Monday, April 25, 2011

Questions about Middle Knowledge

I have found while people have many questions about the teaching of middle knowledge, once explained it turns out to have been intuitively held by these same questioners. What I mean is that something very much like middle knowledge is believed to be true by the majority of Christians not already committed to some form of Calvinism or Arminianism. What is middle knowledge?

Certainly, it is not (directly) concerning predestination and free will (though it has application in that). Primarily, middle knowledge is related to omniscience. The simplest and most agreed-upon definition of omniscience is that “for any true proposition P, God knows and believes P and does not know nor believe not-P.” This means that whatever is true, be it in the future, now, or past, God knows it. Thus, quite literally, God knows what you are going to do before you do![1]

In order to understand middle knowledge in the context of omniscience, we finite beings break down God’s knowledge by logical relationship. First, there is God’s natural knowledge. This contains knowledge of all necessary truths (like “2+2=4” or “there are no married bachelors”) and all logical possibilities. Thus, one could say this is God’s knowledge of everything that could be. Next, there is God’s free knowledge. This is called “free” because the content of this knowledge is what God chose to be so. This includes God’s knowledge of this actual world (i.e. everything that is true in the history of the world up till now, and indeed throughout the potentially-infinite future).[2] One could say this is God’s knowledge of everything that will be. Finally, we have God’s middle knowledge. This is knowledge of a counterfactual form. This form is “if Gary were in circumstances C, Gary would freely do X.” One could say this is God’s knowledge of everything that would be in any other circumstances. In this way, God’s knowledge spans what could be, what will be, and what would be in every circumstance.

Middle knowledge is actually the conclusion of an argument from counterfactual knowledge. I have already explained the idea of counterfactual knowledge. Is it biblical? Absolutely! 1 Cor. 2:8 states, “Which none of the princes of this world knew: for had they known it, they would not have crucified the Lord of glory.” See the counterfactual? “If the princes of this world knew X, then they would not have crucified the Lord of glory”! Either what Paul is saying is true, false, or meaningless. Since it seems Paul really is conveying a truth by these words (and not a symbolic, deeper truth as in a parable), we can rule out meaningless. It also seems that Paul is teaching rather than relaying some other account, so that to say Paul was wrong is attacking the doctrine of inerrancy (not to mention we don’t have overriding reasons to think Paul was wrong). The only option left is to believe it is the truth. 1 Samuel 23:10-12 relay the story of David asking the Lord the counterfactual question, “If Saul comes down to Keilah, will they deliver me up?” The Lord answered in the affirmative. Since they did not deliver David up, this is a true counterfactual. Yet the Lord knew it! It seems the case for counterfactual knowledge, at least biblically, is quite solid. God knows what would happen in any other circumstance.

However, as some opponents have been quick to point out, counterfactual knowledge does not, in and of itself, mean middle knowledge. What would make it middle knowledge? Either counterfactual knowledge is known to God logically prior or logically posterior to the divine decree to create the world (or what we had called God’s “free” knowledge).[3] Essentially, counterfactuals here boil down to free choices of individuals. So then, either free will exists or God directly causes individuals to act. While much more could be written, it seems intuitive (for those not already committed to a position) and obvious, both biblically and experientially, that mankind has a free will. Just note if God causes individuals to act he causes them to sin.

So, if free creatures freely make choices, then it is not true that they act because God causes them to act. If that is the case, then counterfactual knowledge is known to God prior to the creative decree (logically). This is also very good, since if God does not force people to act and yet lacked counterfactual knowledge until the divine decree, then God would be completely lucky in getting this actual world. It gets worse: without this knowledge, God would have no idea how any of us would act in situations, including this actual one! God must, in order to create sovereignly and omnisciently, have counterfactual knowledge; specifically, he must have middle knowledge.

God’s knowing what any free creature would do in any set of circumstances is both biblically and intuitively held. If you ask most people on the street without using the relevant theological terms, “do you, as a Christian, think God knows what would have happened in any other set of circumstances?” they would say “yes.” If you asked them if they believe in free will, they would say “yes.” As we have seen, this just makes middle knowledge analytically true (that is, true in virtue of both of those prior question’s answers!). If you believe in counterfactual knowledge and free will, you believe in middle knowledge.

                [1] While for some the idea that future contingents can be true now is controversial, we shall proceed with the intuitive knowledge that what I will do tomorrow is true as a datum.

                [2] A discussion of why the future is potentially, rather than actually, infinite is an interesting one, but not one which ultimately matters. For our purposes, just know that God knows everything that will happen in this actual world.

                [3] Please note that the usage of “logically prior and posterior” does not in any way have to do with time, as though God lacked knowledge at any point.


  1. These kinds of Molinist counterfactuals seem to require a rejection of either the standard resolution of vagueness of counterfactuals (as discussed by David Lewis) or the rejection of contra-causal free will, namely because the conjunction of these two entail that P in A[]-> P B's is false when P has a choice to B or C in situation A. Or, at least this is the way it seems to me. Would you prefer to reject Lewis' standard resolution? What alternative vagueness resolution scheme would you find more appropriate for the Molinist?

  2. Hi Jake, thanks for commenting. It's late, so perhaps I am misunderstanding you, but are you arguing that if the counterfactual "if P were in A, then P would do C" is true, then it is necessarily false that P does B in A? It seems Lewis would not argue this way, precisely because he believed in the type of modal realism which meant that all logical possibilities existed in parallel type universes, so that the standard semantics for Lewis would not have entailed this (else this is the only possible world, which Lewis did not hold).

    Of course, perhaps instead of meaning that it is necessarily false P does B, you mean it is necessarily false P would do B in A? In this case, we must distinguish between the necessity of a statement (such as invoking the law of noncontradiction) and the necessity of the content of that statement. The former is just an application of the law of noncontradiction: If X is true, it is necessarily false that not-X is true; so that if B and C are mutually exclusive, then it is necessarily false B and C are done. But I see no reason to move from that bit of tautology to apply the modal necessity to the content of that choice, even if P only does C in A and not B; it is contingent upon the will.

    So to answer directly, I see no reason to reject Lewis counterfactual semantics out of hand, nor do I think the Molinist must. But a very interesting question indeed!

  3. Hey Randy,

    Thanks for the late reply. I think that my general point got missed, but I think I was perhaps unclear. The standard resolution of vagueness, as discussed by Lewis in "Counterfactual Dependence and Time's Arrow", provides an ordering of possible worlds and determines the truth-value of counterfactuals based upon the make-up of the "closest" worlds under this ordering. Under this resolution, contra-causal free will entails (P in A <>-> P B's) and (P in A <>-> P C's) in a situation where P is free to choose between B and C in situation A. But the Molinist must deny at least one of these counterfactuals because either (P in A []-> P B's) or (P in A []-> P C's). Thus, the Molinist, it would appear, must deny either contra-causal free will, the standard resolution, or Lewisian semantics for counterfactuals. I think the easiest point to deny would be the standard resolution and instead opt for a sort of backtracking resolution, but I wonder if there's an independent reason to do so. Perhaps there is. But without an appropriate resolution for Molinist counterfactuals, I'm a bit worried about their truth-values and their interplay with counterfactuals of the usual sort.

  4. Hi Jake! I knew this was for adjudicating which counterfactuals were correct in virtue of their being closest to the actual world; I just wasn't sure how to respond (it was late for me after all!). However, I am curious: why think the Molinist asserts that P in A necessarily B's or P in A necessarily C's? As I see it, if B and C are not compossible, then really the Molinist only asserts []P B ∨ P C; but Molinists would not distribute the necessity to P's doing one or the other simply because he in fact would do it. Many Molinists do in fact believe if different counterfactuals were true, then different results would ensue, and not at all that necessary entailments are in view with respect to their choices. I hope that helps!

  5. The Molinist is committed to the truth of counterfactuals of the form, "If P were in A, then P would B". This is what I'm signifying as (P in A []-> P B's), namely because there is a fact of the matter about how I would act under situation A under this view (at least as I understand it). But this cannot be true under the standard resolution of vagueness together with contra-causal free will. This is because contra-causal free will, under the standard resolution, entails "If P were in A, then P could B" and "If P were in A, then P could C", which I'm signifying as (P in A <>-> P B's) and (P in A <>-> P C's), respectively. But (P in A []-> P B's) is incompatible with (P in A <>-> P C's), at least if it is non-vacuous (and likewise for the other pair). This is because, under the possible worlds semantics, the first says something to the effect of, "From the closest possible world at which P is in A, all of the nearest worlds are such that P B's" while the latter says, "From the closest possible world at which P is in A, some among the nearest worlds are such that P C's." It's open, perhaps, to the Molinist to accept some other resolution of vagueness for the Molinist counterfactuals, but this would be to say that their counterfactuals are not of the usual sort (by failing to track causal relations). I don't have too much of a problem with that, as long as it could potentially be independently motivated.

  6. Hi Jake. I see now you're saying Molinism entails this, rather than that Molinists claim this (since of course they would reject this. Molinists (such as William Lane Craig and Thomas Flint) believe there are such possible worlds where if P were in A, then P does C instead of B, even if it is true if P were in A, then P would do B; they preserve the distinction between possible and feasible worlds. With the standard resolution, this is an epistemic judgment; for an omniscient being, he by definition would not need to employ the standard resolution. So while I think the standard resolution is quite good and accessible for us finite knowers, its being merely epistemic instead of an ontological guide need not trouble us here. That is, a Molinist may freely embrace all of the closest worlds P does B (because a world in those precise circumstances P would not do C; yet there is a world in which P does C). This is why I object to the use of the necessary modal operator (as will any and every Molinist, in fact!).


Please remember to see the comment guidelines if you are unfamiliar with them. God bless and thanks for dropping by!