A charge against Molinism
According to Molinism, prior to creation, God knows, for any possible agent, exactly what that agent would libertarianly freely do for any set of circumstances in which that agent can be placed. This position is supposed to resolve the apparent inconsistency between divine providence and human freedom. On Molinism, God can micro-manage the history of the world by choosing what circumstances agents will face during their careers, with full knowledge of what they will do in those circumstances. On the other hand, since these counterfactuals are contingent, what agents do in the circumstances they inherit is entirely up to them.
Critics of Molinism have charged that it cannot possibly be true. But others have argued that, if it were true, it would undermine the justice of God's judgement of us. On Molinism, even though God has no control over the truth-value of the counterfactuals of freedom, God can nevertheless control our behaviour in an indirect sort of way by controlling what circumstances we inherit, in full knowledge of what our response will be. We might plausibly suppose that, for any feature of our behaviour F in virtue of which God might be said to judge us, there are agents who actually have F but lack F in some other actualizable world, and there are agents who actually lack F but have F in some other actualizable world. But then it appears that God, in actualizing a world, or in placing agents in the circumstances they actually inherit, is choosing how those agents will ultimately be judged.
Not all Molinists take this apparent consequence of Molinism to be a problem at all, since it coheres well with the scripturally motivated view that God chooses to provide His grace to some and not to others.Footnote 1 Others have addressed this consequence as a problem, but have given dubious replies.Footnote 2 As a result, typically those who hold that there really is something problematic about God's controlling our eternal fate have been drawn away from Molinism and in the direction of an alternative libertarian theistic approach: open theism. On open theism, not only does God have no knowledge of counterfactuals of freedom; God also has no knowledge of how agents will actually freely behave. The upshot is a God who does not micro-manage the history of the world and so cannot control whether agents will be saved or damned. Presumably, then, God is not responsible for anyone's eternal fate – and so, we are to believe, open theism describes a God who is more just than the God of Molinism.Footnote 3
However, I argue that this picture ultimately gets things backwards. Part of the difficulty concerns how the problem gets categorized: it is usually lumped under the problem of evil. That's not wrong, exactly, but it's distracting. The concern isn't merely that God's omnibenevolence is called into question by His determining who is saved and who is damned. The concern is more basic – that the features in virtue of which we may be judged should be so thoroughly subject to facts beyond our control at all, regardless of who or what does in fact control them. The problem with which we are concerned is every bit as much rooted in a problem of soteriological luck. While open theism may escape the soteriological problem of evil, I argue that it is impugned by the problem of soteriological luck. Molinists, by contrast, may offer a very simple solution to both problems.
Open theism and soteriological luck
Assume for a moment that God judges us solely in terms of our moral profiles. As Nagel has famously argued, our common moral evaluation of agents often turns on factors beyond our control. Here are some of the examples included in or inspired by his seminal writing on the topic:
(i) Resultant luck: Two men attempt murder by shooting, but only one attempt succeeds due to the intervention of a passing bird.Footnote 4
(ii) Constitutive luck: One man is an utter moral monster due to rampant abuse suffered as a child, while another, whose parents were model citizens, is a picture of pure virtue.
(iii) Circumstantial luck: A perfectly harmless German emigré to Argentina would have chosen to obey horrible orders from Nazi superiors if his family hadn't left Germany while he was young.Footnote 5
In each case, Nagel claims, our evaluation of agents will hinge on factors beyond their control. He may be right about us. But we may certainly hope that God's evaluation will not follow suit. But then, how might the open theist protect the agent's moral profile from luck?
The response to (i), at least, should be fairly straightforward. The open theist may simply take the standard anti-luck line that the agent's moral profile is determined by her actions, not their consequences – and so, on this line, God evaluates both shooters the same way. The other cases are more vexing. After all, the character and circumstances that an agent inherits will influence her actions, and so if her moral profile is determined by her actions, then it seems that they will influence her moral profile.
In fact, although Nagel distinguished between them, we can easily treat cases like (ii) and (iii) together as exemplifying a single kind of luck – a more general kind of circumstantial luck. As a matter of stipulation, let's say that, for any set of circumstances C, C includes the atom-for-atom arrangement of the world at a time and the laws of nature, insofar as they participate in inclining forces on the agent. Then C will include facts about what morally relevant actions are open to the agent (e.g. whether she has the opportunity to enlist in the Nazi army) and also facts about the agent's constitution (e.g. whether or not she has nasty dispositions). The trouble with luck exemplified in cases (ii) and (iii) is that one's being in C is (often) beyond one's control, but that one's being in C nevertheless appears to play a role in determining one's moral behaviour, and so one's moral profile.
How might the open theist reply to these cases? She might begin, quite naturally, by claiming that God can take into consideration the circumstances that one inherits when determining one's moral profile. Perhaps what really matters is what we do with what we're given, where more is to be expected from those who have been given much (e.g. a good character with ample opportunities to exercise virtue and few opportunities to exercise vice), and little is to be expected from those who have been given little (e.g. a poor character with few opportunities to exercise virtue and ample opportunities to exercise vice). In the spirit of Kant, we might guess that the moral monster in example (ii) who struggles to do some small moral good has a better moral profile than the model citizen who does great moral good with ease. At any rate, there would be no absurdity in supposing so, and God, we may justifiably assume, is capable of measuring such things. The open theist might hope that this is enough to protect the agent's moral profile from luck.
But it isn't. We begin to see this when we observe that, on this sort of account, the agent's moral profile will still vary wildly from world to world. It does so in at least two ways. First, given libertarianism, there are worlds where my moral profile is different from what it actually is due to different decisions I make in circumstances I actually inherit. For example, there's a world in which I freely choose to insult all my students unfairly upon coming to class, which makes my moral profile worse. This sort of variance is no problem for us, since these are worlds that I make non-actual through exercise of my libertarian freedom. (We might call this ‘controlled variance’.)
The trouble is with another kind of variance. For there are worlds where I freely make morally relevant decisions in circumstances which I do not inherit in the actual world. For example, there are worlds in which I have to make some morally relevant decision while being chased by the Yakuza and suffering a neurological rage disorder.Footnote 6 Now, it might be that in some of these worlds, while saddled with these ‘Yakuza-rage-like’ circumstances, I behave in a way that does not change my moral profile. But in others (assuming libertarian freedom), while in ‘Yakuza-rage-like’ circumstances, I behave in a way that does alter my moral profile. (Call this ‘lucky variance’.) This sort of variance is a problem for us. Since we don't (always) control the circumstances in which we are placed, and since, on the current account, our moral profiles vary in worlds in which we are placed in different circumstances, the concern is that we don't have a sufficient degree of control over our moral profiles.
The open theist, naturally, won't say (and in fact can't say) that there would have been lucky variance in one's moral profile had one inherited different circumstances. After all, the open theist will deny that there are true counterfactuals of freedom, and to say that an agent's moral profile would have been different if her luck had been different is to make a claim about what an agent would have freely done in other circumstances, which is off-limits. (Indeed, this is exactly what example (iii) above does.) Let C* be some set of circumstances in which some agent is not placed in the actual world. On open theism, the following claim simply cannot be true:
(1) Had C* been actual, the agent's moral profile would have been different.
So, while there might be lucky variance across worlds, it's an error to say that there would have been lucky variance had the agent's luck been different. At a casual first glance, this might seem to be an adequate response to our problem.
But, once again, this first glance is too superficial. For it follows from open theism that the following claim also is not true:
(2) Had C* been actual, the agent's moral profile would have been the same.
After all, if there is no true counterfactual of freedom pertaining to what the agent would have freely done in C*, then it's not true that she would have acted in a way that would not alter her moral profile. So, not only is there lucky variance across possible worlds, but it's an error to say that there would not have been lucky variance had the agent been placed in different sets of circumstances.
It's this last consequence of open theism that gives the game away. Since the agent's being in C is often something over which she has no control, the crucial question is whether or not her being in C plays an ineliminable role in determining her moral profile. Merely allowing God to take C into account when evaluating her isn't enough, due to lucky variance of her moral profile across worlds. What is needed is some way of protecting the agent's moral profile against lucky variance. The open theist might, I suppose, make some proclamations about what the agent would likely freely do if she were placed in other circumstances, but it is hard to see how this can help. On open theism, whatever proclamations we (or God, for that matter) are licensed to make about what the agent would likely do are justified in terms of the antecedent conditions: the alternatives open to her, her character – in short, everything that goes into a set of circumstances C, which is precisely what her moral profile needs to be protected from. To get that protection, we need something stronger – something along the lines of (2). But open theists, bereft as they are of the counterfactuals of freedom, are unable to provide it. The result: C does play an ineliminable role in determining her moral profile, and open theism is impugned by the problem.
I've been assuming thus far that God's judgement of us is solely in terms of our moral profiles. But this assumption is inessential. Suppose that ‘salvation is by faith alone’, and that our fideistic profile is what matters. Then the problem remains, and what we have said about moral luck will carry over mutatis mutandis to a corresponding problem of fideistic luck. Indeed, so long as God's judgement of us is in terms of some feature(s) of our libertarian behaviour, we will be stuck with a parallel problem. And so open theism falls prey to the problem of soteriological luck. As a consequence, whatever victory the open theist claims with respect to the problem of evil will be hollow. While, on open theism, God might well be off the hook, that news would come as scant consolation to those who would nevertheless be damned due – in large part – to factors beyond their control.
A Molinist solution
Molinists, by contrast, may offer a solution to both the soteriological problem of evil and the problem of soteriological luck. Indeed, the solution has already been sketched in some form by Michael Zimmerman and John Greco.Footnote 7 The Molinist simply needs to accommodate it to fit her peculiar metaphysics.
As opponents of moral luck, both Zimmerman and Greco argue that an agent's moral profile can't be determined merely by what she does – for the actions she performs are clearly subject to luck in the circumstances in which she is placed. Rather, her moral profile is also influenced by what she would do in different circumstances. There is a fair amount of intuitive appeal here. Suppose that one gestures toward some convicted criminal, and asks you: ‘Are you really a better person than so-and-so?’; ‘Well, how would you have acted had you been in so-and-so's shoes?’. That seems to be precisely the sort of question we should be asking if we want to determine your moral profile.Footnote 8 And again, there's nothing special about one's moral profile, per se – we can talk in general about the agent's soteriological profile without specifying what feature of our behaviour is relevant to God's judgement of us.
We are now entering the Molinist's playground. For on Molinism, there's a fact of the matter (knowable by God) concerning what you would libertarianly freely do in different circumstances – indeed, in circumstances that are radically dissimilar from the circumstances you have experienced. And the outlines of a solution to the problem of soteriological luck emerge when we notice that the Molinist is already committed to the following claim:
(3) For any agent S, and for any actualizable set of circumstances C in which S may participate, if S were in C, all counterfactuals of freedom about S would have the same truth-value that they have in the actual world.
The Molinist is committed to (3) because God knows the truth-value of the counterfactuals of freedom pre-volitionally, and so their truth-value is not contingent upon which world God actualizes. So, had God actualized some other world, then the same counterfactuals of freedom about S would have been true. Now suppose that our soteriological profile is determined not just by what we do, but by what we would do for just any set of circumstances in which it is metaphysically possible for us to be placed. Call this view the ‘counterfactual soteriological profile’ position (or CSP for short). Then, given (3) and CSP, had God actualized some other world, S's soteriological profile would not change, for the very same counterfactuals would truthfully describe how S would behave had S been placed in other metaphysically possible circumstances. Our soteriological profile would not change no matter how God created the world – and God would therefore have no control over it.
As a result, CSP is immune to the soteriological problem of evil. Further, we would be on the doorstep to solving the problem of soteriological luck. All we need to add is the following principle, which seems to be a simple consequence of God's omnipotence:
(4) For any S, and for any set of circumstances C in which it is metaphysically possible for S to be placed, C is actualizable by God.
Recall that, on my way of speaking, a set of circumstances includes the atom-for-atom arrangement of the world at a time and the laws of nature, insofar as they participate in inclining forces on the agent. They do not include the causal history of how that arrangement came about. Perhaps I would never freely choose to insult my students unfairly upon coming to class. If so, then God cannot actualize a world where I freely insult my students upon coming to class. Nevertheless, He can actualize the following set of circumstances: the students are at their desks looking at me with embarrassment or anger, and everyone in the room has the ‘memory’Footnote 9 of my having insulted them. All it takes is for God to organize matter in a certain way (and, if dualism is true, the contents of our souls), and certainly God can do that.
According to CSP and (3), one's soteriological profile would not change had one been placed in some other actualizable set of circumstances. According to (4), the set of actualizable sets of circumstances is exhaustive – all metaphysically possible sets of circumstances are actualizable by God. The result is that, for just any set of circumstances in which it is metaphysically possible for S to be placed, had S been placed in those circumstances, S's moral profile would not change. So, again, let C* be some non-actual set of circumstances in which it is possible for the agent to participate. Given CSP, (3), and (4), not only may we claim that:
(1) Had C* been actual, the agent's moral (soteriological) profile would have been different
fails to be true; we may also claim that:
(2) Had C* been actual, the agent's moral (soteriological) profile would have been the same
is true. And that, it will be recalled, is the result that we needed to solve the problem of soteriological luck – and the result that is denied to the open theist.
Objections and replies
CSP offers a very simple solution to the vexing problems that we have been discussing. So why has it not been offered already? There are, to my mind, four objections that prevent Molinists (or those for whom Molinism is a live option) from embracing it.
The theological objection
One might wonder just how well CSP coheres with scriptural authority. In fact, there are biblical passages (especially in the Pauline epistles) which are distinctly unkind to CSP.Footnote 10 But these passages don't speak against CSP per se – they typically speak against the idea that there's anything problematic about God's determining how we are judged. So, if these passages are what keep one from embracing CSP, then they should also lead one to throw out the charge against Molinism with which we began.
The epistemic objection
If CSP is true, then it will be hard to determine anyone's soteriological profile. That is because, even if we know what someone would do for many circumstances she could inherit, we certainly won't know what someone would do for all of the circumstances she could inherit, or even for most of those circumstances. There are simply too many widely differing sets of circumstances to consider.
This consequence will represent a crippling problem for those who are in the business of making comparative judgements of others – trying, say, to figure out whether Jones or Smith will get the bigger reward in heaven. But perhaps that is as it should be.Footnote 11 The more disconcerting thing is that, on CSP, we won't know our own soteriological profile. Doesn't it follow from CSP that no-one may be certain of her own salvation, since, no matter what she does in the actual world, for all she knows she might have behaved reprehensibly (morally, fideistically, or in some other sense) had she inherited different circumstances?
Not quite – for CSP does not carry any commitments concerning how the soteriological bar is set. How good does your soteriological score have to be in order to be saved? Perhaps only this good: performing in a certain way in just one metaphysically possible set of circumstances is sufficient. If so, then anyone who has in fact behaved in that way may be assured of her salvation. Of course, this line will lead us to something like universalism – for while it's conceivable that there be such wholly incorrigible persons that would never perform up to snuff, regardless of the circumstances, it seems unlikely that such persons exist. Obviously enough, the price (if it is a price) of maintaining both CSP and assurance of salvation is that we must give up assurance of damnation. For all we know, the Satan-worshipping drug dealer down the street might have been a saint had luck gone his way – and if so, then on this line, he is saved.
Of course, CSP is not committed to this line either. One could adopt CSP and still assert that the Satan-worshipping drug dealer is damned – but only at the expense of assurance of one's own salvation. For who knows how we would have behaved had we been in the drug dealer's shoes? Naturally enough, that is part and parcel of what it is to take luck seriously.
The metaphysical objection
On CSP, the agent's soteriological profile is determined by the true counterfactuals of freedom that describe her actual and counterfactual behaviour. Now, most Molinists will say that whenever an agent acts with libertarian freedom (say, S does A in circumstances C), she makes a counterfactual of freedom (‘If S were in C, S would do A’) true. But no Molinist that I am aware of will say that we bring about the truth of counterfactuals with false antecedents. Suppose, then, that it is true that, if I were being chased by the Yakuza, I would freely (and foolishly) take out my pocket knife and choose to fight them. Have I made, or will I make, that counterfactual true? How could I? I've never had (nor will ever have) the opportunity to do so. But then the vast majority of the true counterfactuals about me will be true independently of my doing anything to make them true. And since CSP puts a great deal of weight on these counterfactuals with false antecedents, doesn't this show that CSP is impugned by the problem of luck after all?
No. For these counterfactuals are describing facts about that which is most intimate to me – facts concerning my libertarian agency. These facts may well be brute (as most Molinists seem to believe), but then, libertarian agency itself is brute, and the bruteness does nothing to undermine its significance. On CSP, my soteriological profile is dependent on nothing but a complete description of my free will – both how it has been exercised, and how it would have been exercised had things gone differently. Nothing could be further protected from luck than that.
The motivational objection
But there is a related concern that might appear to have more bite. On CSP, I have very little opportunity to alter my soteriological profile. While there may not be a problem of justice here, perhaps there is a problem of motivation. Why do what is right, or accept Christ, or what have you, if my behaviour will have very little influence on my soteriological profile?
Whether this problem has any teeth will depend on what the proper source of motivation for good behaviour (moral, fideistic, or whatever) ought to be in the first place. If behaving morally were simply a matter of avoiding fire and brimstone, then no doubt CSP would come at a prohibitive cost. But it would be very surprising if the main reason to do what is right were to avoid punishment. There seem to be much better candidates. We should do what is right because it is right. Or because moral behaviour is a way of showing our love for God. Or because we were designed to flourish when behaving morally. Or because by sinning we estrange ourselves from God and our fellow man. Or what have you.
But why accept Christ? One traditional answer is that accepting Christ is a necessary and sufficient condition for salvation. On CSP, however, it is either not sufficient (if your counterfactually rejecting Christ in other possible circumstances is enough to condemn you), or it isn't necessary (if your counterfactually accepting Christ in other possible circumstances is enough to save you), or perhaps it is neither.
But it doesn't follow that, on CSP, we would have to reject the standard Christian line that Jesus came to save us from our sins. It would simply mean that counterfactually accepting or rejecting Christ is also relevant to determining one's soteriological profile. Further, there may be extremely good non-soteriological reasons for accepting Christ. One reason is very simple: that Christ deserves acceptance and worship. Further, the abundant life that Jesus promisesFootnote 12 may very well begin immediately – with a personal relationship with Jesus, or with a sense of meaning and purpose to history and one's life work, or with a renewed love for one's fellow man, or with a deeper understanding of who one is and one's place within the world. And that alone, it seems, would provide excellent reason to accept Christ.
It would doubtless come as a surprise to many evangelists that this benefit is what they are working to spread, as opposed to salvation itself. Many evangelists take themselves to be spreading the gospel message for the purpose of improving others' soteriological profile. To them, no doubt, CSP comes with a notable (perhaps prohibitive) cost. But then again, if, contra CSP, evangelism really does work to make a significant impact on others' soteriological profile, we inherit a different cost – namely, that God's eternal judgment of an individual person will hinge on the behaviour of somebody else. I leave it to the reader to determine which cost is the more severe.
Conclusion
In the end, CSP seems to provide exactly the sort of solution that is demanded by the problem of soteriological luck and the soteriological problem of evil. But it also comes with consequences that many would prefer not to embrace. It does not fit with some scriptural passages in the way one might hope. It saddles us (although not God!) with epistemic limitations, and it undermines certain egoistic motivations for acting in ways that are pleasing to God. But these results, I have suggested, are the natural consequences of taking luck seriously. And if one hesitates to accept them, then perhaps one should hesitate to accept the charge against Molinism with which we began as constituting a problem in the first place.
Nothing said in this paper provides evidence that Molinism is true. But with respect to the issues discussed, it wouldn't be bad news if it were true.Footnote 13
