By David Misialowski (2006)
In the third chapter of his dissertation, Robert considers the Molinist solution to the alleged foreknowledge/free will incompatibility. It is proposed that God has Middle Knowledge, which lies roughly between God's natural knowledge (of all necessary truths and possible truths) and free knowledge (of what God Himself will freely do, and hence the actual conditions of His creation). As Robert explains, this supposed new type of knowledge is intended to make it possible for God to be omniscient and for agents to be free. The concept of middle knowledge introduces the concept of counterfactuals.
The idea, roughly, is that while God knows what I will do at some particular time in some particular scenario, he also has counterfactual knowledge of what I would have done had I been placed in a different set of circumstances. Apparently this is intended to preserve free will on the ground that depending on the particular scenario, I can freely choose my response. God just knows, given any particular scenario, what I would, or wouldn't, do.
There is an extensive literature on Molinism, which was forumulated by the Spanish Jesuit Luis Molina. What Molinism boils down to, though, is this: The Molinist wishes to say that the formulation, If God knows I will do p, then I will do p, is wrong; rather it is correct to say: If I do p, then God knows I will do p; where I do p freely.
This is, of course, the point I have argued for all along. So I agree with the Molinist take insofar as it embraces the content of God's foreknowledge as a semantic but non-caused result of human free will.
As I have tried to argue, rightly or wrongly, well or badly, modern interpretations of modal logic can give us a lot of confidence in the claim that, If I do p, then God knows I will do p, where I do p freely. Molina, though, did not have access to those logical arguments, and it seems that he felt compelled to elucidate a big superstructure of ideas to accommodate his (correct) intuition that God's infallible foreknowledge entailed nothing, but instead was entailed by the free acts of moral agents. One of the reasons he felt obligated to do this was that, in addition to making God's omniscience compatible with human freedom and moral responsibility, he wished to preserve God's sovereignty, the notion that God, so to say, ruled the world, and no scenario played out without his willing it. At this point the Molinist doctrine seems to go adrift (at least that is my impression), but this doesn't worry me because I am not concerned with God's sovereignty, or for that matter, with whether God exists at all. (Maybe I should note that I am an agnostic atheist and Robert is an agnostic theist, so perhaps we make good bookends.) I'm currently only concerned with the logical problem of some essentially omniscient agent (who might not even be God) coexisting with human freedom. To the extent that Molinism goes beyond the flat claim that human free acts determine God's foreknowledge, I find Molinism superfluous.
Nevertheless it is interesting because it raises curious questions about counterfactual conditionals that I will briefly examine.
Before getting to that, however, let's look at this issue of making God's sovereignty and our freedom work together. How does Molinism try to do this?
It does this by positing the idea that in addition to knowing what I will freely do given some particular scenario, God also knows what I would freely do in every possible (though not actualized) scenario; i.e. God knows the truth value of all counterfactual conditionals. To return to my ongoing example - the case in which I kill my neighbor because he subjected me to the repetitive barrage of Night Fever by the Bee Gees - God knew, even before He created the world, that should I be placed in the particular scenario of besiegement by the Bee Gees, then I would kill my neighbor in retaliation for his subjecting me to that besiegement. But this raises a potential problem, as Robert notes in his dissertation: "How does it follow that I make a choice in a scenario freely, if my choice is determined by the scenario I am in and yet I do not get to choose my scenario?"
From the standpoint of modal logic that we have considered, the fact that I did not get to choose my scenario might or might not be a problem. But for the Molinist, it is important that God choose the scenario, because this ability to choose preserves God's sovereignty. In essence, according to the Molinist, God knows, in advance of His creation of the universe, every logically possible outcome of every scenario; and then He chooses to actualize a single universe - one that includes, among a fabulous wealth of detail, a scenario in which my neighbor bombards me with Bee Gees bombast, and I return the favor by (freely) killing him.
Here's the problem: What grounds God's infallible foreknowledge of counterfactual conditionals? Well, God is God, and so knows all things. But how can even God know the outcome of events in worlds that do not, and did not ever, exist? This raises the general problem of what provides the truthmaker of counterfactual statements. But however daunting the problem might be, it seems the Molinist must argue for God's infallible foreknowledge of the outcome of all counterfactual conditionals, because unless God possessed this kind of knowledge, then His creation of the actual world would be a kind of crap shoot. How would God know, for instance, that the world He actually created was the best of all possible worlds (assuming that this was His goal) unless He also knew what all other possible worlds would be like?
The logical problem of treating counterfactuals was not well understood in Molina's time, but it has to do with the weakness of standard propositional logic in dealing with non-instantiated events. A summary explanation may be found in Christopher G. Small's Reflections on Godel's Ontological Argument, which is the subject of a thread in the Explore Forum. Small writes:
It is beyond the scope of this reply (and perhaps also beyond my current competence) to discuss in detail modal and propositional logic, but in his dissertation Robert raises the challenge of what is called modal realism, sometimes also known as extreme modal realism, which was introduced in the late 20th century by the philosopher David K. Lewis. Modal realism presents new challenges (threats?) to Molinism and even to standard belief in God. I will briefly to examine those challenges.
Lewis proposed that there is an elegant solution to the problem of what makes counterfactual statements true. The truthmaker of counterfactual statements is that counterfactual statements are true at concrete worlds that are not actual worlds. That is, these worlds are not actual to us, inhabitants of this world (those of us who are exchanging views at The Galilean Library, for example). But the worlds that are counterfactual from our perspective are actual from the standpoint of the inhabitants of those worlds. As Robert writes in his dissertation:
Lewisian worlds exist in what might be called Logic Space. Hence, under this thesis, it follows that all logically possible events, outcomes or entities are actual somewhere in Logic Space. Since it is logically possible that talking donkeys exist, then talking donkeys do in fact exist (just not in our world).
It has been famously noted that Lewis's hypothesis, when grasped, generally elicits "the incredulous stare". Of course, since these worlds (of which their potentially seems to be infinitely many) are causually and spatiotemporally isolated and exist only in Logic Space, their existence (or non-existence) can never be empirically verified or refuted even in principle. Nevertheless, Lewis and other exponents of modal realism argue that they hypothesis is serviceable because, among other reasons, it answers the question of what makes counterfactual statements true. But in his essay, Robert points out the problem that this thesis (which, obviously, Molina never contemplated) presents for Molinism:
If Lewis's account of modality is right, Molinism collapses into incoherence on any number of grounds. Again, suppose that my neighbor bombards me with the Bee Gees, and I kill him in response. The Molinist account is that God knew, in advance of His free act of creation, that if I were placed in a Bee-Gees bombardment scenario, then I would (freely) respond by killing my neighbor. Choosing among all possible worlds, God, for whatever reason, created the actual world in which I was placed under Bee Gees bombardment, and responded by killing my neighbor.
But, if the Lewisian account of modality is correct, there is a logically possible world in which, in response to being placed under Bee Gees bombardment, I decline to kill my neighbor (no matter how much he deserves such punishment). Instead, I merely punch him out; or maybe I give him a stern talking-to, or whatever. The point is that, logically, there are perhaps an infinite number of responses I could make to Bee Gees bombardment, and if actuality is indexical - akin to the indexical "here", where "here" is New York City for me, but "here" would be Melbourne, Australia for someone in Melbourne - then in fact I do make all those responses. So it can't be the case that God knows in advance that if placed in scenario S I will respond by doing p; on the contrary, I will respond in every logically possible way to scenario S, though each response will happen in a different, concrete world, where each world is spatiotemporally isolated but exists in Logic Space.
But, putting God aside for a moment, the Lewisian account of possible worlds raises the problem of trans-world identity. What does it mean to say that in this world, I killed my neighbor, but in another world, I merely punched him out? How can it be that the "I" in this world is the same as the "I" in the other world? Lewis contends that strictly these two versions of me are not the same; rather they are counterparts of each other. Perhaps this offers some wiggle room for Molinism: strictly, the "I" who kills my neighbor is a different entity, in some baroque logical sense, from the "I" who merely punches out my neighbor; and God is able to differentiate between the two.
Unfortunately, this won't rescue Molinism, because other problems surface. If all logically possible worlds are actual worlds, this makes a hash of the notion that God freely chose to actualize one world only. But then, did He actualize all possible worlds? Or does God, as Robert discusses, Himself have counterparts at all possible worlds? And did each counterpart actualize the particular world that the counterpart God finds himself in? Another possibility is that there is one God only, and that he transcends Logic Space, in the same way that he supposedly transcends ordinary time and space. Whether this account can be made compatible with extreme modal realism is unknown (at least to me). And even if that is the case, we are back to the problem of why God actualized all possible worlds, when presumably he actualized this world (the world in which we happen to find ourselves) for some particular reason - that it proved to be the best of all possible worlds, for instance.
In closing I would note that according to Lewis himself (or at least according to interpretations of his thought; I'm not sure I have come across any unambiguous claim by Lewis on this matter) God really does exist, but not at our world. This thesis demotes God from a necessary being to a possible being. The thinking goes that since it is logically possible that God exists, then He just does exist at any number of possible worlds in Logic Space. Both theists and non-theists are likely to find this conclusion deeply discomfiting: the theist because modal realism demotes God from necessary to possible, and the non-theist because the non-theist (or some of them, anyway) wishes to deny that God is possible at all. One non-theist rejoinder would be to argue that God is, for example, a physical impossibility; nothing in the nature of our world gives support to the possible existence of an "omni" being. But to argue in this fashion would be to misinterpret Lewis: He is not interested in physical possibility but logical possibility. Hence, while talking donkeys, too, are physically impossible in this world, they are logically possible at some world and hence do exist at some world (where the laws of physics are different, for example). As the philosopher John Leslie has noted, Lewisian modal realism embraces not just God, but all kinds of Gods; for instance, here is a concrete world, indexically actual to its inhabitants, where the Greek Gods literally exist.
Of course, modal realism might be wrong, and logicians certainly do not uncontroversially embrace it. But from a philosophical standpoint it has something to recommend it, and if it is right, Molinism seems incoherent. More, reconciling modal realism with standard accounts of God would seem to present a theological challenge that needs to be addressed.