible for God. Craig develops this notion in response another account of grace’s efficacy, namely the theory of congruous grace, as proposed by the late 16^th/early 17^th-century Jesuit theologian Francisco Suàrez. According to Suàrez, God calls the elect efficaciously on the basis of his middle knowledge. Through his middle knowledge, “God knows for any individual S under what circumstances S would freely receive Christ” (Craig, 1989, p. 181). All God has to do, then, is bring about these circumstances, and he can ensure that S freely receives Christ. Craig, however, argues that, for some persons, there are no feasible circumstances God can bring about under which S will freely receive Christ (“For some individual S, there are no circumstances under which S would freely receive Christ”) and so “there are no worlds feasible for God in which S exists and is saved” (Craig, 1989, pp. 181-182). God can give grace efficaciously to some, but he cannot give grace efficaciously for those who have the property of transworld damnation, and so God cannot bring it about such that no persons will undergo suffering in hell without thereby losing the good of creaturely freedom.²⁷

Whatever the merits of such a view, the Augustinian must reject it for the simple reason that Augustine himself rejects this view (or, at least, something very similar to it). On Craig’s view, grace can only be efficacious for certain individuals who lack the property of transworld damnation. In other words, God is only able to give grace efficaciously to those who have a prior disposition (understood as the possession – or lack of – certain modal properties) to freely accept God’s grace. As Katayanagi (1990) has shown, Augustine decisively rejects this view as early as Ad Simplicianmum. Augustine has two very good reasons for rejecting such a view. First, if only those who have a good disposition are able to receive efficacious grace, then it would not be the case that salvation comes solely ex vocante, but also from the good disposition of the elect. What is more, the elect would have reason to boast of their salvation, since they could take pride in their not being transworld-damned.

Second, the notion of transworld damnation pulls against the tenor of the scriptural evidence regarding conversion. It is a consistent motif of scripture (Augustine argues) that God is able to call even the most stubborn and obstinate of wills to faith, a point illustrated above all in the example of Paul, who prior to his conversion was will was “fierce, savage, blind”, yet was nevertheless stirred up to faith “so that suddenly out of a marvellous persecutor of the Gospel he was made a still more marvellous preacher of the Gospel” (Ad Simpl. 1.2.22; c.f. Wetzel, 1992, pp. 192-194). Of course, the example of Paul does not prove that there could not be transworld-damned people. Rather, my point (echoing Augustine) is that the very notion of transworld damnation runs counter to the way conversion is depicted in scripture, which emphasises God’s ability to work the conversion of anyone he wills.²⁸

Craig’s notion of transworld damnation thus rescues the doctrine of hell in the face of efficacious grace, but at a theological cost which no good Augustinian should be willing to accept. Instead, the claim that God is able to give grace efficaciously should be taken as indicating that God is able to give grace efficaciously to all, not just to some. And, if this is the case, then God is able to save all from hell without thereby losing the good of creaturely freedom by giving efficacious grace to all. And, so, for the Augustinian, creaturely freedom cannot be the countervailing good for hell.

The prospects for infernalism, given the Augustinian theology of grace sketched out above, do not look great. Against the justice response to the problem of hell, an Augustinian theology of grace suggests that God can give sinners greater than they deserve – by bestowing gratuitous grace upon them – and, thus, that God can intervene to prevent the sinner going to hell without thereby losing the good of his justice. Against the free will response, an Augustinian theology of grace suggests that God can guarantee the free acceptance of grace on the part of the sinner – by calling them effectually and bestowing efficacious grace upon them – and, thus, that God can likewise intervene to prevent the sinner from going to hell without thereby losing the good of creaturely freedom. Given this, both “justice” and “creaturely freedom” fail to meet criterion (ii) for being a countervailing good, and so, on an Augustinian theology of grace, neither God’s justice nor the freedom of creatures can serve as the countervailing good for hell.

As well as undermining the traditional responses to the problem of hell, an Augustinian theology of grace gives rise to a related yet distinct problem for the infernalist: the problem of reprobation. Medieval theologians were well aware of this problem: for instance, Bonaventure raises it in his Commentary on the Sentences, while Thomas Aquinas addresses a version of this problem in his Summa Theologiae.²⁹ In short, the problem is this: if “God wills that all human beings be saved” (1 Tim. 2.4), how can it be that God reprobates – that is to say, does not elect³⁰ – some human beings. In other words, if God is able to give efficacious grace to all, why does he not do so?³¹ Historically, responses to this problem have generally followed one of two strategies. Some theologians have sought to explain God’s reprobation of the non-elect by identifying some other good which explains and justifies God’s decision not to elect (and so save) all human beings. Bonaventure, for instance, argues that God wills not to justify the damned so that his justice might be manifested (Bonaventure, Comm. Sent. I, dist. 41, q.2).³² Thomas Aquinas, on the other hand, argues that God reprobates the damned so that there might exist “diverse grades of beings”, thereby rendering creation complete and manifesting divine goodness (Thomas Aquinas, Summa Theologiae I, q.23, art.5). John Calvin, for his part, suggests that the reprobation of the damned is for the benefit of the elect, who will recognise more clearly the extent of God’s goodness and mercy by regarding the misery of the damned, and so will praise and worship God to an even higher degree as a result (John Calvin, Comm. Rom. 9.22-23).

Each of the above responses adopts a broadly theodical strategy to dealing with the problem of reprobation. Augustine, however, offers an alternative strategy. Rather than seeking to identify some alternative good that God seeks to gain by reprobating the damned, Augustine’s strategy rests on an appeal to the mysteries of the divine will:

Let this truth, then, be fixed and unmoveable in a mind soberly pious and stable in faith, that there is no unrighteousness with God. Let us also believe most firmly and tenaciously that God has mercy on whom he will and that whom he will he hardeneth, that is, he has or has not mercy on whom he will. Let us believe that this belongs to a certain hidden standard of equity that cannot be searched out by any human standard of measurement, though its effects are to be observed in human affairs and earthly arrangements… He decides who are not to be offered mercy by a standard of equity which is most secret and far removed from human powers of understanding. Inscrutable are his judgements, and his ways are past finding out (Rom. 11.33). (Ad Simpl. 1.2.16)

According to Augustine, God’s will to reprobate some is based on “a standard of equity which is most secret and far removed from human powers of understanding.” In other words, God’s will to reprobate some human beings is not arbitrary: Augustine is not denying that there are some God-justifying reasons why God chooses not to have mercy on some people, but merely that these reasons are knowable to human beings.

It is Augustine’s strategy that I wish to pursue here.³³ I propose that Augustine’s appeal to the mysteries of the divine will offers a plausible line of defence for the infernalist when understood in light of the insights afforded by skeptical theist responses to the problem of evil. That is to say, belief in hell can remain rationally defensible for the card-carrying Augustinian if they become a skeptical infernalist. Like the skeptical theist, the skeptical infernalist is skeptical of our ability to make all-things-considered judgements regarding the sorts of countervailing goods that might justify God permitting the (eternal) suffering of people in hell.

The skeptical theist-type response has received surprisingly short shrift in philosophical discussions of the problem of hell.³⁴ Jonathan Kvanvig, for instance, dismisses the possibility of such a move at the outset, arguing that, while this is an appropriate response to the evidential problem of evil, it is a non-starter when it comes to the problem of hell:

Whereas with most evils, the appropriate theological response I favor merely points out that the human condition is one of limited understanding, such a response is of no use when we are told the point of a particular evil. There is no issue regarding hell as to whether it has a legitimate point; the theological traditions that include such a doctrine explicitly state what the point is. (Kvanvig, 1993, p. 9)

Kvanvig’s point seems to be that a skeptical theist-type response does not work in the case of the problem of hell for the very simple reason that we already know what the countervailing good for hell is supposed to be – namely, justice. The challenge instead lies in showing how justice could be a countervailing good for hell (and, indeed, how hell could be just). However, as I argued earlier, we can conceive of hell as being an expression of God’s justice without thinking that justice is the countervailing good God seeks to obtain by permitting the suffering of people in hell. Even if we accept that scripture depicts hell as an expression of divine justice, it does not follow from this that justice is the countervailing good for hell. On the contrary, I have argued that justice cannot serve as the countervailing good for hell even if hell is just.

This distinction may best be grasped by means of an analogy. Consider: a father whose child is ill has a choice between two medicines, A and B. A and B are each equally priced and equally effective. A has no side-affects at all; B, on the other hand, will cause mild nausea and headaches for several days. At the same time, B does come with some ancillary benefits – perhaps it contains some microproteins thought to boost academic performance in infants. Let us now say that the father chooses to administer medicine B to his child. On the one hand, the reason he gave his child B was in order to heal them from their illness. On the other hand, this by itself does not explain why he chose medicine B, given this would cause greater suffering in his child than medicine A. The good of health is not the only good the father sought when selecting medicine B: he also sought to attain a further good in addition, namely, the improved academic performance of his child. It is this further good which explains the father’s choice of medicine B specifically, and so it is this good which explains the father’s decision to allow his child to undergo the suffering brought on by the side effects of B. By analogy, scriptural depictions of hell as an expression of divine judgement and justice may tell us the reason why some created persons are consigned to hell without providing a complete explanation of the goods God seeks to attain by permitting the suffering of people in hell. If this is so, then it is not the case (contra Kvanvig) that scripture tells us what the countervailing good for hell is. Thus, a skeptical theist-type response remains viable, as even if scripture does indicate that hell is an expression of divine justice, we may nevertheless be skeptical of our ability to recognise the full range of goods that God seeks to bring about by permitting the suffering of sinners in hell.

Skeptical theist responses to the problem of evil typically target two kinds of evidential argument from evil: Humean or abductive arguments and what are sometimes called “noseeum” arguments for the existence of gratuitous suffering (Wykstra, 1996). It is the latter which is of interest here. Noseeum arguments appeal to the apparent absence of countervailing goods for certain instances of suffering as evidence for the existence of genuinely gratuitous suffering. Rowe’s 1996 summary of his evidential argument from evil offers a classic example of this kind of inference (where E1 and E2 signify two instances of suffering (E1 = the case of a fawn trapped in a forest fire and undergoing prolonged suffering before dying and E2 = the case of a five-year old girl who is severely beaten, raped, and strangled):

P: No good we know of justifies an omnipotent, omniscient, perfectly good being in permitting El and E2;

For Rowe, our failure to see any countervailing good which explains (or could explain) why God permits E1 and E2 constitutes evidence for the claim that E1 and E2 lack countervailing goods (and so are gratuitous). The inference from P to Q is thus inductive: according to Rowe, P renders Q more probable, such that it is reasonable to believe Q on the basis of P. Skeptical theists, on the other hand, deny that our not-seeing of a countervailing good for an instance of suffering constitutes evidence for that instance of suffering being gratuitous, and so reject the “noseeum” inference from P to Q as invalid. To put it another way, skeptical theists deny that the proposition “we do know of no countervailing good for instance(s) of suffering x” provides rational grounds for affirming the proposition “there is no countervailing good for instance(s) of suffering x”.

The argument for universalism I outlined in the opening section of this paper relies on a similar noseeum inference. Recall Adams’ argument for universalism: (i)

God exists, and is essentially omnipotent, omniscient, and perfectly good.

As I note above, premise 3 relies on the following argument: 3.1*.

If God existed and were perfectly good, He would want to avoid any preventable instance of gratuitous suffering (i.e. suffering for which there is no countervailing good).

I have argued that the Augustinian should reject attempts to rebut 3.2* by identifying either justice or creaturely freedom as the countervailing good for hell. But should the Augustinian therefore accept 3.2*? As far as I can see, the case for 3.2* rests on our inability to identify a countervailing good which could explain God’s decision to permit the suffering of created persons in hell. In other words, the failure of theodical defences of infernalism is the primary evidence in support of 3.1*. The case for 3.2* can thus be presented as follows: 3.2.1*

, We do not know of a countervailing good which would justify God’s permitting the eternal suffering of created persons in hell (given an Augustinian theology of grace).

The case for 3.2* thus rests on a noseeum inference. This observation is not by-itself sufficient to show that the universalist argument problem is vulnerable to a skeptical-theist type response. After all, most skeptical theists accept that some noseeum inferences are valid, even if the noseeum inferences used in evidential arguments from evil are not. Nonetheless, it seems to me that the same considerations which lead skeptical theists to reject the noseeum inferences used in evidential arguments from evil equally undermine the noseeum inference in support of 3.2*.

Skeptical theists offer a number of different reasons for rejecting noseeum arguments for the existence of gratuitous suffering. For present purposes, I will focus on Stephen Wykstra’s CORNEA-based skeptical theism, although I suspect a similar case could be made, mutatis mutandis, for other varieties of skeptical theism. CORNEA-based skeptical theism argues for the invalidity of noseeum arguments for the existence of gratuitous suffering by appealing to a general epistemic principle, namely the Condition of Reasonable Epistemic Access (CORNEA), originally stated as follows:

On the basis of cognized situation s, human H is entitled to claim “It appears that p” only if it is reasonable for H to believe that, given her cognitive faculties and the use she has made of them, if p were not the case, s would likely be different than it is in some way discernible by her. (Wykstra, 1984, p. 85)

A few points of clarification will be helpful here. First of all, in spite of how it is sometimes treated in the literature, CORNEA does not purport to be a statement of the general conditions of epistemic warrant for claims of the form “It appears that p”, but, rather, provides the conditions on which “It appears that p” can be said to be warranted on the basis of “cognised situation s” (see Wykstra, 2007, p. 95, responding to Graham & Maitzen, 2007). In other words, CORNEA is a principle concerning the evidential connection between some “cognized situation s” and the claim “It appears that p”. In this context, a “cognised situation” is simply anything that can serve as evidence (perceptual experience, beliefs, other mental states etc) (McBrayer, 2009, p. 80). Secondly, while in its original wording CORNEA focusses on the evidential connection between some cognised situation and “appearance” claims (where the word “appears” is understood in an epistemic [as opposed to a phenomenal] sense, such that an appearance claim that p constitutes prima facie support for p), Wykstra’s subsequent discussions of the principle have made it clear that it has broader application, setting out the conditions in which some cognized situation can serve as evidence for any given proposition or hypothesis p (whether via an appearance claim – “it appears that p” – or directly) (Wykstra, 2007, p. 88). Specifically, according to Wykstra’s later treatments of CORNEA, the principle seeks to show the conditions on which any given cognized situation can serve as levering evidence for the truth a given proposition or hypothesis – that is, evidence “sufficient to move one from one rational square belief-state to another” (Wykstra, 1996, p. 131). Thirdly, the subjunctive conditional clause in the consequent of CORNEA (“if p were not the case, s would likely be different than it is in some way discernible by her”) should be understood as a probabilistic conditional rather than a counterfactual conditional, so as to avoid the sorts of counter-examples raised by Justin McBrayer (Wykstra & Perrine, 2012, responding to McBrayer, 2009; see also Perrine, 2022).

While CORNEA functions as a general epistemic principle for assessing whether a given cognized situation counts as evidence for some proposition or hypothesis, what is particularly significant for our purposes is that CORNEA provides a criterion against which to judge the validity of noseeum inferences. Noseeum inferences treat the apparent absence of some thing, x, as evidence for the actual absence of x. A noted above, these inferences are sometimes valid and sometimes invalid. For example, it appearing to me that there are no elephants in my office is (presumably) good evidence for there actually being no elephants in my office. By contrast, it appearing to me that there are no tardigrades in my office does not constitute good evidence for there actually being no tardigrades in my office. The intuitive weight of CORNEA lies in the fact that it provides a plausible account of why the former inference is valid while the latter is not. The difference between these two inferences is that I can reasonably expect (given my cognitive faculties and the use I have made of them) that things would likely appear differently to me if there were in fact an elephant in my office, whereas I cannot reasonably expect that things would likely appear differently to me if there were tardigrades in my office. This is why the apparent absence of elephants in my office counts as evidence for the actual absence elephants in my office, while the apparent absence of tardigrades in my office does not constitute evidence for the actual absence of tardigrades in my office.

According to CORNEA-based skeptical theism, countervailing goods are like tardigrades: we cannot reasonably expect, given our cognitive faculties and the use we make of them, that a given instance of suffering would appear differently to us if it had a countervailing good. Skeptical theists offer a number of considerations in support of this claim, including (inter alia) the disparity of our cognitive abilities in contrast to God’s supreme knowledge and wisdom (Wykstra, 1984, p. 88) and our lack of data regarding future goods (and in particular goods pertaining to the afterlife) that God might seek to bring about by permitting suffering in the present (Alston, 1991, p. 38). Yet by the same lights, the infernalist is equally entitled to be skeptical regarding the noseeum argument in support of 3.2*. In light of CORNEA, the relevant question is this: is it reasonable to suppose that, given our cognitive faculties and the use we have made of them, things would appear otherwise if there were some countervailing good which would justify God’s permitting the eternal suffering of created persons in hell? I suggest that it is not. There are strong scriptural grounds for supposing that our knowledge of eschatological matters is exceedingly limited. According to the synoptic Gospels, even the Son himself does not know the day and hour of the eschaton, let alone ordinary human beings: “But about that day and hour no one knows, neither the angels of heaven, nor the Son, but only the Father” (Mark 13.32; Matthew 24.36; Luke 17.36). Paul denies that any human being knows the nature of the eschatological reward of the righteous: “eye has not seen, nor ear heard, neither have entered into the heart of man, the things which God has prepared for them that love him” (1 Cor. 3:9). In a similar vein, the Johannine epistles state that the eschatological state of human beings is unknown: “What we will be has not yet been revealed” (1 John 3:2). Given these epistemic limitations regarding eschatological matters, it does not seem reasonable to suppose that human beings are in a position to be cognisant of the range of actual and possible eschatological goods that might serve as countervailing goods for hell. As such, we are not in a position to know whether God is able to save all persons without thereby losing some countervailing good. The principle of CORNEA, then, provides grounds for skeptical theism and skeptical infernalism alike.

Of course, skeptical theism is itself controversial, and a skeptical infernalist stance which draws on skeptical theism (whether CORNEA-based skeptical theism or some other variety of skeptical theism) will be vulnerable to the same objections faced by skeptical theist responses to the problem of evil. Nonetheless, it seems to me that the skeptical infernalist is at least no worse off than the skeptical theist. Moreover, in at least one regard she is in a slightly stronger position. One of the more worrisome objections against skeptical theism in the contemporary literature concerns the implications of skeptical theism for ethical decision making. Put simply, if I do not know whether a particular instance of suffering is necessary in order for some greater good to obtain, I cannot know whether I should intervene to prevent said instance of suffering (Almeida & Oppy, 2003). This worry is less acute for the skeptical infernalist, however. No created person could hope to intervene to prevent the suffering of people consigned to hell, and so the skeptical infernalist need not be troubled as to whether or not to intervene to prevent this suffering.

The Augustinian theology of grace rendered infernalism rationally inexplicable. It does not, however, render it irrational. The Augustinian can believe in hell, and can do so rationally, just so long as she adopts a stance of skeptical infernalism.