Free will and the moral vice explanation of hell's finality

Abstract

According to the Free Will Explanation of a traditional view of hell, human freedom explains why some human persons are in hell. Human freedom also explains its punishment and finality: persons in hell have freely developed moral vices that are their own punishment and that make repentance psychologically impossible. So, even though God continues to desire reconciliation with persons in hell, damned persons do not want reconciliation with God. But this moral vice explanation of hell's finality is implausible. I argue that God can and would make direct or indirect alterations in their character to give them new motivational reasons that re-enable their freedom to repent. Subsequently, I argue that it is probable that each damned person will be saved eventually, because there is a potential infinity of opportunities for free repentance. Thus, if the Free Will Explanation's descriptions of hell and divine love are correct, it is highly probable that each person in hell escapes to heaven.

God loves all human persons. As such, God wants good for them and union with them (Stump (2010), 91). The highest good for human beings is union with God. Thus, God desires the salvation of all human persons. But according to a traditional view of hell, God's desire for universal salvation is frustrated, because (i) at least some human persons are damned in hell, (ii) persons in hell are punished, and (iii) persons in hell do not escape (Kvanvig (1993), 19). How could God's omnipotent will be frustrated in this way?

The Free Will Explanation aims to vindicate God's love and power by locating the obstacle to universal human salvation in human free will. A person acts freely, according to a basic leeway libertarian account of free will, if and only if the action was not causally determined, she had alternative possibilities, she had the power to choose between those alternatives, and she chose in virtue of exercising that power (e.g. O'Connor (2005); Franklin (2018)).¹ This kind of free will provides the resources for philosophers and theologians to explain the reason, punishment, and finality of hell as a product of human free choice in a way that puts God beyond reproach. I unpack these three explanations over the next few paragraphs.

First, the reason that at least some persons are in hell is that they have freely chosen it, or their being in hell is a foreseeable consequence of their free choices (Lewis (1940), 120; Swinburne (1983); Stump (1986), 194–195; Walls (1992); Kvanvig (1993); Timpe (2014)). If God gives free will to human beings with the option of rejecting a relationship with God, then God cannot unilaterally guarantee their salvation, because it is impossible for God to determine causally a human person's free choice. If human persons do freely reject God's invitation to salvation, God does not causally determine them to accept the invitation. There are various reasons: God respects human autonomy (Swinburne (1983), 49); entering a loving relationship requires a free choice (Kvanvig (2011), 53); or the best kind of loving relationship must begin with a free choice to enter into the relationship (Rasmussen (2013), 424–425).

Second, the punishment of hell is not externally imposed by God. Rather, the punishment is constituted by suffering that is concomitant with the exercise of vicious character, which was previously freely acquired (for example, see Lewis (1940), 124–125; Swinburne (1983), 47–49; Stump (1986), 195–197; Walls (1992), 150–155; Timpe (2014), 72). This natural consequence view of punishment in hell, according to which the sin is its own punishment, is most famously illustrated by Dante in the Divine Comedy.² Yet God's love is continually manifested to damned persons by God's desiring their salvation and by offering grace necessary for their salvation, if it could make a difference (Lewis (1940), 126; Kvanvig (1993), 109; Timpe (2014), 76–77). God loves all human persons at all times, both before and after death.

Third, hell is final, because the moral vices that were previously freely acquired ensure that no damned person freely repents.³ Their moral vices silence all potential reasons for repentance and free actions are necessarily done for reasons (Lewis (1940), 130; Swinburne (1983), 49; Walls (1992), 113–138; Timpe (2014), 71–81). As C. S. Lewis (1940, 130) puts the idea, ‘[T]he doors of hell are locked on the inside.’⁴ God waits on the heavenly side of the door and desires that damned persons walk through – and is ready to provide them helping grace if it can make a difference – but the moral vices of damned persons distort their moral perceptions in a way that they cannot see a reason to open that door. Thus, they cannot freely open the door.

The Free Will Explanation is a philosophical justification for the reason, punishment, and finality of a traditional view of hell. It is, thus, liable to philosophical objection.

I argue that the Free Will Explanation's appeal to silencing moral vices cannot adequately explain the inescapability or finality of hell. I proceed as follows. First, I argue that God can make interventions in vicious human character to reopen the possibility of their freely choosing to repent, in view of contemporary work on free will and character. Second, I contend that God would make such interventions by appealing to some theological commitments of the Free Will Explanation and to plausible intuitions about trade-offs between autonomy and well-being. Third, I argue that it is highly probable that each person will eventually leave hell for heaven (or for purgatory and then heaven),⁵ in part because damned persons have a potential infinity of chances to freely repent.⁶

My thesis is the following conditional claim: if the models of hell and divine love in the Free Will Explanation are correct, there is a high probability that each damned person will freely repent eventually, which implies that the Free Will Explanation of the finality of hell is implausible. It is worth clarifying two features of this thesis. First, the conditional's consequent is merely that there is a high probability that each damned person will be saved, which is compatible with the possibility and actuality that some persons in hell are damned forever. Second, accepting my conditional thesis as true does not require accepting as true its consequent – namely, that there is a high probability that each damned person will freely repent eventually. A person is rationally required to embrace the consequent on the basis of the conditional thesis only if she also accepts the antecedent as true – that is, she accepts the models of hell and divine love in the Free Will Explanation as correct. But a person might embrace my conditional thesis and yet deny one of those models. For example, one might believe that in God the desires of love are overwhelmed by desires for justice, and so what God ultimately wants is to punish retributively all damned persons forever (Augustine (1993), 778–809). In that case, one could rationally embrace my conditional thesis while rejecting the conditional's consequent. Thus, I do not argue for the claim that there is a high probability that each damned person will freely repent eventually; rather, my argument reveals an internal instability in the Free Will Explanation concerning the finality of hell.

Human freedom and what God can do

Character traits are mental dispositions to notice, think, feel, and act in various ways in trait-relevant circumstances. For example, when a compassionate person is with people who are suffering, the compassionate person has a heightened tendency to notice their suffering, to believe that their suffering is bad, to feel pity for them, and to be motivated to act in ways that alleviate their suffering. Character traits are also stable, relatively thick, and normatively assessable (Miller (2014), 3–36). In concrete terms, a compassionate person tends to be compassionate through time; compassion is activated in a broad enough range of circumstances in which others suffer; and possessing compassion is a mark in favour of being a morally good person.

Moral vices are character traits. They have morally bad focus, beliefs, desires, and motivations, and they involve at least near wholehearted motivation to perform their characteristic actions in trait-relevant circumstances. As such, moral vices silence corresponding morally good reasons for action (McDowell (1979), 336, 346). The viciously spiteful person cannot see reasons to rejoice genuinely in the good of others.

The moral vices of damned persons rule out the ability to accept freely God's offer of salvation by silencing the reasons to accept.⁷ Free actions must be done for reasons (see Timpe (2014), 21–24; Hartman (2020), 1419–1421). It is only through our character that we see the weight of reasons, and the vicious character of damned persons precludes their seeing reasons to repent, which is why damned persons are unable to do so.

Consider the point in terms of Harry Frankfurt's (1988, 86) concept of ‘volitional necessity’. A person is volitionally necessitated to act in a particular way if she must act in that way given who she is. Other options are unthinkable for her; her reasons are ‘too good’ to do anything else (Frankfurt (1988), 86; cf. Williams (1993), 65). To employ Eleonore Stump's (1999, 323) example, a mother is offered a nickel to cut up her daughter into tiny pieces; her character volitionally necessitates a refusal. That is, her reasons to refuse are so good that she sees no reasons on behalf of accepting the offer. Of course, she retains the general capacity to accept the offer – she can say the word ‘yes’ or nod her head affirmatively – but she cannot will to use those general capacities in those ways in this circumstance (Frankfurt (1988), 86).

The moral vices of damned persons volitionally necessitate their rejection of God's offer of salvation in ways that are ensured by their dominant moral vices. Damned persons cannot even freely intend to do something (or a series of somethings) such that if they do it (or them), they would regain their ability to repent, because they also see no reasons in favour of regaining that ability to use their more general capacities for repentance.⁸ According to the Free Will Explanation, the freely acquired moral vices of damned persons explain why no-one leaves hell and how the doors of hell are locked from the inside.

The problem with this explanation is that it requires that the relevant character traits of damned persons cannot change. Of course, it is true that damned persons cannot freely change their character in a way that reinstates their ability to repent, because they see no reasons to do so and free actions must be done for reasons.

But a person's character can change over time in ways that are significantly influenced or caused by the external world. For example, Phineas Gage after being impaled by a railroad spike, abruptly changed in character from being a person who is reverent and mild to being a person who is irreverent, angry, and impatient. Even Aristotle (2014, 107; 1101a10) agrees that good character can be corrupted by ‘great and repeated’ misfortunes. For example, a person taken into slavery finds herself in circumstances that naturally diminish self-esteem and autonomy given the conditions of deprivation, abuse, and confined options (Nussbaum (2001), 336–340); or a person who develops chronic fatigue syndrome can naturally develop selfish tendencies by prioritizing her needs over others as adaptive strategies to avoid the catastrophic effects of overexertion.

Importantly, if those kinds of external pressures and circumstances can change a person's character, an omnipotent deity can easily do so. God can directly alter a person's character. That is, God can remove or add mental dispositions, or both. For example, God hardens Pharaoh's heart in such a way that he refuses to allow the Israelites to go into the wilderness to worship God (Exodus 4–14; cf. Stump (1988), 414–420). God can also indirectly alter a person's character by putting them into circumstances that make changes in their relevant mental dispositions to be at least highly probable. In other words, God can place people in circumstances in which they probably gain true beliefs, lose false beliefs, develop good preferences, and lose bad preferences. For example, God eventually spurs humility in Nebuchadnezzar by turning him into a beast (see Daniel 4:28–37), or God assigns a ‘thorn in the flesh’ to Paul to prevent his becoming boastful (2 Corinthians 12:7–10). The possibility of these processes is taken for granted in the contemporary free will debate (for example, see Pereboom (2014); Mele (2019)), and an omnipotent deity is obviously able to bring them about.

Suppose that God alters or suppresses moral vices of damned persons that had previously volitionally necessitated their rejection of God's invitation to salvation, and this intervention enables the damned person to see reasons to repent. In a circumstance in which God does so and offers saving grace, the damned person's two-way free choice is reinstated, because she sees some reasons to accept God's invitation to salvation and she sees other reasons to reject the offer.⁹ In this new condition, damned persons can freely choose to accept God's invitation to salvation for the reasons that they are now able to see.

Let us consider three objections to my argument that God can give damned persons new reasons that reinstate freedom of choice to repent and offer responses to each.

First, God cannot give damned persons new reasons to repent, because damned persons already know everything relevant to the decision for or against God's offer of salvation; and getting new information is the only way for a person to change their mind. For example, damned persons now know that God exists; they know all of the sins they have committed in their pre-mortem life; and they have seen the blessedness of the redeemed (see Aquinas (1948), 2919–2920; ST Supp q. 87, a. 1).

But seeing new reasons to repent is not entirely a function of new information. Importantly, reasons to act are not only constituted by beliefs but also by desires. God can augment or generate desires and preferences for the good in damned persons. Thus, it is possible for damned persons to gain new motivational reasons for repentance, even without acquiring new information. Contrary to the objection, however, it also seems plausible that damned persons are not omniscient concerning the information relevant to the decision for salvation; they are self-deceived in ways that promote inattention or ignorance.¹⁰ God can dredge up their merely dispositional beliefs and put them back into the spotlight of their attention; in this way, God can change the structure of thinking that is involved in constructing live action options. God can also instill new true beliefs about their own vicious and unhappy condition, the moral order, and God's self. Thus, the vast knowledge of human persons in the damned condition does not rule out God's ability to generate in them new reasons for repentance that can reinstate their freedom to repent.

Second, a damned person cannot freely choose to repent, because that choice is from reasons and character that she has not freely formed if God intervenes in the way previously explained. This second objection presupposes the Character Claim:

(CC): S performs free action A only if S has freely formed the parts of her mental dispositions that motivate and explain her performing A. (Strawson (1994), 6)

If the CC is true, the damned person cannot freely choose to accept God's invitation to salvation for the reasons brought about in her by God, because the choice would be motivated and explained by reasons and character that have not been previously formed by her own free choices.

But proponents of the Free Will Explanation must reject the CC, because they are committed to the metaphysical possibility of acting freely. The CC combined with extremely plausible premises imply that acting freely is impossible. Here is a simplified version of Galen Strawson's (1994) argument for the impossibility of free action: the CC states that a person acts freely only if the action is motivated and explained by mental dispositions that have been formed by previous free choices. But a person's first free action must be motivated and explained by reasons and character that have not been formed by previous free choices precisely because it is the first free choice. Thus, the CC rules out the possibility of a person's performing her first free action, and so it is impossible to act freely.

To avoid that conclusion in a plausible way, the proponent of the Free Will Explanation must deny the CC. She must embrace the idea that capacities for free choice emerge from the right kind of non-freedom conditions (Hartman (2018), 175–176). The relevant conditions include some knowledge of morality and mundane matters of fact and capacities that enable two-way control over actions in the way specified by leeway libertarians. It is beyond the control of human beings in their pre-mortem lives to gain these capacities in the first place, if they do gain them at all (Hartman (2017), 98). So, there is no problem for its being beyond the control of damned persons whether they regain these capacities by God's intervention. Thus, the second objection is defused.

Third, perhaps acting from non-freely formed mental dispositions does not itself undermine freedom, but it does when their source is an external manipulator and when they are at odds with the person's previous character traits. This third objection relies on the Character Claim*:

(CC*): S performs free action A only if A is not motivated and explained by mental dispositions that (i) S has not freely formed, (ii) are caused by non-consensual manipulation, and (iii) are at odds with S's previous mental dispositions (cf. Mele (2019), 127–128).

The CC* does not imply that free action is metaphysically impossible. For although it is necessarily true that a person's mental dispositions that motivate and explain her first free action have not previously been formed by her own free choices, it is possible that those mental dispositions are not subject to non-consensual manipulation or are not subject to it in a way that is at odds with her previous mental dispositions. But if the CC* is true, damned persons cannot freely repent in the way that I have argued, because the choice to repent is motivated and explained by character and reasons that (i) they have not freely formed, (ii) are subject to non-consensual manipulation by God, and (iii) are at odds with their previous character and reasons.

The CC*, however, should be rejected for a reason forcefully argued by Taylor Cyr (2020, 2387–2391) and others (see Kane (2007), 174–175; Lemos (2018), 34–35; Hartman (forthcoming)), which I apply directly to the case of damned persons. Suppose that a youth can perform her first free action from character and reasons not freely formed if she has relevant knowledge and leeway control. The same seems to be true of damned persons who have recently had their character altered by God, because there does not seem to be a freedom-relevant difference between the state of the mental dispositions of the youth and the damned person. The youth and the damned person both satisfy (i) in the CC*. So, if there is a freedom-relevant difference between them, it must show up in conditions (ii) and (iii).

Consider why (ii) does not make a freedom-relevant difference. The youth's and damned person's character traits and motivating reasons are both from sources beyond their control and concerning which they gave no consent. The source of the youth's character and reasons is in their parent's genes and habituation as well as the community's moral education, and the source of the damned person's altered character is the agency of God, or perhaps indirectly through trauma, the damned community, or some other indirect mechanism set up by God. These differences – outside of the control of both agents – themselves do not seem relevant to justify the moral judgement that the youth can act freely but the damned person cannot.

Consider why (iii) also does not make a freedom-relevant difference. Plausibly, it is no impediment to free action that a person has recently gone through a major change in their mental dispositions such as a religious conversion, becoming a parent, or having sudden hormonal changes (Arpaly (2006), 112–113); the same is true for radical disorientation due to having lost a loved one or having become disabled in a sudden way.¹¹ The mere fact that the damned person has recently acquired new character traits that are at odds with old traits does not seem relevant to justifying that the youth can act freely and that the damned person cannot. The upshot is that there is no freedom-relevant difference between the youth and the damned person concerning CC* that justifies the judgement that the youth can act freely but the damned person cannot.

So, since the youth can act freely to some degree and there is no freedom-relevant difference between them with respect to acting freely to some degree, damned persons can act freely to some degree. Thus, the CC* is false; the third objection is defeated.

I do not foresee other plausible objections to the claim that God's manipulating or suppressing the character traits of damned persons can re-enable their free choice to accept God's salvation.¹² I conclude that God can make character alterations in damned persons to reopen the possibility of their freely choosing to repent; God can also put people in circumstances that reliably lead to character alterations that reopen the possibility of their free choice to repent. In the next section, I consider the reason why God would do so and why perfect love permits such diminutions of human autonomy.

Human freedom and what God would do

It is in-character for God to create an ability for human persons to repent freely. In the pre-mortem lives of human persons, God directly or indirectly works to alter or mask their character to open the possibility of their free repentance in the first place. It is Christian doctrine that apart from divine help, human persons cannot freely choose to accept God's invitation to salvation or even freely cease resisting it (Pawl (2017), 531–535; see also Romans 3:1–23). But God's prevenient grace alters or masks human character traits to expand the option range of human freedom by enabling them to see sufficient reasons to accept, or stop resisting, God's offer of salvation; God thereby enables human beings to repent freely (Buschart (2006), 190–193). If God works on human character in this way in our pre-mortem lives and there is no change in God with respect to God's love for human persons after they die, we should expect God to work in a similar way in the post-mortem lives of damned persons. Thus, if damned persons become wholeheartedly vicious in a way that volitionally necessitates their rejecting God's offer of salvation, God would eventually intervene to re-enable their freedom of choice to repent.

One might object that there is a relevant difference between pre-mortem and damned conditions of human persons such that God would enable free repentance in the pre-mortem condition but not in the damned condition. For example, God's opening the possibility of repentance in pre-mortem human persons does not reverse a condition for which they themselves are morally responsible. It is through no fault of their own that pre-mortem human persons are unable to enter freely into a relationship with God; it is a function of the way in which original sin has blamelessly bent their character (see Swinburne (1989), 137–147). In contrast, damned persons have freely chosen their condition with respect to which repentance is psychologically impossible, or they have at least made free choices with this being a foreseeable consequence. As such, they are blameworthy for being in that condition. In sum, God intervenes in the cases in which a person is not blameworthy for her inability to repent freely but not in cases in which a person is blameworthy for her inability to repent freely.

This difference in blameworthiness does not justify that conclusion. Human persons in both the pre-mortem and damned conditions can be morally responsible for their inability to make the free choice to repent. A person can become morally responsible for an antecedently possessed mental disposition if the mental disposition is foreseeably maintained via free choices such that had she made different free choices, the mental disposition would foreseeably have been weakened or eliminated (Audi (1991), 308; Hartman (2020), 1422). So, even if a pre-mortem human person is not initially morally responsible for her inability to accept God's invitation to salvation, she can become morally responsible for it to some extent if she has opportunities to make free choices that would foreseeably remove that inability, but she instead made free choices with the foreseeable result that the inability is maintained. And if a person in her pre-mortem life becomes morally responsible to some extent for her inability to accept God's invitation to salvation as a foreseeable consequence of her rejecting God's offer of salvation, she can become blameworthy for her inability. But God does not thereby stop seeking to save her. Rather, God continues to offer the prevenient grace necessary to enable her to accept freely God's salvific grace. Plausibly, the same is true of post-mortem persons in hell.

In sum, it is in-character for God to alter or mask some of the character traits of damned persons to re-enable the free choice to repent by giving them new reasons to do so; God already does so in the case of pre-mortem human persons.

I anticipate, however, that proponents of the Free Will Explanation will object that God's doing so does not appropriately respect the autonomy of damned persons (cf. Murray (1999), 63–65). It is obvious that human autonomy is valuable and should be respected. But promoting well-being is also important. Sometimes enhancing the well-being of others makes it permissible to reduce their autonomy to some degree. For example, if a physically healthy young adult wants to kill himself years after his girlfriend broke up with him, it is permissible to prevent him from doing so. Thus, there are clear cases in which it is at least permissible to curtail the autonomy of others to promote their well-being.

The well-being stakes for damned persons could not be higher, because accepting the invitation to salvation amounts to a reversal from eternal languishing to eternal flourishing. Partly in view of these high stakes, I argue that it is permissible for God to make the character interventions previously mentioned, because God can do so while respecting human autonomy to a sufficient extent for the following three reasons.

First, God would not constantly intervene in the character of damned persons. God may leave damned people to themselves, and perhaps the other denizens of hell, for years at a time between interventions; during such times, their full-fledged moral vices would be operational and so would be the unique suffering concomitant with their exercise. These experiences may provide important data for future free choices when God reopens the possibility of their repentance (cf. Stump (1988), 419), because the memory of horrendous suffering can be an additional motivation to embrace opportunities to live in a new way.

Second, each divine intervention diminishes the autonomy of damned persons only to a small extent. That is, each intervention leaves intact most of their character, projects, goals, and motivations that are theirs in some important sense and for which they are morally responsible. As a result, if a damned person freely accepts God's salvation due in part to God's intervention that reopens that possibility, the free choice to accept must be out-of-character in the sense that either they are acting on reasons that do not issue from their character at all or they are acting on reasons that issue from weaker and less integrated character traits (cf. Hartman and Matheson (forthcoming)).

Third, damned persons retain autonomy over being in heaven or hell. The interventions provide damned persons with the free choice to accept or to reject God's salvific offer, and so damned persons can still freely choose to reject the offer. Possibly, they always freely reject that offer. That is, as many times as God reopens the ability of damned persons to repent freely, it is possible that damned persons make free choices to reject it and effectually reharden their own hearts. Damned persons retain autonomy to some degree over their choice between heaven and hell. If they stay in hell only for a limited period of time, they had autonomy concerning how long they stayed.

I conclude that God would make the character interventions to reopen the possibility of free repentance in damned persons due to their autonomy being respected to a sufficient extent and due to the stakes involved for their well-being.

Consider some objections. If it is permissible for God to make such character interventions, it is permissible for governments and private persons to do so too. But it is impermissible for governments and private persons to make such interventions. Thus, it is impermissible for God to make such interventions.

But this conditional is false at least due to differences in the interveners. God, unlike governments and private persons, is perfectly good, omniscient, and omnipotent. As such, God acts with good ends and does not err in carrying them out, whereas governments and private persons often do not act with good ends; and when they do, they often err in carrying them out.

Another objection is that my argument proves too much. In view of the high stakes, God could intervene with greater character alterations or suppressions to render probable, even in-character, the free choice of damned persons to accept God's offer. So, if my argument is a good one, God should change or suppress their character to a much greater degree than I have outlined, even if God should not causally determine repentance.

Nevertheless, my argument does not prove so much. Depending on which account of personal identity is correct, the person may not survive the more dramatic character alterations suggested by the imagined objector, and God aims to achieve the salvation of the persons who exist. Furthermore, respect for human autonomy is a great good. Such great intrusions into human autonomy do not seem morally permissible to me, but, admittedly, I do not have a moral principle that highlights the relevant difference between the great and small intervention cases; I rest on my intuitions here. For these reasons, I think God is justified in making only the small intrusions that reopen the ability to repent freely.

Probably, each damned person will freely repent eventually

Given the view of hell and divine love from the Free Will Explanation and the fact that God can and would make character interventions in the way previously described, it is very probable that each damned person will be saved eventually. Hell does not end. Even if it is only once every few years that damned persons have the opportunity to accept freely God's offer of salvation, they would still have a potentially infinite number of opportunities to repent freely. When damned persons do have an opportunity to repent freely, they have at least some reasons that favour accepting God's offer of salvation. Those reasons ground a non-zero objective probability of those free choices.¹³ Since there are potentially limitless chances to accept God's invitation to salvation and there is a non-negligible probability of doing so on each such occasion, it is highly probable that each damned person will eventually freely accept God's invitation to salvation (cf. Kronen and Reitan (2011), 160–162; Rasmussen (2017)).

Additionally, God is maximally resourceful in arranging circumstances in ways that non-coercively increase the likelihood that damned persons freely choose to accept the offer of salvation on discrete occasions. To enliven our imaginations about how God might do this, consider some empirical observations about pre-mortem human character.

Social psychologists have been running experiments since at least the 1960s in which they show ways in which human moral behaviour is deeply influenced by mundane features of the situation. According to these experiments, helping behaviour is highly influenceable – that is, sometimes people help and sometimes they do not help – by circumstances that differ merely by smelling cookies (Baron (1997)), being in a hurry (Darley and Batson (1973)), exiting a restroom (Cann and Blackwelder (1984)), or being around bystanders (Latané and Darley (1970)). People are much more likely to help after they smell cookies or exit a restroom, and people are much less likely to help if they are in a hurry or around bystanders. Christian Miller (2013, 57–152) explains that smelling cookies, for example, puts people in a good mood, and their desire to extend that good mood greatly increases the likelihood that they choose to help. But being around other people increases the potential for embarrassment in case the person who appears to need help does not in fact need help, and the fear of embarrassment decreases the likelihood that they choose to help. These experimental results, and many others, show that pre-mortem human moral behaviour is significantly influenced by mundane circumstantial features that are often thought to be irrelevant to human moral behaviour.

The character of damned persons is far more integrated around their dominant moral vices than pre-mortem persons in ways that would make them less susceptible to these circumstantial influences. Still, it seems plausible that these ubiquitous mental dispositions concerning mood enhancement or embarrassment avoidance could still function to influence the free decision to accept or reject God's offer of salvation, especially when damned persons are not wholehearted in their moral vices due to God's interventions. If I am right about this, God could use these facts, and others that perhaps only God knows, to increase the likelihood that damned persons make the free choice for their eternal good.

Behavioural economists have shown other ways in which circumstantial features of choice presentation, or the choice architecture, can non-coercively nudge a person to choose one option over another. A ‘nudge’ is any part of a circumstance that ‘alters people's behaviour in a predictable way without forbidding any options or significantly changing their economic incentives’, and ‘the intervention must be easy and cheap to avoid’ (Thaler and Sunstein (2008), 6). For example, people applying for a driving licence must choose whether or not to be an organ donor. The percentage of donors greatly increases when being an organ donor is the default with an opportunity to opt-out, in comparison with the default as not being an organ donor with an opportunity to opt-in (Thaler and Sunstein (2008), 177–184). The choices are the same, but the way in which the choice is presented has a great impact on moral behaviour. Consider another example. Patients have a choice about whether to have an operation, and they are given the same basic information framed in terms of survival or death: (i) 90 per cent of patients who have the operation survive or (ii) 10 per cent of patients who have the operation die. Patients who are told (i) rather than (ii) are much more likely to have the operation, but it is the same choice merely framed in a different way (Thaler and Sunstein (2008), 36–37). What these, and other, results show is that human persons do not make their choices based solely on the reasons that there are. Rather, what reasons occur to us and how strongly they are weighted is also influenced by features of the situation, its choice architecture. God, or perhaps God's missionaries as imagined in C. S. Lewis's The Great Divorce, could use such knowledge to present the invitation to salvation at times and in ways that make its acceptance more probable.¹⁴

In sum, given the Free Will Explanation's view of hell and divine love, it is highly probable that each person will be saved eventually, because there is a potential infinity of opportunities for free repentance, a real chance on such occasions for free repentance, and that chance can be bolstered by the right choice architecture.

Does it follow, from the high probability that each person in hell will be saved eventually, that it is highly probable that all people in hell will be saved eventually? It depends on the probabilities and number of people. For example, suppose it is 0.99 probable that each person will be saved eventually. If there are 100 people in hell, the probability that all will be saved eventually is ~0.366, which is calculated by (0.99)^100. And if there are instead 500 people in hell, the probability for universal salvation plummets to ~0.0065; the probability further drops to ~0.0001 if there are 900 people in hell. So, if hell is densely populated, the only way for it to be very probable that all people will be saved eventually is if the probabilities for individual salvation are so high that they eventually approach 1.0. But instead of speculating exactly how probable it is that each person will be saved eventually, I highlight that I do not need to claim that universal salvation is probable to reveal an internal problem in the Free Will Explanation. That it is highly probable that at least some people freely repent in hell and thereby escape to heaven suffices to refute the Free Will Explanation's ambition to provide a philosophical defence of the finality of damnation, which is the target of this article.

Consider eight objections to the argument. First, God does not want reconciliation with damned persons in this way, because damned persons are choosing for reasons that are ultimately not their own.

Nevertheless, all people who accept God's invitation to salvation choose for reasons of which they are not the ultimate source. To avoid the Pelagian heresy that human persons can be fully acceptable to God on their own steam, God's prevenient grace must be the ultimate source of those reasons (Cross (2005)); if this prevenient grace is not a problem in the pre-mortem case, it is unproblematic in the post-mortem case.

Second, God's nudging damned persons by manipulating their choice architecture undermines their freedom of choice.

God's setting up the choice architecture does not undermine human freedom for three reasons. First, a nudge is not a push; it does not apply coercive force or even constraining force. Second, the objection would prove too much, because divine providence sets up the choice architecture for all human choices. Thus, the Free Will Explanation is dialectically committed to the claim that God's orchestrating the circumstantial influences does not itself undermine freedom of choice. Third, there are good reasons for thinking that circumstantial luck in general does not undermine freedom, even though I cannot restate the arguments here (see Hartman (2017)). This provides a reason to think that nudging in general does not undermine freedom of choice, because, in both cases, the choice is influenced by circumstantial factors outside of the agent's control – and the mere difference that nudging involves the agency of others does not make a freedom-relevant difference (see Todd (2019), 1348–1349).

Third, a perfectly loving God would not intervene in these ways, because love requires vulnerability to rejection (Lewis (1960), 121); God's interventions rule out that vulnerability.

This objection fails for two reasons. First, love does not require vulnerability to rejection. The persons of the Trinity love one another without vulnerability to rejection.¹⁵ Second, God is vulnerable to rejection from sinful human persons. Many human persons freely reject God's invitation to salvation in their pre-mortem lives, and, probably, human persons in hell freely reject God's salvific offer many times even after being subject to the interventions. Possibly, damned persons freely reject God forever.

Fourth, God's perfect love does not require providing a potentially infinite number of opportunities for repentance, because perfect love for human beings does not require unrestricted pursuit of the good for humans (see Wessling (2020), 33–36; 149–163). Since God is the most valuable being, God's self-love motivates God to pursue God's own projects, which imposes limits on God's attention to pursuing the good for human beings.

As an omnipotent being, however, God can easily provide the potentially infinite opportunities for human repentance, presumably, at no detriment to God's own projects, especially if God sets up the environment of hell to create fissures in moral vices from time to time.

Fifth, limitless opportunities for free repentance do not make it probable that everyone will freely repent eventually, because the longer a damned person persists in rejecting God, the harder it is for her to repent freely (see Walls (2004), 227).

This objection presupposes that damned character cumulatively changes over time until it becomes an unalterable volitional necessity to reject God. The previous sections highlight why that presupposition is false. God's character interventions can alter their moral vices to enable damned persons to see new reasons in favour of their freely repenting. Thus, it is not the case that the longer a person is in hell, the less likely it is that they will leave of their own free will.

Sixth, it is impossible for damned persons to repent and escape from hell, because hell is unalterable due to its being an eternally static state.

This is a new reason for the inescapability of hell. The objection, thus, requires that the Free Will Explanation be supplemented with a reason to think that hell is an eternally static state. But the Free Will Explanation provides a reason to reject that idea. If God sets up hell as an eternally static state when it could have been a dynamic state, the ultimate reason why damned persons cannot be reconciled post-mortem to God would be located in the divine will rather than human free will, which is contrary to the Free Will Explanation.

Seventh, damned persons in hell cannot repent and ascend to heaven. For if damned persons can repent and ascend to heaven, then redeemed persons in heaven can apostatize and descend to hell, but redeemed persons cannot apostatize and descend to hell.

The conditional claim is false due to asymmetries in the character and circumstances of redeemed and damned persons. The moral virtues possessed by redeemed persons cannot be lost. Redeemed persons cannot freely choose to lose their moral virtues, because they are volitionally necessitated to choose the good, which requires their freely preserving moral virtue in heavenly circumstances (Hartman (2021)).¹⁶ Additionally, the exercise of moral virtues is intrinsically pleasant, external heavenly circumstances only support virtuous choices and character, and God only supports virtuous choices and character. Thus, it is impossible for redeemed persons to see reasons to apostatize.¹⁷ In contrast, damned persons do not always have volitional necessities to reject God's invitation to salvation, because their vicious character can be upset. And there is suffering concomitant with the exercise of the vices themselves. Even if damned persons cannot see in themselves the vicious cause of the attendant suffering, they may be able to see it more clearly in the similar vices of others and their attendant suffering (see 2 Samuel 12:1–15), which may provide a reason for them to rethink their own character and projects. Traumatic external circumstances that exacerbate suffering from their moral vices may provide yet another source of uneasiness that leads the damned person to rethink their cares and commitments. God may also make direct or indirect interventions to alter or suppress wholeheartedly vicious character on various occasions. But then, damned persons at least sometimes see reasons to repent, and this is what enables a free choice to repent.

Eighth, it is not possible for damned persons to be redeemed, because if it is possible for damned persons to be redeemed, redeemed persons in heaven cannot rest until damned persons are saved; but redeemed persons in heaven can rest.

But the conditional is false due to the perfected character of the redeemed in heaven. The redeemed do care about damned persons, and they do pray for them. But since their character has been perfected, they exercise perfect trust in God's working for the redemption of damned persons, and so they will be at peace and can rest (cf. Isaiah 26:3).

Conclusion

The Free Will Explanation aims to provide a plausible philosophical explanation of the compatibility of divine love and power with a traditional view of hell. But I argued that the Free Will Explanation's description of hell and divine love carries the seeds for its own rejection. First, the moral vice silencing mechanism in the Free Will Explanation cannot plausibly explain why no-one leaves hell. God can make interventions in human character to reopen the possibility of their freely choosing to repent. Second, God would do so, because it is in-character and it respects human autonomy to a sufficient extent. Third, given a potential infinity of opportunities to repent freely and God's controlling the choice architecture, it is very probable that each damned person freely chooses to accept God's offer of salvation eventually, which is not necessarily to say that there is a high probability for universal salvation.

Suppose that my arguments succeed. That is, suppose I have established the conditional thesis that if the models of hell and divine love in the Free Will Explanation are correct, there is a high probability that each damned person will freely repent eventually. What else we should conclude about hell remains an open question based on whether we should apply modus ponens or modus tollens to that conditional. On the one hand, take it as a datum that it is false that there is a high probability that each damned person will freely repent eventually. In view of that datum and truth of my conditional thesis, it follows by modus tollens that either the model of hell or divine love in the Free Will Explanation is incorrect. On the other hand, take it as a datum that the description of hell and divine love in the Free Will Explanation are correct. In view of that datum and truth of my conditional thesis, it follows by modus ponens that there is a high probability that each damned person will freely repent eventually. But which datum is more plausible, and so whether to apply modus tollens or modus ponens to my conditional thesis, is an important question that outstrips the ambition of this article.