The problem of mourning

In The Image of God, then, Stump’s stated aim is not to formulate yet another theodicy but rather to address the problem of mourning, which she takes to be distinct from the traditional problem of evil: God might be morally justified in using suffering to restore post-fallen lives, but what are we to do with the lingering sense of loss and damage that remains even as lives are made beautiful again? To answer this problem, Stump sets out to show that a world with suffering – or, as she puts it, a world with a fall – is ‘more glorious’ and ‘more worth celebrating’ than a world without suffering (Stump Reference Stump2022, x, 15). In this sense, Stump constructs what she characterises as a felix culpa defence (Stump Reference Stump2022, 10–16): it is better that the fall should have happened rather than not, because the good that emerges from suffering lies in not merely what follows from it but also ‘the very life with that suffering,’ in the transformed self that such a life makes possible (Stump Reference Stump2022, 242).

Hence, what makes the problem of mourning somewhat distinct from the more traditional theodical problem is the angle it takes. The problem of mourning is not a problem that concerns divine morally justifiable reasons as such. Rather, it is an axiological problem – it concerns the kinds of value appraisals we can make concerning God’s creation and, by extension, the kinds of responses that are appropriate with respect to it. At the beginning of the book Stump articulates this axiological problem in global terms, but in actuality Stump finds the exercise of comparing whole worlds ‘unpromising’ and therefore chooses to transpose the question ‘to a smaller and more tractable question about individual human lives’ (Stump Reference Stump2022, 298). More specifically, she argues that the goodness of lives should be measured in terms of the gloriousness of a person’s self (Stump Reference Stump2022, 259). Her aim, then, is to demonstrate, through a close reading of the passion narratives in the Gospels, that every human life is ‘not a disappointing or even distressing version of what it might have been in a world without a Fall’ because only in a fallen world human beings can gain their most glorious selves – and so lives that contain suffering are more glorious than their suffering-free alternatives. Stump therefore sets herself the following standard of success when it comes to her response to the problem of mourning:

Stump’s solution

Stump’s solution to the problem of mourning hinges on a particular view of selfhood and its perfection. Her central claim is that a person’s most glorious self might only become available in a world with suffering. Stump builds a complex argument for this claim in distinctly theological terms. She begins by suggesting that the true self of a person is ‘the emergent condition that results from the interweaving of his personal thriving with the fulfilment of his heart’s desires when they converge’ (Stump Reference Stump2022, 51). This emergent view of flourishing is already familiar from Stump’s earlier works on theodicy – Wandering in Darkness (Reference Stump2010) and Atonement (Reference Stump2018) – but in The Image of God she links it specifically with Christ’s passion. First, she argues that the flourishing of any existing thing is co-extensive with the fulfilment of its nature. Hence, for human beings flourishing is a function of their ‘species-specific capacity’ for ‘the intellect’s vision of the essence of the triune God and the will’s union of love that is shared with the persons of the Trinity’ (Stump Reference Stump2022, 261). Put otherwise, human flourishing consists of having God as the deepest desire of one’s heart and this desire then leading one to a reciprocal union of love with God (Stump Reference Stump2022, 262). When a person attains this state of flourishing, they attain their true self.

Now, what makes a person more or less glorious in her true self is ‘the intensity of God’s image’ in that person (Stump Reference Stump2022, 261). More specifically, since God is Trinitarian love, it is the depth of love that makes a person reflect God’s image more in their true self; unions of love admit of degree, and the deeper a person’s union of love with God is, the more intensely God’s image is magnified in them. This builds on the earlier work of Stump’s, but here it is not merely attaining a state of flourishing – or a true self – that Stump is interested in. Rather, she is interested in how persons can attain maximal flourishing, their most glorious self.

This takes us to suffering. For ‘the most compelling manifestation of God as love is in Christ’s passion’ (Stump Reference Stump2022, 262). It follows, then, that reflecting the ‘complex love of God as manifest in Christ’s passion is the perfection of the true self of a human person’ (Stump Reference Stump2022, 262). Which is to say that the deepest kind of love is suffering love and therefore, since attaining our most glorious and God-like selves is a matter of love, the perfection of a person’s true self can only be attained in a world where Christ suffers and where human beings can, through their own suffering, be joined with Christ’s passion, thereby embodying God’s love in their selves. Importantly, then, reflecting ‘the complex love of God as manifested in Christ’s passion’ is not merely a matter of, say, deeply appreciating the love Christ displays on the cross and then taking that love as an ethical directive or having one’s appreciation of this love fuel one’s devotion to God. Instead, reflecting God’s own suffering love requires an identification with that love through suffering: the deepest unions of love are unions of mutual, shared suffering with Christ that draw human beings into ‘the mutual indwelling of love among the persons of the Trinity, but in a way not available to them in a world without a Fall; it magnifies the image of God in them more intensely than it could have been otherwise’ (Stump Reference Stump2022, 297).Footnote ²

It is perhaps already evident how this might help Stump solve the problem of mourning: for if the deepest kinds of unions of love – and hence our most glorious selves in which the image of God is most magnified – require participation in Christ’s passion, then they are only available in a world where there is suffering. Some true selves and forms of flourishing might, of course, be available also in a suffering-free world, but most glorious selves and highest forms of flourishing might not be. Thus, if our world is sufficiently like the narrative world of Stump’s theodicy, we have no reason to mourn its being the way it is: for without its attendant evils, this world would not be a place where our most glorious selves can emerge.

The problem of valorisation

Stump’s solution to the problem of mourning is multi-layered and elegant. It also exhibits Stump’s capacity for creatively employing the distinct resources of Christian theological traditions in service of her argument. Unfortunately, however, Stump’s solution to her problem generates another, rather severe problem because it is premised on a particular understanding of love, namely one in which there is a very close connection between love and suffering. Indeed, in The Image of God love is most richly realised in suffering because love always involves vulnerability with respect to the other, and this vulnerability is most intensified – or, as it were, stretched to its maximal expression – in suffering for the other’s sake (Stump Reference Stump2022, 108–110). Now, we might wish to agree with Stump (and Scriptures; see Jh. 15:13) that love’s richest expression is found in laying down one’s life for the sake of the other, but – as I shall soon demonstrate – for Stump this is pushed further: love is so strongly identified with a certain kind of suffering that this suffering becomes reason-giving in itself. In this section I argue that Stump’s proposal therefore risks granting at least some suffering intrinsic value. I’ll call this – granting at least some type of suffering intrinsic value – valorisation.

In this paper I’m going to assume that valorisation is a problem, because it undermines our basic judgement that suffering is a bad thing. It is important that we are able to judge suffering as a bad thing because the judgement concerning its badness informs our ethical responses with respect to suffering.Footnote ³ Stump agrees with this: she wants to say that there is ‘nothing good about suffering by itself; by itself suffering is truly worth lamenting’ (Stump Reference Stump2022, 299). That suffering is bad by itself suggests that suffering is something that no rational agent would want or desire as such. I take it that this, too, is something Stump would agree with since she thinks that the will is always directed towards a perceived good, whether real or apparent (Stump Reference Stump2022, 251).

What Stump wants to say is that even though suffering is not something any rational agent could prefer as such, rational agents might come to prefer a life with suffering to a life without because the life with suffering involves greater values on the whole. Lives, after all, are complex axiological entities: they are value aggregates that form more or less rich wholes. Hence, it is possible that Life₁ that entails suffering might be better than another, suffering-free Life₂ because in Life₁ suffering allows certain higher values to emerge that would not be available in Life₂.Footnote ⁴ In preferring Life₁ over Life₂, then, the agent does not desire suffering as such but rather the axiological whole that is Life₁.

This, I take it, is more or less the structure of Stump’s response to the problem of valorisation: what is valued is not suffering per se but the ‘gloriousness in the lives of those who suffer’ (Stump Reference Stump2022, 15), the ‘very life with the suffering’ (Stump Reference Stump2022, 242). Another way to put this would be to say that suffering has no intrinsic value, but merely instrumental value. Whatever has intrinsic value is reason-giving in itself: it gives us at least a justifying reason to pursue it for its own sake. Suffering, according to Stump, is not like this: we have no reason to pursue it for its own sake but only for the sake of something else. And so, in various places she characterises the role of suffering as instrumental in bringing about the person’s most glorious self (Stump Reference Stump2022, 244, 255).

However, if suffering is merely instrumental in bringing about some person S’s most glorious self, then S’s most glorious self – the goodness of her final state – is in principle conceivable apart from suffering. We could imagine, for example, that S’s suffering is required to bring her to a deeper union with God but once this state is reached S – by some divine miracle – forgets her past sufferings: they no longer play any role in the gloriousness of her final state. This, however, is not something Stump wants to say: she wants to say that some of the glorious-making value of S’s best life is instantiated in suffering, because it is a type of suffering – namely, sharing in Christ’s passion. Wounds and scars are therefore ‘part of the glorification of the true self of a person’; they are an ingredient of its very gloriousness (Stump Reference Stump2022, 21). Thus, they do not simply make available S’s most glorious self: they are constitutive of it.Footnote ⁵

Hence, even though Stump explicitly states that suffering has an instrumental role in bringing about S’s most glorious self, her account in fact assigns some suffering (suffering-in-union-with-Christ) a more than instrumental role, thus tacitly granting it intrinsic value so that this specific type of suffering becomes something worth pursuing as such. I think Stump is pushed into making this move because if suffering had a purely instrumental role in bringing about S’s most glorious self, then that self would be conceivable apart from suffering. This would leave open the possibility of regretting the fact that S’s most glorious self had to be realised through mournable means – that is, that there isn’t a world available where S can attain her most glorious self without suffering. To close this grief-producing loophole, Stump must join the glorious-making value of S’s life with a certain type of suffering so that this suffering and the version of S’s self that most magnifies divine love become one and the same thing. And so, since the glorious-making value of S’s life is to be located in the very suffering S endures and not simply in its results, S has no reason to regret the presence of suffering in her life. This, I take it, is why Stump states that the maximal value of S’s life is found not simply in the results of S’s suffering but in the ‘very life with the suffering’ (Stump Reference Stump2022, 242).

By joining together suffering and the glorious-making value of S’s life, Stump’s view entails that a certain type of suffering is constitutive of an intrinsic value: S’s most glorious self that most magnifies divine love is constituted in S’s suffering-in-union-with-Christ.Footnote ⁶ The relevant glorious-making value, in other words, does not merely entail a type of suffering but is a type of suffering. At this point it might be useful to make some distinctions. Saying that suffering-in-love constitutes the glorious-making value of S’s life is different from merely saying that suffering is a part of a valuable whole – that is, suffering contributes to the value of the whole but does not constitute it. For example, suffering could make available to S a deeper union of love with Christ or it could allow S to manifest God’s love more. These are both claims that would amount to making suffering a part of an intrinsically valuable whole (namely, S’s union with Christ) without suffering itself being constitutive of the whole’s axiological status.

These – suffering making available a deeper union with Christ and so on – are ways of saying what might be done with some suffering that befalls S, how it could be redeemed. Stump, on the other hand, is not interested in simply explaining how suffering can be redeemed or defeated post facto because that does not answer the problem of mourning; even if suffering would be redeemed post facto, we could still mourn the fact that it took place in the first place. That’s why Stump needs to make a stronger claim: there is a type of suffering that is constitutive of the final axiological status of S’s life and not simply tolerated as an unfortunate accompaniment of it. It follows that there is a type of suffering that acquires pro tanto reason-givingness because it is constitutive of a glorious-making value. And this is indeed precisely Stump’s point: the value of suffering-in-love that most maximally glorifies God’s image in S gives God a reason to realise a world in which the fall takes place and where S can join Christ’s suffering.

We can express these distinctions more formally in the following way:

Value Constitution, then, differs from Value Contribution in that the final axiological status of L is fully determined by some individual value-component G. Thus, G does not merely contribute to P – whether instrumentally or otherwise – but constitutes it. If we take L to be some person S’s life, then G is appropriately called ‘the highest good’ of S.Footnote ⁷ The highest good, on the other hand, is an intrinsic value – it is reason-giving in itself. The issue is, then, that in Stump’s story this highest good is a type of suffering: it is not just that union-with-Christ renders S’s life glorious but a union of suffering renders S’s life glorious. Again, Stump is pushed towards this latter view because, in order to block mourning, she needs to give suffering a more than instrumental role in the final and highest value of S’s life.

The issue at hand, then, is not whether suffering can be redeemed or defeated, but whether Stump’s account implies that there is suffering that is itself constitutive of value in such a way that it becomes reason-giving as such. Stump might respond, of course, by reiterating that suffering considered by itself – apart from any redemptive Christological framework – is ‘truly worth lamenting’ and does not constitute a value (Stump Reference Stump2022, 299). However, this does not allow Stump to avoid the problem of valorisation. For what is sufficient to produce the problem of valorisation is saying that there is some suffering that has intrinsic value. What’s more, Stump’s account implies that suffering constitutes the highest value of S’s life, because only by joining Christ’s passion can S enjoy a maximally deep union of love with God. Therefore, it seems that in Stump’s story there is suffering that a rational agent would want or desire as an end in itself – suffering that has intrinsic value.Footnote ⁸

Is there a solution to the problem of mourning?

In my view, risking granting any suffering intrinsic value is too high a price to pay for a solution to the problem of mourning. Now, while it may be, as Vicens (Reference Vicens2025) argues, that the problem of mourning is something we can live with, we might still wonder if there’s a solution available to the problem that does not entail valorisation. Recall that the problem of mourning does not concern the question whether suffering simpliciter – considered apart from any redemptive framework – is lamentable. Stump agrees that it is. Rather, the question is whether lives that contain suffering are worth mourning, mere ‘sad and distressing’ variations of what they could’ve been.

In order to answer this question in the negative, Stump embarks on a quest to effectively show how some suffering can constitute the highest value of some person S’s life. We can therefore identify two distinct premises that Stump’s answer to the problem of mourning involves: (a) that there is suffering that constitutes a value and (b) that the value suffering constitutes is a maximal value that renders Life₁ better than any suffering-free alternative Life₂. These two claims are linked, but previously our main focus was on (a). In this section we’ll focus on (b): I argue that by rejecting (b) we can, in fact, approach a solution to the problem of mourning that avoids the pitfalls of valorisation.

The maximisation claim, then, presupposes that Life₁ and Life₂ are commensurate axiological wholes: one can be better or worse than the other. This, however, is a notion that can be rejected, because we could always say that Life₁ and Life₂ are simply incommensurate: they cannot be placed on the same scale. Thus, Life₁ and Life₂ might just be different in such a way that there is no coherent answer to the question whether one is better than the other.Footnote ⁹ We thus find that there are three options for how Life₁ and Life₂ could stand in relation to each other axiologically: (i) they could be commensurate with each other and equally good; (ii) they could be commensurate with each other and one could be considered better and/or worse than the other; (iii) they could be incommensurate.

With this in mind, we can see that there are, in fact, three axiological situations that would defuse the problem of mourning – something Stump does not seem to note. First, as per (i), Life₁ and Life₂ could be commensurate with each other and equally good. Second, as per a version of (ii) – call it (ii)* – they could be commensurate and Life₁ could be better than Life₂. Third, as per (iii), Life₁ and Life₂ could be incommensurate with each other. Stump is solely focused on (ii)*, that is, demonstrating how Life₁ is better than Life₂. It, of course, seems prima facie implausible that Life₁ that contains suffering could be better than Life₂ that does not contain suffering – which is why Stump spends most of the book demonstrating the glorious-making value of suffering. This then leads to the problem of valorisation. However, paying attention to non-lamentable scenarios other than (ii)*, we might find a different way towards a solution that does not risk valorisation.

I find option (iii) plausible for the following reason. As Marilyn McCord Adams (Reference Adams1999) has proposed, the glorious-making value of any life S could have is ultimately S’s relationship with God and the integration of all of S’s life experiences, including her suffering, around her relationship to God. God, on the other hand, as Adams argues, cannot be placed on a scale with any creaturely value: the relationship between creatures and God is incommensurate because of God’s unique ontological status as God. God is not just a good thing; God is Infinite Goodness itself, superabundantly containing in Godself all creaturely goodnesses. If we take God to be the ultimate and incommensurately great good of any person, then all of S’s best lives contain integration around divine goodness.Footnote ¹⁰

Now, unfortunately we do not have enough space in this context to develop a theological framework around this incommensurability claim that is as robust as Stump’s own theologically construed axiological framework. Recall, however, Stump’s view that the glorious-making value of a person’s life is the self they become through that life. We could connect this Stumpian notion with Adams’s central claim in the following way: in all those lives in which S’s experiences are integrated around divine goodness, S attains one of her best selves. Call all those lives of S in which S’s experiences are ultimately integrated around divine goodness and in which S thereby attains one of her best selves redeemed lives. For every redeemed life, the incommensurate good of God’s own life is constitutive of their value. That is, as per Value Constitution, the axiological status of the whole is determined not simply by the sum total of all its values but rather by the way those values are organised in relation to a single good – that is, the infinite value of God’s own life. This good we can designate ‘the highest good’ of S. This, I take it, is one way to interpret the traditional Christian claim that God is the ultimate and final happiness of human beings. It differs from Stump’s view in that here the highest good is incommensurate rather than maximal and that it is not identified with suffering-in-union-with-Christ but rather with God simpliciter.

Now, the key point here is that since redeemed lives are axiologically constituted in this way, they inherit the incommensurability of their highest good G (i.e., God). Thus, they are also axiologically incommensurate with respect to each other. What gives us inheritance is precisely Value Constitution: redeemed lives do not merely include an incommensurate good – rather, their value is constituted by it. At the same time, individual components of redeemed lives, when considered in isolation from the whole and its value-constituting relations, can be commensurate with alternatives. This allows us to maintain, as Stump also says, that there is nothing good about suffering considered by itself.

What matters for the problem of mourning, however, is what kind of global appraisal is fitting with respect to lives as wholes: since the value of L is constituted by G, it is the presence of G in L that renders positive appraisal (affirmation, satisfaction, gratitude) fitting with respect to L as a whole. This defuses the problem of mourning if we take it to be a global axiological problem that concerns the comparative axiological status of different lives S could have lived and appropriateness of certain responses (grief, disappointment) that follows from our comparing S’s actual life with some other life she could’ve lived.Footnote ¹¹ All of S’s redeemed lives, then, are very good lives for S regardless of whether they involve suffering because in them S’s experiences are integrated around the incommensurate value of divine goodness.Footnote ¹² At the same time, redeemed lives do not all have single, identical value because they remain differently organised axiological wholes: each redeemed life is constituted axiologically by distinct patterns of integration around the same incommensurate good (e.g., they contain different life stories for S, sets of relationships and experiences, etc.). This organisational feature individuates each redeemed life.

That some lives of S are incommensurate does not entail that all of S’s lives are incommensurate. More specifically, although S’s redeemed lives are incommensurate with each other, it does not follow that S’s redeemed and unredeemed lives (i.e., lives that do not contain integration around divine goodness) are on a par. The reason for this is that redeemed status is a global, final-value property constituted by an integration around an incommensurate value – lacking this status is not a deficiency that can be compensated by any finite aggregate of created goods. We can therefore divide the possible lives of S into two tiers: one – call it L_r – containing all of S’s redeemed lives and another – call it L_u – containing all of S’s unredeemed lives. Members of L_r may differ greatly (for instance, in the presence or absence of suffering), but there is no maximally great member of L_r, since maximality would require a within-tier ordering of better and worse. At the same time, any member of L_r is lexically dominant over any and all members of L_u.Footnote ¹³

The preference problem

There is, however, still another dimension of Stump’s problem of mourning that should concern us. At times Stump indicates (e.g., Stump Reference Stump2022, 242–243, 289, 294) that the problem of mourning is also about S’s own wishes, desires, and preferences with respect to her life: S’s life must be such that in the end she does not wish God would’ve done otherwise with her. I agree with Stump (and Adams Reference Adams1999, 143–149, who is the originator of the idea) that divine goodness gives us grounds for thinking that God would be committed to making sure that in the end S is capable of adopting some kind of preferential attitude with respect to her actual life over alternatives at least in the sense that S does not wish that God would’ve done otherwise with her. This, as Adams has argued, is entailed by the fact that God’s goodness is personal: God’s goodness is directed towards persons on an individual level. Hence, theodicies that are interested in defending divine goodness must show how God is not good simply by generic standards but how ‘God is good to each and every created person when their whole career … is taken into account’ (Adams Reference Adams1999, 52). Part of God’s being good to each person is God’s making sure that each person’s life is one that they are ultimately content with. This produces the following problem:

The Theodical Preference Problem

1. 1. If God is wholly good to S, then God ensures that S’s actual life, Life₁, is such that it matches her ultimate preference (i.e., the preference of her perfected self).

2. 2. S’s actual life, Life₁, contains suffering.

3. 3. There is a possible suffering-free life, Life₂, S could’ve had.

4. 4. In her perfected self, S would prefer Life₂.

5. 5. Thus, S’s actual life does not match her ultimate preference.

6. 6. Therefore, as per 1, God is not wholly good to S.

Note, first, that the preference problem concerns preferences S would have under conditions of ideal agency. For Stump, this ideal agency is construed in a particular way as the person’s most perfected self (i.e., the self that most magnifies God’s image). However, we do not necessarily need to accept Stump’s particular construal of ideal agency to get the preference problem off the ground: the point is that under the conditions of ideal agency (however construed) S would prefer some suffering-free alternative to her actual life. In what follows, then, I’ll mostly opt for rationality rather than perfected self as a way of characterising ideal agency.Footnote ¹⁴

For Stump, the preference problem is not distinct from the more global axiological problem (i.e., the problem of mourning), because she assumes that any rational agent would prefer an axiological whole of greater value over one with lesser value and that, more specifically, in her perfected self S would prefer a life that more richly manifests God’s love: rational agents would choose a more glorious self and a richer life over a less glorious self and a less rich life. Thus, if Stump succeeds in showing that Life₁ with suffering is better – and thus more preferable to rational agents – than its suffering-free alternatives, then the problem of preference is solved by adding a subjective condition to the solution presented to the problem of mourning: Life₁ should be better than its suffering-free alternatives and this fact should be something that S will be capable of appreciating in her perfected state.

It might be tempting to think that by adopting the view that redeemed lives are incommensurate we will have solved the preference problem. However, the problem of preference remains as long as we maintain that S can still coherently prefer one incommensurate whole over another. In fact, in some sense the problem is exacerbated: if, pace Stump, redeemed lives with suffering are not better than redeemed lives without, why would anyone prefer a life with suffering to a life without? That is, say S would have a preference over two options: S’s actual life (Life₁) that contains both suffering and the incommensurate good of integration around divine goodness and an alternative life (Life₂) that is free of suffering and contains the incommensurate good of integration around divine goodness. Since these lives are incommensurate – and thus S cannot prefer Life₁ over Life₂ because it is somehow ‘better’ than Life₂ – what reason would S have to prefer her actual life over the suffering-free alternative?

Recall, however, that the incommensurability of Life₁ and Life₂ does not entail that they are equally good axiological wholes. Rather, they are entirely differently great axiological wholes – redeemed lives can contain diverse values that are organised in a variety of ways. That two items are incommensurate means that, all things being equal, a subject can act rationally in choosing either one of the items. In the context of preferences over lives, incommensurability therefore introduces both explanatory and epistemic constraints. First, it introduces a simple explanatory constraint: when it comes to preferences that concern incommensurate values, we cannot explain S’s preference of some value A over B by arguing that A is better than B and therefore S has at least a justifying reason to prefer A. Say, for example, that apples and oranges are incommensurate with each other. This would mean that there’s no way of coherently answering the question whether apples are better than oranges in terms of something being the case with respect to apples and oranges as such. Rather, S’s general preference for apples would have to be explained through complex – and sometimes intractable, even arbitrary – subject-specific factors (S’s personality, life history, context-specific aims, etc.).

Add to this the fact that lives are very complex axiological wholes. Say, for example, that in Life₁ S’s experiences are integrated around divine goodness in a particular way due to some suffering S undergoes in Life₁ – for example, perhaps in Life₁ S’s integration around divine goodness takes on a different kind of depth because of S’s suffering.Footnote ¹⁵ In Life₂, on the other hand, S’s integration around divine goodness does not have a similar dimension of depth. Nevertheless, since the good involved is the incommensurate good of divine goodness, we cannot say that Life₁ is better than Life₂. Rather, they remain differently valuable lives. S can thus rationally prefer either Life₁ or Life₂ – both would be very good lives for S. Thus, S’s being a rational agent – or acting under conditions of ideal agency – is not sufficient to explain S’s preference. Rather, S’s preference might simply be a personal preference – a coherent expression of S’s unique subjectivity that is, as it were, supra-rational in the sense that it is not just a choice that any and all rational persons would make but a choice that somehow makes visible the particular person S is. We recognise that human beings have such supra-rational preferences all the time – indeed, we often value choices based on those preferences more than choices that are merely rational.

This takes us to epistemic constraints. If S’s redeemed lives are incommensurate it becomes much harder even for S to answer the question whether she would prefer her actual life with suffering over a suffering-free alternative. Why? Because the very subject-specific factors that explain S’s preferences are themselves formed by things that are specific to each life (the particular course S’s life takes in Life₁, the particular character of her relationships to significant others in Life₁, and so on), and S might not have sufficiently robust epistemic access to her possible selves and their preferences to know whether she would, in fact, prefer some possible life, Life_2, over her actual life, Life_1, because that preference involves not just her preferences in Life₁ but also, as their reverse, her preferences in Life₂. That is, to say that one prefers Life₂ over Life₁ is to also say that in Life₂ one would rather have Life₂ than Life₁. To clarify this, consider again the apples-and-oranges case: to say that S prefers apples over oranges is to also say that S can reasonably expect to be satisfied with the outcome of her choice of apples over oranges. If, on the other hand, S chose apples over oranges only to begin to prefer oranges over apples, it would not be coherent to say that S actually has a sustained preference for apples over oranges: S simply seems indecisive about what she wants.

Preference requires, in other words, that the agent make the judgement that in the new, preferred but yet to be realised situation she would be satisfied. In the case of apples and oranges, S can indeed know this reasonably well because her preferences in both possible scenarios (getting apples or getting oranges) are a result of the same preference formation process. However, when it comes to S’s knowing what she would prefer if she had had a completely different life, the question becomes quite intractable, given that S’s capacity to assess and imagine her preferences – whether possible or actual – is itself formed by the life she has had.

These explanatory and epistemic constraints undercut the preference problem by depriving its key preference-claim of the needed epistemic support. However, these considerations in fact lead us to an even stronger response – one that will allow us to reject the whole problem as incoherent. To see how, recall that the preference problem is about S’s preference of one life over another. It therefore involves some kind of trans-world preference, that is, a preference that is not dependent on any particular life S might have lived but remains stable throughout possible worlds. Perhaps there are some trans-world preferences: for example, perhaps we could say that in all the worlds where S exists and is a rational agent, S has a preference for her own happiness over her unhappiness. Here, S’s trans-world preference would be dependent on – and explained by – a salient agential feature of S (i.e., rationality). Even here, however, the trans-world preference in question concerns only intra-world options. By contrast, preferences over lives would require a stable trans-world preference ranging not over different options within a single world, but over different possible worlds themselves. The incommensurability of S’s redeemed lives undermines the coherence of such a preference. If Life₁ and Life₂ are incommensurate, then any preference for one over the other must be explained by subject-specific factors that are themselves world-specific, since they are formed by the very life S lives in each world. Thus, S’s choice of Life₁ over Life₂ is not the kind of preference that could meet the criteria for a stable trans-world preference.Footnote ¹⁶

If S’s preferences concerning lives are world-specific, then there simply isn’t a relevant trans-world preference of S that God could honour. That is, there might not be a life that God could give for S that S would always prefer. Instead, God is simply committed to making sure that whichever version of her possible lives in L_r S ends up having, S’s preferences eventually match the life S has – and, if S’s life is one of the incommensurably good, redeemed lives, then S could rationally prefer it regardless of what the alternatives are. Thus, it is possible that S would be equally rational in being satisfied with Life₁ or Life₂ because in both worlds her world-specific preferences match the world she finds herself in. Since the preference problem concerns S’s ultimate preference – what S will prefer in her perfected state when her earthly career is complete – this solution is compatible with S’s current preference being such that she would rather have a different life. What matters for Christian hope is that ultimately – when all of S’s life experiences are integrated around divine goodness – she will be content with the life she had. This is not a way of saying that already now all is well with respect to S’s life; the transformation of S’s life into something that is an overall good for her is a work yet to be done.