The Incompatibility of Middle Knowledge and Libertarian Freedom

by Max Andrews

The following is a guest blog post by Roger Turner, a Philosophy Ph.D. candidate at the University of Tennessee.

If you’re a Molinist—at any rate, if you believe that God has middle knowledge—you believe that God knows whether or not any given counterfactual, or subjunctive conditional is true.  What’s more, you probably believe this because you think it helps make clear how it is possible that humans can act freely even though God knows, before they act, how they’ll act.  And you probably want to have some grasp on how all that works because you’re a libertarian with respect to freedom.  That is, you think that determinism and free will are incompatible, but you believe that divine foreknowledge and free will are compatible.  If you are a libertarian, you think that indeterminism is true.  In other words, you think that the theses ‘the conjunction of the past and laws of nature entails a unique future’ and ‘nobody has a choice about the future because nobody has a choice about what God foreknows’ are false.  So, the idea of middle knowledge seems to be the best way to duck and dodge the relevant snares.  But there’s a puzzle here.  There appears to be a significant tension between one’s being a libertarian about freedom, and one’s being a Molinist.  In what follows, I hope to illuminate the problem.

Take your friend, Jones.  If you believe God has middle knowledge, you probably believe that, if you were to ask God what Jones would do in such and such a circumstance, God would know the answer to your question.  God would respond, so you think, with something like the following:  “If Jones were in C, he would freely A.”  And you’d feel pretty confident that God’s having answered this way fails to undermine Jones’s freedom because you think that God’s belief about whether or not Jones A’s in C depends on Jones and whether or not Jones A’s in C and not the other way around (i.e. Jones’s Aing in C doesn’t depend on God’s knowing that, if Jones were in C, he would A).

Quick question, though: what is the truth value of a subjunctive conditional if indeterminism is true?  Here’s why I ask.  If a particular event is an indeterminate event (that is, the event is undetermined) the odds of that event’s happening are something like 50/50.  So, take Jones again and his being in C and whether or not he A’s.  If Jones’s Aing is undertermined, then he’s just as likely to A as he is to not-A given his being in C and an identical past up to the point of his being in C.  This is what indeterminism implies.  Given Jones’s past from t0 to the present moment, the moment just before he acts, Jones is supposed to be just as free to refrain from Aing as he is to A.  It can’t be the case, for example, that, if Jones is libertarianly free, he’d be more likely to A, given his past (or other conditions beyond his control), than he would be to refrain from Aing.  Because if that’s the case, then he’s got factors that are out of his control which bear on whether or not he A’s.  And this would absolve Jones of (at least part of) his responsibility for Aing (or refraining from A).  If we’re libertarians (and we usually are if we believe God has middle knowledge), then we think Jones has just as much chance of Aing, given that he’s in C, as he does refraining from A.

Okay, back to the question about the truth value of a subjunctive conditional given the truth of indeterminism.  The subjunctive conditional (this thing:  €®) expresses what would happen in the closest-by, relevant possible worlds.  So, take Jones again.  And express the proposition ‘if Jones were in C, he would freely A’ as follows:

JC nec. €→ A

The way this is typically read is something like this: in all the closest-by C worlds (i.e. the worlds in which Jones is in C), Jones A’s.  But this implies that it’s not indeterminate what Jones would do in C.  It’s not indeterminate because we can “zoom-out” and see which C worlds are the closest to the actual world.  And, just by doing that, we can see that in any C world that is most closely related to the actual world, Jones A’s.  The odds aren’t 50/50 given the truth of the subjunctive conditional; the closest-by C worlds are such that Jones A’s.  So, if Jones is in C, though it’s not necessary that he A—there are, after all, possible worlds where he’s in C but doesn’t A; they’re just further off—he’ll A.  He’ll A because there is a group of possible C worlds—those closest-by to the actual world—where he A’s.  And whether or not Jones is in a closest-by C world has nothing to do with Jones!  It’s beyond Jones’s control as to whether or not he’s in one of the closest-by C worlds.

But, wait.  On the libertarian view, indeterminism is true; that is, determinism is false.  Jones must, to be libertarianly free, be just as likely to A, in C, as he is to refrain from A.  That is, Jones must be equally likely to refrain from Aing as he is to A in all C worlds.  It can’t be the case that the proximity of the C world to the actual world determines whether or not Jones A’s.  But, if the subjunctive conditional (this bit:  JC nec. €→A) is true, then it can’t be that Jones was just as likely to refrain from Aing in C as he was to A in C.

So, if indeterminism is true, the subjunctive conditional must be false.  The answer, then, to the above question ‘what is the truth value of the subjunctive conditional if indeterminism is true?’: the truth value of the subjunctive conditional, if indeterminism is true, is False.  But this means that it’s false that if Jones were in C, then he would freely A.  And what’s more, it’s false that if Jones were in C, then he would freely refrain from Aing.  And since God can’t have any false beliefs, he can’t believe that if Jones were in C, then he would freely A is true, nor can he believe that if Jones were in C, then he would freely refrain from Aing.  He can’t believe this because it’s false that if Jones were in C he would freely A and it’s false that if Jones were in C he would freely refrain from Aing.  It’s false because there is no would to it; indeterminism implies that the proximity of the worlds in which Jones is in C has no bearing on whether or not Jones A’s.  It has no bearing because, in all C worlds, Jones is just as likely to refrain from Aing as he is to A.

The upshot of all this is that if it’s true that if Jones were in C, he would freely A, then Jones was not libertarianly free to A.  Some factor that wasn’t up to Jones made it the case that his particular C world was closer to the actual world than some other C world, some other C world where he refrains from Aing.  And this implies that, if it’s true that if Jones were in C, then he would freely A, then libertarianism is false.  At any rate, it implies that if the subjunctive conditional is true, then God’s middle knowledge rules out our acting in a libertarianly free way.

So, either you are a Molinist—at any rate, you believe that God has middle knowledge—or you are a libertarian.  But you can’t be both.  So the argument goes, anyway.


5 Responses to “The Incompatibility of Middle Knowledge and Libertarian Freedom”

  1. This is exactly the reason I’m an open theist instead of a Molinist. Good post! One thing though: couldn’t you be a libertarian and a Molinist if you held to the B-theory of time? Because then in the actual world God could “see” what your four-dimensional time worm libertarianly chooses in the “future.” I don’t know how this would work with possible worlds; I guess you would have to hold that every possible world actually exists with its own B-theory of time, such that God could “see” those free choices as well.

    But then the Molinist has to reject the A-theory and bite the bullet of all the problems of saying every possible world actually exists. (And why would they all only be B-theory worlds?).

    Yup, this is the only problem holding me back from Molinism: how to reconcile middle knowledge with libertarian free will and an A-theory of time.

    • I think I was trying to say this: maybe it’s not that Molinism and libertarian free will are incompatible, strictly speaking, until we also throw in A-theory of time.

      Libertarian Free Will + A-theory of time = no Molinism?

  2. I found this to be a thoughtful article, but I want to offer a few friendly criticisms (as a Molinist). First, it is said: “If a particular event is an indeterminate event (that is, the event is undetermined) the odds of that event’s happening are something like 50/50. So, take Jones again and his being in C and whether or not he A’s. If Jones’s Aing is undertermined, then he’s just as likely to A as he is to not-A given his being in C and an identical past up to the point of his being in C. ” But I find this confusing. First, if it is meant in an epistemological way, I don’t think it’s at all true that libertarian free will requires this. In 10 minutes, I will either hit my wife or I will not hit my wife. Who thinks it’s equally likely that I do one or the other? If it is epistemological, there’s really no problem for Molinism, as God doesn’t need to probablistically believe any specific proposition. He would just know it.

    However, I suspect it’s meant to be ontological. But in that case the sentence really means something like “Jones is equally able to choose to A or not to A.” This is the standard definition of libertarian free will. However, it’s worth noting “equally” is superfluous to the proposition, since one is either able or not able. Hence, it’s really only true that “Jones is able to choose to A or not to A.”

    The meat of the problem seems to be in the standard semantics for counterfactuals, where some counterfactual can be said to be true just in the case that in all of the worlds with the same antecedent conditions (or most similar to the actual world) the consequent follows. First, this is only an epistemological way of accounting for which propositions we can justifiably say are true; not an account of which counterfactuals are in fact true. There are probably millions of counterfactuals that we have no warrant for that are nonetheless true; we simply don’t have justification to believe them either way.

    But I don’t think it’s too much of a problem to accept the standard semantics. For when it is said, “He’ll A because there is a group of possible C worlds—those closest-by to the actual world—where he A’s,” I think this makes the mistake of assuming circumstantial causation. He doesn’t A in the actual world because of a group of possible C worlds; he As because he chooses to, and it’s just simply the truth of what he would choose. Since the semantics are simply an epistemological, rather than ontological, construct, I don’t think it follows that Jones acts non-libertarianly. For Jones only acts “because” in terms of explanation or influence, not in terms of causation. Suppose one is raised in a family that hates God and loves drinking. Children imitate adults.. It seems likely, on the face of it, that the child will grow up to love drinking, even though he may choose to refrain from drinking. The explanation for his alcoholism in a non-causal sense is his upbringing; his causal explanation is his choice to drink. Though we do not have counterfactual power over the past, so that you are right to say Jones has no control over the circumstances in which he finds himself (at least prior to his birth), he nonetheless does control his individual choices made.

    I think it’s also far more plausible to believe that God knows the truth value of would-counterfactuals and that some of them are true than to believe that God thinks all would counterfactuals are false, including their negations.

  3. Brilliant. A great blog all-round. A joy to read!

  4. Trackbacks

Leave a Reply