Middle Knowledge: A Reformed Critique

By Travis James Campbell

[Travis J. Campbell is a history teacher at Deerfield-Windsor School in Albany, Ga., and adjunct professor of philosophy at Darton College in Albany, Ga.]

Luis de Molina’s solution to the freedom/foreknowledge dilemma has had a revival of sorts in the latter half of the twentieth century, most notably through the efforts of William Lane Craig,[1] Alfred Freddoso,[2] Jonathan Kvanvig,[3] Thomas Flint,[4] and Alvin Plantinga.[5] In short, these thinkers have followed the Jesuit Counter-Reformer in postulating a middle knowledge between God’s natural and free knowledge.

Molina was apparently troubled by the intuition—held by many—that an infallible foreknowledge residing in the mind of God implies a denial of human freedom, at least as it is understood by most libertarians. Indeed, on the libertarian model of human free will, the free agent has contra-causal freedom to choose other than what he in fact chooses at the time of choice. Or, in the words of Molina, “.. . given the same disposition and cognition on the part of the intellect, the will is by its innate freedom able to will or to dissent or to neither will nor dissent.”[6] However, if God infallibly knows that, for example, Jay will eat steak for dinner tomorrow at 7:00 p.m., then Jay will certainly do so. Hence, there is some sense in which Jay is unable to not eat his steak. Indeed, if he were categorically able to not eat the steak, then God’s foreknowledge is not infallible.

Molinists have rightly resisted the move often taken by pagans (e.g., Cicero), heretics (e.g., Socinus), and heterodox theologians (e.g., open theists) to insist that humans possess libertarian freedom to do otherwise and, hence, God does not possess an exhaustive foreknowledge of future contingencies[7] - that is, that humans have libertarian freedom and God lacks omniprescience.[8] The doctrine of middle knowledge is the Molinist solution to the problem of reconciling the exhaustive foreknowledge of God with the (libertarian) free will of man. Molina and his followers have postulated three logical moments in the life of God, which some[9] have diagrammed as follows:

• Moment 1: God’s natural knowledge of everything that could be
• Moment 2: God’s middle knowledge of everything that would be Divine Decree
• Moment 3: God’s free knowledge of everything that happens in the actual world

Moment 1: Natural Knowledge. Prior to the creation of the space-time continuum, God’s knowledge exists as a timeless intuition that comprehends all truth. Nevertheless, Molinists argue that it is possible for us to distinguish conceptually the divine cognition into three orders or types or (logical) moments,[10] the first being natural knowledge.

The divine natural knowledge comprehends every possible state-of-affairs that could obtain. In short, the natural knowledge of God comprehends the merely possible. The divine cognition is able to comprehend an infinite variety of possible worlds that could exist if he were to will them into being. Thus, God knows that Hitler could get into art school, that Paul could reject Christ on the Damascus road, that Caesar could choose not to cross the Rubicon, that the month of July could normally be cold in North America, that Peter could choose not to deny Christ, and so forth. This knowledge is called natural because it exists in the mind of God, as it were, naturally and essentially—that is, independently of the divine decree.

Moment 3: Free Knowledge. The free knowledge of God comprehends everything that has/is/will happen in this existent and contingent world. Hitler could get into art school (that is logically possible), and Peter could choose not to deny Christ (that also is logically possible). But that is not what actually occurred. Indeed, on the Molinist scheme, God sovereignly decreed that Hitler will not get into the school of his dreams, and that Peter will deny Christ. Hence, this knowledge is called free because it is based on God’s free decision to create this world and not another.

Moment 2: Middle Knowledge. As we will see, both Thomists and Calvinists have traditionally distinguished divine natural knowledge from divine free knowledge. Hence, there is some common ground between Thomists/Calvinists and Molinists/Arminians regarding Moments 1 and 3. However, Molinists/Arminians also postulate a middle knowledge which is intermediate between natural and free knowledge. God’s middle knowledge comprehends everything that would happen if God had been willing to decree its occurrence. Caesar could choose to cross or choose not to cross the Rubicon, and God actually decreed that he will cross the Rubicon in the first century B.C. But God also knows what Caesar would do if he had lived during the first century A.D., or if he had been challenged by Pompey a year earlier, or if Crassus had decided not to join the First Triumvirate. God also knows under what circumstances Peter would and would not have chosen to deny Christ. He also knows under what circumstances Judas would have chosen to remain faithful to Christ even unto death. Hence, through his natural knowledge God knows that Judas could either betray or not betray Christ; with his middle knowledge God knows under what circumstances Judas would betray Christ and would not betray Christ; and by his free knowledge— based on his decree to make this world and not another—God knows that Judas will betray Christ.

Another way to understand middle knowledge is to suggest that God knows via the scientia media all counterfactual states-of-affairs. A counterfactual is a state-of-affairs that is counter to what actually takes place in the real world. To put it another way, a counterfactual proposition is a subjunctive conditional which presupposes the falsity of the antecedent. A few examples of such counterfactual propositions are as follows: if the sun had not risen this morning, I would not be typing these words; if I had only studied harder, then I would be a better philosopher; if my in-laws had traveled to Phuket on Christmas 2004 they would have been swept away by a terrible tsunami. According to the Molinist, God’s middle knowledge encompasses all counterfactuals of (libertarian) free choices. That is to say, God comprehends through the scientia media under which circumstances Peter either would or would not use his libertarian freedom to deny Christ, as well as the number of times he would have expressed his denial. So, speaking hypothetically, had, for example, state-of-affairs “A” obtained, it just so happens that Peter would not have denied Christ; had state-of-affairs “B” obtained, Peter would have denied Christ only once; had state-of-affairs “C” obtained, Peter would have denied Christ only twice; and so on. To be sure, Peter possessed contra-causal freedom to do otherwise under all of these circumstances. Nevertheless, the divine mind knew exactly how Peter would use his free will under any set of circumstances in which God decided to place him. God also knew that Peter would betray Christ three times if circumstance “D” were to obtain. And when God decided to exemplify this state-of-affairs through his divine creative decree, he (freely) knew with certainty that this is exactly what would happen (long before it actually did happen).

Hasker points out that modern defenses of middle knowledge differ from the classical defense “in virtue of the application to the counterfactuals of freedom of the powerful possible-worlds semantics for counterfactuals....”[11] The reader should see the importance of advocating the truth of counterfactuals, for if counterfactuals can be true, then we have some basis for adhering to the philosophical coherence of middle knowledge. Indeed, if counterfactuals are true,[12] then it must be admitted that God knows them, since his omniscience guarantees that he knows all true propositions (Ps 145:7; 1 John 3:20). And if they are true prior to the divine decree, then God must know them through his middle knowledge.[13]

Craig has argued that middle knowledge may very well serve as a rapprochement between Calvinists and Arminians.[14] In order to prove this, he sets forth the distinction between Thomism, Congruism (which gets many of its insights from Francisco Suarez),[15] and Molinism in the following synopsis:[16]

Thomism	Congruism	Molinism
1. God decides absolutely and gratuitously to predestine S to glory.	1. God decides absolutely and gratuitously to predestine S to glory.	1. God decides absolutely and gratuitously to give sufficient grace to every person he creates.
2. God then decides to give S a series of intrinsically efficacious graces to cause his free assent to God’s offer of salvation. Those not included in (1) are reprobate.	2. On the basis of his middle knowledge, God chooses those graces to which he knows S would freely respond, if he were given them. These graces are therefore efficacious for S. Those not included in (1) are reprobate.	2. On the basis of his middle knowledge, God knows whether S would respond if given sufficient grace. If so, then in creating S, God predestines S to glory, and his grace becomes efficacious. If not, then S is not predestined, and God’s grace remains merely sufficient.

That Thomas Aquinas held to a doctrine of unconditional election is beyond doubt.[17] Congruism, inspired by Suarez, preserves both libertarian freedom and unconditional election. Molinism preserves libertarian freedom, but advocates a conditional election that is grounded in what God foreknows his creatures would do if they were given prevenient grace. Thus, Craig writes:

In Congruism, I think we can clearly see how closely Arminianism and Calvinism can be brought by a doctrine of middle knowledge. For Lutheranism/Calvinism is (with respect to the issue at hand) simply a more consistent Thomism, and Congruism gives the Thomist everything he could desire in terms of God’s gratuitous and sovereign election and yet, unlike Thomism, consistently maintains human freedom. With Luther, one could affirm God’s infallible foreknowledge of future contingents and, with Calvin, God’s sovereign providence over the universe and yet not thereby sacrifice genuine human freedom. Middle knowledge does not entail Congruism, of course, and Arminians are not apt to go so far in affirming the gratuity of election and the efficacy of God’s gracious initiatives; but the point remains that by laying a common foundation of a doctrine of middle knowledge, Calvinists and Arminians could reduce the chasm that now separates them to the small divide that serves to distinguish Molina from Suarez, and this would be a monumental and laudable achievement.[18]

There are very few arguments in the modern literature which attempt to show that God possesses a middle knowledge of all counterfactual truths. It would seem that advocates of this theory have been content to offer middle knowledge as a coherent solution to the foreknowledge/freedom dilemma, having few or no reasons against it.19 However, Craig, to his credit, has offered an argument for the divine scientia media which, if sound, would establish the truth of Molina’s theory. His argument has been outlined as follows:

(1) If there are true counterfactuals of freedom, then God knows these truths. 

(2) There are true counterfactuals of creaturely freedom. 

(3) If God knows true counterfactuals of creaturely freedom, God knows them either logically prior to the divine creative decree or only logically posterior to the divine creative decree. 

(4) Counterfactuals of creaturely freedom cannot be known only logically posterior to the divine creative decree. 

From (1) and (2) it follows logically that 

(5) Therefore, God knows true counterfactuals of creaturely freedom. 

From (3) and (5) it follows logically that 

(6) Therefore, God knows true counterfactuals of creaturely freedom either logically prior to the divine creative decree or only logically posterior to the divine creative decree. 

And from (4) and (6) it follows that 

(7) Therefore, God knows true counterfactuals of creaturely freedom logically prior to the divine creative decree, which is the essence of the doctrine of middle knowledge.[20]

Is this argument sound? The burden of this article is to show not only that it is not a sound argument, but also that the doctrine of middle knowledge ought to be abandoned by anyone who wants to remain faithful to biblical theism.

What can be said in favor of these premises? Premise (1) is true in virtue of God’s omniscience. Omniscience entails that God knows all truths. And, thus, if there are true counterfactuals of creaturely freedom, then God must know them. Premise (3) is true in virtue of the principle of excluded middle—there is no third alternative between God’s knowing something either logically prior to his decree or (only) logically posterior to his decree. Since Craig is correct to say that premises (5), (6), and (7) follow from (1) through (4), it appears that his argument stands or falls on premises (2) and (4).

If for no other reason, premise (2) is unacceptable because the phrase “creaturely freedom” is ambiguous. Christian theologians and philosophers have traditionally opted for two conceptions of human free will: one that allows for a compatibility of freedom and some forms of necessity/certainty (compatibilism), and another that denies any compatibility between necessity and freedom (libertarianism). On the compatibilist model of freedom, one’s choices are determined by one’s strongest motive at the moment of choice. On the libertarian model of freedom, one’s choice is an uncaused intention, being determined by nothing at all.

Now, clearly the Molinist—as well as the Congruist—is not interested in defending a compatibilist model of freedom, for it is libertarian freedom that calls omniprescience into question in the first place.[21] So, in order to reconcile exhaustive divine foreknowledge with libertarianism, the Molinist does not need premise (2) but rather

(2)’ There are true counterfactuals of creaturely libertarian freedom.

But (2)’ runs into a number of problems, of which we shall name only two.[22] First, it is not at all obvious that a libertarian model of freedom is the best way to preserve the notion of human responsibility. To quote one Molinist in this regard: “Here, as in so many areas of philosophy, the evidence [for either compatibilism or libertarianism] is simply not conclusive, and the rationality of dissenting opinions ought to be acknowledged by even the most fervent proponents of either side.”[23] When we note that a contra-causal view of freedom, at least on the surface, leaves one bereft of a causal agent determining his choices out of the resources of his own character—making the movement of the will uncaused by all of what the agent is—it is hard to avoid the conclusion that libertarianism leaves one without a locus of praise or blame. Indeed, if all of what I am does not determine my choices, then all of what I am cannot be praised or blamed for those choices. In effect, I am not wholly responsible for my choices.[24]

But even if the Molinist can overcome this critique, we are confronted with a second major problem with (2)’, namely, that a counterfactual of freedom is, in the nature of the case, a necessary truth whose truth-value is not at all determined by the free will of the creature. And, since Molinism presupposes that the content of God’s middle knowledge is determined by the creature, the scientia media cannot reconcile itself to a libertarian model of free will. Hasker justifies this contention by first noting that, in order for the theory of middle knowledge to be complete, we must remember that the antecedents and conditionals God considers encompass everything that is relevant to the choice of the free creature. Consider Hasker’s example of David’s stay in Keilah:

(1) If David had stayed in Keilah, Saul would have besieged Keilah. In order to have some grasp of this sort of counterfactual, I suggest that we think in terms of initial-segment counterfactuals, in which the antecedent specifies a complete initial segment of a possible world up to a given point in time, and the consequent an event that may or may not take place at that time. (Of course, the antecedent will include any relevant causal laws that have held up until that time in that possible world.) If now we symbolize such counterfactuals using a capital letter followed by an asterisk to stand for the antecedent, then the initial-segment counterfactual corresponding to (1) would be 

[2] A* - Saul besieges Keilah, where “A*” represents a proposition specifying the entire initial segment of the possible world envisaged by God as the one in which Saul makes his decision. The contrary counterfactual then would be 

[3] A* - Saul does not besiege Keilah. If, as we have been assuming all along, (1) is true, then [2] also will be true, and [3] false.[25]

How can it be that [2] is a contingent truth rather than a necessary truth? It would appear that the only answer a Molinist can appeal to is “similarity.” That is to say, what renders [2] true and [3] false is that [2] describes a world that is more similar to the actual world than [3]. In other words, if the actual world (i.e., a world in which David leaves Keilah) were any different than it is, then [3] may very well have ended up being true and [2] false.

The problem with this solution is that with “regard to initial segment counterfactuals. .. comparitive similarity has no work left to do.”[26] Indeed, as we evaluate [2], should it make a difference whether the actual world is as it is or is different in certain ways from this one? “After all, if A* were actual, then neither ‘our’ actual world nor that other one would be actual—so why should the truth of [2] depend in any way on which of those worlds is actual as things stand?”[27] In other words, if A* were actual, it would be no more or less similar to the actual world than any other world where David stays in Keilah, for A* exhausts all worlds in which David stays in Keilah. Thus, Hasker is correct to say that, at the moment when Saul is getting ready to make his choice, there is nothing left for initial-segment counterfactuals to do.

The theory of middle knowledge is obliged to hold that some initial-segment counterfactuals are logically contingent. But in order to do this, the theory must apply to these counterfactuals the notion of comparative similarity to the actual world, and I have argued that this notion has no legitimate application here—which is to say, the notion is misapplied. The correct conclusion to be drawn from counterfactual logic, then, is that if initial-segment counterfactuals are true at all, they are true in all possible worlds and thus are necessarily true. But this conclusion is fatal to middle knowledge.[28]

Craig’s strategy in overcoming Hasker’s argument is to repeat the contention of Plantinga, namely, “that shared counterfactuals are themselves a measure of the similarity between worlds.”[29]

Thus, if some counterfactual is true in the actual world, there still are antecedent-permitting worlds which are farther from the actual world than the sphere of antecedent-permitting worlds in which the consequent is universally true, namely, those worlds in which the consequent is false. But those worlds may be closer to some other possible world; hence, in that world the counterfactual which is true in the actual world is false.[30]

But it seems to me that Craig has told us nothing interesting here, for he is merely repeating the point that Hasker has already refuted. To tell us that there are antecedent-permitting worlds which are farther removed from the actual world than those in which the consequent is universally true—that is, those in which the consequent is false—is to bring up once again the notion of similarity as an explanation of how a counterfactual of freedom is contingently true. But we have already seen that everything is accounted for in A* at the time Saul makes his decision, and A* cannot be any more or less similar to the actual world than it already is. Merely to assert that worlds in which A* - Saul besieging the city are more similar to the actual world than those worlds in which A* - Saul not besieging the city, thus explaining how the former is contingently true (at least as far as this world is concerned), is no way to overcome a criticism that has already taken such a response into account! Craig must do more than simply repeat the criterion of similarity. He must show us why such a criterion is even relevant to the situation at hand.

Craig may very well retort that he is not merely repeating himself here, but is instead saying that “similarity” allows us to say that some possible worlds are more similar to the actual world than others on account of the fact that such possible worlds share counterfactuals with the actual world while others do not. But as Hasker would respond, “similarity” was originally introduced into the equation in order “to secure that counterfactuals are evaluated in worlds sufficiently similar to the actual world in noncounterfactual respects.” Hasker then asks, “How can Plantinga [and, by extension, Craig] justify relying on the principles of counterfactual logic when at the same time [they undercut] the rationale for accepting those same principles?”[31]

Now if the similarity relation fails to account for a true counterfactual’s contingency, then, with respect to our David/Saul example, we have no reason to think that there is a possible world in which [3] obtains. Hence, [2] is true in all logically possible worlds in which A* obtains and, thus, is necessarily true.[32]

What’s the Molinist to do? If we are correct, then (2)’ is not a true proposition. And there seem to be only three options available to the biblical theist: (a) offer an open-view-theistic revision of (2)’; (b) deny that there are true counterfactuals of freedom; and/or (c) offer an Augustinian-Calvinistic revision of (2).

(a) Should we adopt an open-view revision of (2)’? Gregory Boyd believes so, writing that we should incorporate the concept of “might-counterfactuals” into our understanding of middle knowledge. He offers us the following argument:

While many classical theists have assumed that the very definition of omniscience rules out knowledge of might-counterfactuals—an omniscient being cannot know anything as a “maybe”—we see here that might-counterfactuals are actually logically implied in the definition of omniscience, if we grant that God knows would-counterfactuals. To know the truth value of the one is to know the truth value of the other. There is, then, no basis for restricting God’s middle knowledge to knowledge of would-counterfactuals and no necessary reason for the conclusion that all might-counterfactuals are false. This, I believe, is the basic assumption, and basic oversight, of classical Molinism.[33]

But there are at least three problems with Boyd’s “neo-Molinism.” First, no classical theist—Molinist or Thomist, Arminian or Calvinist—has ever said that God does not know genuine possibilities. For on all of these models of God, the divine cognition embraces all possibles via his natural knowledge.

Second, and following the first point, the idea of a “might-counterfactual” is simply bizzare. It is, indeed, superfluous, given the doctrine of natural knowledge. Boyd’s “neo-Molinism” is nothing of the kind, for it merely conflates middle knowledge into natural knowledge, which is what Thomists/Calvinists have been advocating all along!

Third, in order for Boyd’s open-view revision to work, it would have to be the case that God knows only some would-counterfactuals. But this leads to two more problems. (i) If even some would-counterfactuals are true, then they are necessarily so and have no grounds—which brings us once again to the question of whether middle knowledge reconciles prescience with libertarianism. (ii) What makes some would-counterfactuals true and others neither true nor false? If something makes some true, then it would have to be either God himself (i.e., his natural knowledge, thus making middle knowledge superfluous), or at least his decree (entailing Augustinianism), or the environment of the free agent (entailing hard-determinism), or the character of the agent (entailing compatibilism). But if nothing makes some true, then the fact that some would-counterfactuals are true (while others are neither true nor false) becomes an eternal surd Boyd so desperately wants to avoid.[34] Of course, this is all academic since, after all, to insist that some counterfactuals have no truth-value is to assume that they are not governed by the law of bivalence. But the universal applicability of bivalence is much more plausible than Boyd’s revision.[35] Then again, one could entertain the possibility that all would-counterfactuals are true, but it just so happens that God does not know them (for whatever reason). However, this would entail that there are potentially infinite counterfactual propositions unknown to God, which hardly constitutes a coherent account of divine omniscience. No, the real problem in all of this is Boyd’s open view of God, not Craig’s second premise.

(b) Should we insist that there are no true counterfactuals of freedom? It seems that the best approach someone like Boyd should take is to follow Hasker, who appears to suggest that there is no sense in which a counterfactual of freedom is true. Indeed, Hasker is a libertarian, so he cannot allow for a counterfactual of compatibilist freedom. And since counterfactuals of libertarian freedom are incoherent insofar as they destroy libertarianism, Hasker cannot allow for this either.

However, as Craig notes, this perspective is fairly radical, for it seems intuitively obvious that there are counterfactual propositions that are true. “Very little reflection is required to reveal how pervasive and indispensable a role such counterfactuals play in rational conduct and planning.”[36] And here we can agree, for it appears undeniable that some (and even many) counterfactual propositions are true, for example, had I studied harder, I would have been a better philosopher; had J. F. K. not been shot, America either would or would not have escalated the Vietnam conflict; had I failed to complete my doctoral dissertation, I would currently be unhappy; and so forth.

Second, it is plausible that the Law of Conditional Excluded Middle (LCEM) holds for counterfactuals of a certain special form, usually called “counterfactuals of creaturely freedom.” Counterfactuals of creaturely freedom are counterfactuals of the form, If S were in C, S would freely do A, where S is a created agent, A is some action, and C is a set of fully specified circumstances including the whole history of the world up until the time of S’s free action. According to LCEM for any counterfactual.[37] Molinists need not and should not endorse LCEM unqualifiedly. There is no reason to think, for example, that if Suarez were to have scratched his head on June 8, 1582, then either Freddoso would have scratched his head on June 8, 1982, or would not have scratched his head on June 8, 1982. But it is plausible that counterfactuals of the very specialized sort we are considering must be either true or false. For since circumstances C in which the free agent is placed are fully specified in the counterfactual’s antecedent, it seems that if the agent were placed in C and left free with respect to action A, then he must either do A or not do A. For what other alternative is there?[38]

The only alternative is to deny the universality of bivalence.[39] But since this is implausible, Craig seems to be correct.[40]

Finally, the Bible is replete with passages articulating counterfactual truths describing what a free agent would have done had a different state-of-affairs obtained. Perhaps Craig’s clearest example is 1 Cor 2:8: “None of the rulers of this age understood this; for if they had, they would not have crucified the Lord of glory” (NRSV).[41] It seems, then, that Craig has made his point. There are true counterfactuals of creaturely freedom.

Of course, in conceding the truth of (2), at least in some sense, we are not advocating Molinism. Indeed, we have already seen that (2), as it stands, is ambiguous, and (2)’ is incoherent, for the libertarian model of freedom is questionable (to say the least). And, if there are counterfactuals of freedom, there is some sense in which they are necessarily true—thus destroying the libertarian model of freedom.

(c) Should we accept an Augustinian revision of (2)? If there is some sense in which (2) is true, and if (2)’ is incoherent, and if open-view revisions of (2) or (2)’ cannot help us, then perhaps we ought to accept an Augustinian revision of (2). If so, then we need neither (2) nor (2)’, but rather

(2)” There are true counterfactuals of creaturely compatibilistic freedom.

Many Calvinistic writers have opted for this approach. On this revision of (2), the ground for any counterfactual of freedom is the nature or character of the free agent, that is, the character of the free agent determines the will of the agent. And, since there are several good reasons to reject libertarianism in favor of compatibilism,[42] this is not a problem.

But if an Augustinian admits that middle knowledge does not reconcile libertarianism with unlimited prescience, why accept it? Such Augustinians answer that the doctrine of middle knowledge enriches our understanding of providence. For it is via his middle knowledge that God providentially arranges which world will be actual. As Ware argues:

.. . if we really do make our choices for prevailing reasons, if the conditions (both internal and external) surrounding a particular choice present to us the individually necessary and jointly sufficient conditions for making just the choices we do, if choices and actions are actually effects of sufficient causal factors—if this is so, then it follows that God can know what choices would be made by knowing just exactly the set of conditions (i.e., all factors which together form the set of individually necessary and jointly sufficient conditions) that gives rise to particular choices and actions. So, he can envision an agent in one situation, and knowing all the factors true in that situation can know from these factors what choice the agent would make here, and he can envision a slightly different situation, and again, in knowing all the factors true to that situation he can know what the agent would do, instead, there.[43]

Tiessen concurs:

Without middle knowledge I cannot conceptualize God’s decision as either wise or cognizant of the freedom of his creatures. If God simply decided the future in one logical moment without regard to the possible responses of creatures to his own initiatives and wisest responses that he could make to those creaturely decisions, then any appearance of significance in those human decisions is thoroughly illusory.[44]

This proposal is not exactly the Congruism of Suarez, but it is close. On the one hand, Calvinists such as Ware, Tiessen, Feinberg, and Frame agree with Suarez that Molinism, when left to itself, compromises salvation by grace alone. For, according to Molina, God’s election of certain persons unto salvation is based on his foresight of human merit. In other words, Molina promoted salvation by works.[45] On the other hand, such Calvinists disagree with Suarez’s libertarianism, for, they insist, it cannot account for human responsibility.[46]

Shall we adopt the Calvinist-Middle-Knowledge View? Μὴ γένοιτο! Historically speaking, middle knowledge was never meant, in Craig’s words, to be a “rapproachement” between Calvinists and Arminians. As Craig himself suggests, no Arminian should embrace Congruism, for the idea of unconditional election is, from the Arminian point of view, one of the most repugnant ideas Calvinists have to offer.[47] It not only implies that some are not given a chance to be saved, but also suggests that God does not (savingly) love everyone. For most open theists, “love” has become the central divine attribute, the one upon which all others hang.[48] Hence, it is hard to imagine either classical Arminians or open-view theists consistently embracing Congruism.

But consistent Calvinists cannot embrace the scientia media either, if for no other reason than that too many Reformed distinctives would be lost in this scheme. We must remember that middle knowledge is the philosophical foundation of Arminianism.49 Hence, one can see why Reformed theologians have traditionally not wanted to embrace the scientia media. As Richard Muller explains:

For scientia media to become the basis for such a rapprochement. .. the Reformed [Calvinist] would need to concede virtually all of the issues in debate and adopt the Arminian perspective, because, in terms of the metaphysical foundations of the historical debate between Reformed and Arminian, the idea of a divine scientia media or middle knowledge is the heart and soul of the original Arminian position. Middle knowledge is not a middle ground. It was the Arminian, just as it was the Jesuit view, in the controversies over grace and predestination that took place in the late sixteenth and early seventeenth centuries.[50]

Therefore, we not only need to accept (2)” but we must also reject Craig’s fourth premise, namely,

Counterfactuals of creaturely freedom cannot be known only logically posterior to the divine creative decree.[51]

On the contrary, counterfactuals of creaturely freedom are known only posterior to the divine decree. In other words, God does not possess middle knowledge. Of course, if the only argument we could offer is the historical background to the debate, this contention would not go very far. But there are at least five theologico-philosophical considerations which have convinced many that God does not possess middle knowledge.

First, Molina’s biblical arguments for middle knowledge are inconclusive. Thereare, to be sure, several texts suggesting that God knows what would have happened had another state-of-affairs obtained (cf. 1 Sam 23:10–13; Matt 11:20–21).[52] But this proves only that God knows counterfactuals. Whether he possesses middle knowledge is still left open. And surely God’s awareness of counterfactuals is not, in and of itself, proof that he possesses middle knowledge. At best, the passages often used by defenders of middle knowledge prove, if anything, only “that God, knowing all causes, free and necessary, knows how they would act under any proposed condition. Even we know that if we add fire to powder an explosion would ensue.”[53] In other words, these biblical passages only show us that God knows the nature of the free agent so well that, were that agent placed in another circumstance, God knows exactly what he would do. And this does not reconcile middle knowledge with libertarianism, but with compatibilism. More importantly, it is certainly possible that God knows this information only logically posterior to the divine decree—thereby eliminating the need for middle knowledge. Therefore, there is little, if any, biblical warrant for the scientia media. Interestingly enough, Craig admits this point, writing:

Since Scripture does not reflect upon this question, no amount of proof-texting can prove that God’s counterfactual knowledge is possessed logically prior to his creative decree. This is a matter for theological-philosophical reflection, not biblical exegesis. Thus, while it is clearly unbiblical to deny that God has simple foreknowledge and even counterfactual knowledge, those who deny middle knowledge cannot be accused of being unbiblical.[54]

Second, middle knowledge compromises the self-existence of God. When we say that God is self-existent, we mean to say all of the following: that he is pure actuality (there is no potentiality [or passive potency] in his being); that he is wholly independent and uncaused; that he is his act of existing; and so forth.

However, if God possesses middle knowledge, then his knowledge is somehow dependent upon the creature—that is, God knows what I would do in a particular situation because I would do it in that situation. Indeed, every doctrine of middle knowledge one encounters in the literature implies this. Tiessen, as we have seen, tells us that God must take into account what the free (compatibilist) creature will do in order to make the wisest decision he possibly can. Ware’s comments are essentially the same.[55] Molinists openly admit that the free (libertarian) creature determines which counterfactuals of freedom are true.[56] Thus, Craig affirms that it “is up to God whether we find ourselves in a world in which we are predestined, but it is up to us whether we are predestined in the world in which we find ourselves.”[57] In short, the content of the divine middle knowledge is determined by the creature. And this immediately introduces passive potency into God’s knowledge, for it makes God (or, at least, some aspect of him) something less than absolutely independent. Thus, it metaphysically reduces God to the same level as the creature, for he is now seen to be a mixture of act and potency. God is now seen as contingent, therefore he needs a cause. As the great neo-Thomist Garrigou-Lagrange once put the issue: “God determining or determined; there is no other alternative.”

It is because in God alone essence and existence are identical, because He alone is Being itself, that we must conclude that only in Him can there be no accident, that He alone is infinite, that nothing that is external to Him can exist unless it has been created and preserved in being by Him, that nothing external to Him can act without the divine motion. Action, in fact, presupposes being, and the mode of the action corresponds to the mode of the being that is in action. God alone, who is His existence, who is Being itself, is consequently action itself, intellection itself, love that is itself eternally subsisting. On the other hand, no creature, however perfect it may be, since it is not its existence, is not its own thought and will; but the most perfect angel, just as the least endowed human soul, always is in need of the divine motion in order to think and will anything whatever. Nothing, consequently, escapes the divine motion except evil, which, being a defect, presupposes only a deficient cause. It cannot come from God, but is permitted by Him, because He is powerful enough and good enough to draw from it a greater good that is known to Him, a glimpse of which is at times given to us.[58]

Amazingly, Craig admits the point that middle knowledge compromises God’s pure actuality:

Despite Molinist protests, I think we will have to admit that this is true. But at the same time. .. this seems to me of no great consequence. As I argued earlier, God’s simple foreknowledge can be understood as determined in its content by what will in fact occur. This sort of determinacy or passivity on God’s part seems to me altogether innocuous, and if this sacrifices the Thomistic view of God as Pure Actuality, then so be it. The Thomistic view, that God determines not only the truth of future contingent propositions, but also the truth of counterfactuals of freedom, seems to lead inescapably to making God the author of sin and to a denial of human freedom and responsibility in general.[59]

Perhaps the most important objection to God’s pure actuality is that the doctrine is not taught in Scripture. And, to be sure, the Bible does not say, in so many words, that God is actus purus. Nevertheless, Scripture makes several statements which seem to entail the very notion of divine self-existence implicitly (or explicitly) denied by defenders of middle knowledge. According to the Bible, God has life in and of himself ( John 1:1–4; 5:26). “The God who made the world and everything in it, he who is Lord of heaven and earth, does not live in shrines made with human hands, nor is he served by human hands, as though he needed anything, since he himself gives to all mortals life and breath and all things.... ‘For in him we live and move and have our being”‘ (Acts 17:24–25, 28; NRSV).[60] Our basic point here is bolstered when we note that, according to Isaiah, God’s knowledge is in no way based upon the creature. He gets his knowledge from himself. He does not receive it from anything else. Indeed, has anyone measured the waters of the sea in the palm of his hand? Has anyone ever measured the heavens or the earth? Has anyone weighed the mountains? God has done all these things (Isa 40:12, 15–17). Likewise, no one can say that the Lord is dependent upon the creature for anything, including his knowledge:

Who has directed the spirit of the Lord,
or as his counselor has instructed him?
Whom did he consult for his enlightenment,
and who taught him the path of justice?
Who taught him knowledge,
and showed him the way of understanding? (Isa 40:13–14 NRSV)

These are obviously rhetorical questions. Of course, no one has taught God anything. Indeed, “All the nations are as nothing before him; they are accounted by him as less than nothing and emptiness” (Isa 40:17). All knowledge, all wisdom, and all understanding have their source in God alone. Hence, not only is God’s nature self-existent, but his knowledge is from himself (ase), being wholly independent of the creature.[61] We therefore agree with an Arminian exegete when he writes that the point of vv. 13 and 14 “.. . is surely to deny the existence of any counselor, whether human or heavenly. The answer to the rhetorical questions in both vv. 13 and 14 is that no one has advised God either in the creation or in the administration of the world.”[62] Indeed, the biblical tradition generally portrays God as the source of all things, including wisdom and knowledge ( Job 28:1-28; 35:9–11; 38:2–39:30; Ps 36:9; Prov 2:6; 8:22–31; Jas 1:5; Rom 11:33–36; Col 2:3). Wisdom flows from God, not to him. Or, in the words of Arminius, “God knows all things, neither by intelligible [species] representations, nor by similitude, but by his own and sole essence....”[63]

Several more responses to Craig are in order. First, a close analysis of the traditional Augustinian-Calvinistic views regarding the divine decree suggests that such a doctrine does not make God the author of sin.[64] Second, as we have seen, the libertarian model of human freedom is internally flawed, independent of any considerations concerning divine foreknowledge. Third, one wonders how the Molinist escapes his own dilemma. Does not the God of the Molinist scheme predetermine that this world—and all the evil contained therein— obtain? Of course! The Molinist may then respond, “But this world merely exemplifies those counterfactual (or subjunctive conditional) truths that are not grounded in either God or his decree.” True. But it is unclear how this solves anything. God may not be the grounds of any counterfactual, but he is the one who causes it to be exemplified. Hence, he is no more (or less) responsible for evil appearing in our world than the God of the Augustinian tradition. To quote Paul Helm:

Craig says that the Augustinian-Calvinist perspective (which, as I said earlier, I think he takes to be equivalent to theological fatalism and to fatalism simpliciter) holds that “foreknowledge is based upon foreordination: God knows what will happen because he makes it happen” (p. 135). Three things on this. First, taking this claim at face value, it is not a doctrine that the ancient Greek fatalists would have recognized. Second, these expressions are inaccurate. God’s foreordaining x is not equivalent to God’s making x happen. For according to the Augustinian-Calvinist perspective, God ordains evil by willingly permitting it. Third, on the question of the authorship of evil, there’s not a hairsbreadth between the Augustinian-Calvinist perspective and Craig’s Molinism. According to Craig’s description of Molinism, “God decreed to create just those circumstances and just those people who would freely do what God willed to happen” (p. 134). While this description does not entail that God is the author of evil (any more than the Augustinian-Calvinist perspective does), it does entail that God decreed all sinful acts to happen and decreed them precisely as they have happened. If this is so, the God of Molina and Arminius seems to be as implicated in the fact of evil as much (or as little) as the God of the Augustinian-Calvinist perspective.[65]

Hence, Craig has given us no reason to think God is not wholly self-existent. And a wholly self-existent deity cannot have middle knowledge.

Third, middle knowledge seems to entail an (environmental) determinism of the circumstances. Concerning, for example, Peter’s choice to deny Christ, proponents of middle knowledge will insist that God, knowing those circumstances in which Peter would (if placed in them) choose to deny Christ, placed Peter in just those circumstances. But we are left asking, “What is it about those circumstances which influenced Peter to make his choice?” If the answer is, “Nothing about the circumstances influenced Peter,” then we are left asking about the relevance of saying, “God knew under which circumstances Peter would choose to deny Christ.” If the answer is, “There is something about the circumstances which influenced Peter to make his choice,” then the circumstances become the determining factor in Peter’s decision, not Peter himself.[66] But this is to embrace an environmental determinism which destroys every plausible account of human freedom and responsibility (compatibilist or libertarian). As Garrigou-Lagrange writes:

If it is maintained that before any determining divine decree (positive or permissive), God foresees infallibly such a conditional free act of the future by reason of the virtual priority of truth over goodness, one falls back into fatalism or determinism of the circumstances. For, after all, according to the hypothesis, this free act of the future is determined neither by the divine decree nor in the created will which is free or indifferent. For it to be foreseen infallibly and not merely conjecturally, it must therefore be determined by the circumstances.[67]

Hence, “...the scientia media, devised to safeguard liberty, destroys it.”[68]

Fourth, the only way to maintain consistently the sovereignty of God is to deny middle knowledge. We earlier noted that middle knowledge compromises God’s pure actuality. In so doing, it also compromises his sovereignty. Proponents of middle knowledge will protest, insisting that God is truly sovereign over which hypothetical circumstances (i.e., counterfactuals/subjunctive conditionals) obtain in this world. But as Turretin explains, this answer leads to an incoherence in the theory of middle knowledge—at least insofar as it denies the very sovereignty it wants to secure:

This middle knowledge takes away from the dominion of God over free acts because according to it the acts of the will are supposed to be antecedent to the decree and therefore have their futurition not from God, but from itself. Indeed God would seem rather to depend upon the creature while he could decree or dispose nothing, unless a determination of the human will were posited which God would see in such a connection of things. Nor ought the reply to be made that the dominion of God is not therefore taken away because he can remove that connection or some circumstances of it; for example, in the foreknowledge by which God knew that Peter would deny Christ if placed in a certain condition, God could hinder him from denying Christ by taking away some foreseen circumstance (for instance, the fear of death) or by adding greater light in the intellect and a greater inclination in the will to confession, and the like. For it is not sufficient for the support of the dominion of God that he could hinder Peter from denying Christ, for he might have deprived Peter of life before the apprehension of Christ (but this would be to have dominion over the life of Peter, not over his free will); but it is requisite that the free acts of Peter, of denying or not denying Christ, should depend upon him (which is denied on the supposition of this knowledge). In fine, if God can take away one foreseen circumstance, he can therefore change the event of the thing: if he can by a decree change the event of a thing, therefore it also pertains to the decree to procure it; for he who hinders the event by a removal of some circumstance ought to cause it by supplying the circumstances.[69]

Turretin’spointhereisthatGodmustbefactoredintotheequationwhencontemplating whether there are true counterfactuals of freedom; for at every moment of every hypothetical state-of-affairs, the question can still be raised, “Is God going to permit this?” Hence, the truth of any counterfactual must be dependent upon God, not the will of the creature—much less the circumstances in which the creature is placed.

Fifth, the doctrine of middle knowledge is superfluous. Most theologians can see why a distinction between natural and free knowledge is necessary, namely, if God does not have natural knowledge, then, among other things, he had no choice but to create this world, since his knowledge is limited to what goes on in this world (thus destroying the doctrine of free creation); and if he does not have free knowledge, then God is not sovereign over which world obtains. But middle knowledge is simply unnecessary, since, as Turretin maintains,

[n]atural and free knowledge embrace all knowable things and entities and are not to be multiplied unnecessarily. There is nothing in the nature of things which is not possible or future; nor can future conditional things constitute a third order. For they are such either from a condition only possible or powerful, yet never to take place, or from a condition certainly future and decreed. In the former manner, they do not recede from the nature of possible things and belong to natural knowledge; in the latter, they are future and decreed by God and come under the free knowledge.[70]

To speak in Boyd’s terms, all would-counterfactuals are simply a species of might-counterfactuals. And it is certainly feasible to think of God predetermining which world will obtain in light of his natural knowledge. Therefore, contrary to Molina, Tiessen, and Ware, middle knowledge reduces to natural and, hence, is simply not needed for a robust doctrine of providence.

Conclusion

In this article we analyzed what is, perhaps, the most prominent method of reconciling divine foreknowledge with libertarian freedom. We have seen that, even on the doctrine of middle knowledge, humans do not possess libertarian freedom. More importantly, middle knowledge seems to compromise the very nature of the God described in the Bible. Therefore, not only a consistent Calvinist, but a consistent Christian must reject the doctrine of middle knowledge.[71]

Notes

1. Besides the many articles Craig has written on this subject, he has contributed to the modern discussion first, by giving us a very helpful historical overview of the problem in The Problem of Divine Foreknowledge and Future Contingents from Aristotle to Suarez (Brill’s Studies in Intellectual History 7; New York: Brill, 1988). Secondly, he has given us his own insights with regard to the foreknowledge/ freedom dilemma in both a scholarly and popular work in Craig, Divine Foreknowledge and Human Freedom: The Coherence of Theism: Omniscience (Brill’s Studies in Intellectual History 19; New York: Brill, 1991) (henceforth, DFHF ); and Craig, The Only Wise God: The Compatibility of Divine Foreknowledge and Human Freedom (Grand Rapids: Baker, 1987).
2. Alfred Freddoso, introduction to On Divine Foreknowledge by Luis de Molina (Part 4 of Concordia; trans. Alfred J. Freddoso; Ithaca, N.Y.: Cornell University Press, 1988), 1–81.
3. The Possibility of an All-Knowing God (New York: St. Martin’s Press, 1986).
4. Divine Providence: The Molinist Account (Cornell Studies in the Philosophy of Religion; Ithaca, N.Y.: Cornell University Press, 1998).
5. The Nature of Necessity (Oxford: Clarendon Press, 1974), ch. 9, esp. his defense of counterfactuals. Plantinga apparently articulated his arguments for the truth of counterfactuals without any knowledge of Molinism; cf. “Self-Profile,” in Alvin Plantinga (ed. James Tomberlin and Peter Van Inwagan; Boston: Reidel, 1985), 50.
6. Liberi Arbitrii cum Gratiae Donis, Divinia Praescientia, Providentia, Praedestinatione et Reprobatione Concordia, 4.2; as quoted in Freddoso, “Introduction,” 24–25. All quotations from Molina’s Concordia are taken from Freddoso’s translation.
7. See Gregory A. Boyd, God of the Possible: A Biblical Introduction to the Open View of God (Grand Rapids: Baker, 2000); and William Hasker, God, Time, and Knowledge (Cornell Studies in the Philosophy of Religion; Ithaca, N.Y.: Cornell University Press, 1989).
8. Omni = all; pre = before; science (scientia) = knowledge; i.e., knowing all things beforehand.
9. One will find a similar diagram in William Lane Craig, What Does God Know? Reconciling Divine Foreknowledge and Human Freedom (Norcross, Ga.: Ravi Zacharias International Ministries, 2002), a popular booklet expounding Molinism.
10. These are called logical moments since an atemporal God cannot have successive temporal experiences. And, even if God were omnitemporal, there would never be a time when he did not know these truths.
11. Hasker, God, Time, and Knowledge, 25.
12. To be sure, it is counterfactuals of libertarian freedom that are the issue in this debate.
13. Craig, DFHF, 246–69.
14. William Lane Craig, “Middle Knowledge: A Calvinist-Arminian Rapprochement?,” in The Grace of God and the Will of Man (ed. Clark H. Pinnock; Minneapolis: Bethany House Publishers, 1989), 161.
15. Cf. Craig, Aristotle to Suarez, ch. 8.
16. Craig, “Middle Knowledge,” 161; Craig’s synopsis is taken, with slight revision, from de Regnon’s Banez et Molina (Paris: Oudin, 1883), 157–58.
17. Cf. Summa Theologica, Ia.23.
18. Craig, “Middle Knowledge,” 161.
19. To be sure, Flint argues for Molinism indirectly, i.e., he tries to demonstrate that the alternatives to Molinism are implausible, leaving Molina’s theory as the only game in town for those who want to maintain a doctrine of meticulous providence and libertarian freedom (Flint, Divine Providence, esp. ch. 4).
20. William Lane Craig, “The Middle-Knowledge View,” in Divine Foreknowledge: Four Views (ed. James K. Beilby and Paul R. Eddy; Downers Grove, Ill.: InterVarsity Press, 2001), 136–37.
21. In support of this I simply appeal to Hasker, who tells us that his work on divine foreknowledge and human free will largely ignores the issue of compatibilist freedom. Indeed, “.. . if one takes a compatibilist view of free will, most of the problems considered here are rather readily resolved...” (God, Time, and Knowledge, viii).
22. For a more thorough critique of (2)’, see Travis J. Campbell, “The Beautiful Mind: A Reaffirmation and Reconstruction of the Classical Reformed Doctrines of the Divine Omniscience, Prescience, and Human Freedom” (Ph.D. diss., Westminster Theological Seminary, 2004), esp. chs. 7–8, 10.
23. Flint, Divine Providence, 26.
24. Cf. Campbell, “Beautiful Mind,” chs. 7–8; Jonathan Edwards, Freedom of the Will (vol. 1 of The Works of Jonathan Edwards; ed. Paul Ramsey; New Haven, Conn.: Yale University Press, 1957), passim.
25. Hasker, God, Time, and Knowledge, 32–33 (italics in original).
26. Ibid., 34.
27. Ibid., 35 (italics in original).
28. Ibid. (italics in original)
29. William Lane Craig, “Hasker on Divine Knowledge,” 10. This essay originally appeared in Philosophical Studies 62 (1992): 57-78. All quotations of this essay are taken from http://www.origins. org. See also, Plantinga, Nature of Necessity, 177–78.
30. Craig, “Hasker on Divine Knowledge,” 10.
31. Hasker, God, Time, and Knowledge, 35–36 (italics in original). Craig responds that, if evaluating worlds sufficiently similar to the actual world in noncounterfactual respects is the motivation for introducing the notion of comparative-similarity in the first place, “it only follows that the motives of those who drafted possible worlds semantics for counterfactuals were thwarted” (“Hasker on Divine Knowledge,” 10). But if we are to evaluate a counterfactual proposition’s truth-value by comparing its similarity, or lack thereof, with the actual world, then does it not stand to reason that such an evaluation must be based upon noncounterfactual considerations? The actual world is, after all, actual—not counter-to-fact—and, thus, any similarity that exists between the actual world and any counterfactual is to be found in the way things actually are, not in what is contrary to fact (or counterfactual). Thus, if the motives of those who drafted possible worlds semantics for counterfactuals were thwarted, we cannot help but conclude that the notion of counterfactuals of freedom would be thwarted as well.
32. Plantinga has asserted that “... we can’t sensibly explain necessity as truth in all possible worlds” (“A Reply to Robert Adams,” in Alvin Plantinga, 378). But he does not explain why. And what else makes a truth necessary, or at least logically necessary, if not the virtue of being true in all logically possible worlds? In fact, Plantinga argues elsewhere that God is a necessary being in virtue of the fact that he, a being of maximal excellence, is instantiated in every possible world—and since the actual world is also possible, it follows that a being of maximal excellence is instantiated in this world (Nature of Necessity, 213–17). If I understand him correctly, Craig’s new insights into this issue simply rehearse this discussion, for it seems that his only way to avoid the famous “grounding objection” philosophers have given— with regard to counterfactuals of libertarian freedom—is to repeat this similarity criterion. See William Lane Craig, “Middle Knowledge, Truth-Makers, and the ‘Grounding Objection,”‘ Faith and Philosophy 18 ( July 2001): 337-52, esp. 347; cf. Campbell, “Beautiful Mind,” 283–87.
33. Gregory A. Boyd, “Neo-Molinism and the Infinite Intelligence of God,” Philosophia Christi 5 (2003): 192; cf. Boyd, Satan and the Problem of Evil: Constructing a Trinitarian Warfare Theodicy (Downers Grove, Ill.: InterVarsity Press, 2001), 123ff.; Boyd, “An Open-Theism Response,” in Divine Foreknowledge: Four Views, 144–48.
34. Cf. Boyd, “Neo-Molinism and the Infinite Intelligence of God,” 190.
35. See n. 39 below.
36. Craig, “Middle Knowledge, Truth-Makers, and the ‘Grounding Objection,”‘ 338.
37. This seems to be one of Suarez’s contributions to Molina’s doctrine of middle knowledge. Cf. Dean A. Kowalski, “On Behalf of Suarezian Middle Knowledge,” Philosophia Christi 5 (2003): 21927; Charles F. Kielkopf, “Suarezian Middle Knowledge: A Response to Dean A. Kowalski,” Philosophia Christi 5 (2003): 229-32; and Kowalski, “Kielkopf’s Compromise: A Reply to Charles F. Kielkopf,” Philosophia Christi 5 (2003): 233-34.
38. Craig, “Middle Knowledge, Truth-Makers, and the ‘Grounding Objection,”‘ 338.
39. I.e., “all propositions are either true or false.” This principle of bivalence seems to be entailed by the law of excluded middle, discussed in Craig’s paragraph. If bivalence is a universally applicable principle, then the proposition “all counterfactual propositions are either true or false” automatically follows. And, since all counterfactual propositions have a truth-value, it seems more plausibly true than not that there are at least some true counterfactuals. Cf. Campbell, “Beautiful Mind,” 76ff.
40. Campbell, “Beautiful Mind,” 82–86.
41. Craig, “Middle Knowledge, Truth-Makers, and the ‘Grounding Objection,”‘ 338–39.
42. Cf. Edwards, Freedom of the Will, passim.
43. Bruce A. Ware, God’s Greater Glory: The Exalted God of Scripture and the Christian Faith (Wheaton, Ill.: Crossway Books, 2004), 27–28.
44. Terrance Tiessen, Providence and Prayer: How Does God Work in the World? (Downers Grove, Ill.: InterVarsity Press, 2000), 319. Cf. John S. Feinberg, No One Like Him: The Doctrine of God (Founda¬tions of Evangelical Theology; Wheaton, Ill.: Crossway Books, 2001), 747–52. Ware lists Frame as another Calvinist who endorses middle knowledge (God’s Greater Glory, 27). But when one actually turns to Frame’s comments on this issue, one encounters some ambiguity. On the one hand, we read: “God’s will is formulated according to knowledge, including his foreknowledge of creatures; but his knowledge is also dependent upon the decisions of his will.... The problem with the Arminian, then, is not so much what he affirms, but what he denies.” These quotations imply something similar to the “Calvinist-Middle-Knowledge-View” advocated by Ware and Tiessen. However, Frame later qualifies some of this by saying: “If we abandon libertarianism, we abandon the traditional meaning of middle knowledge, and then, as I said earlier, there is no reason to dis¬tinguish God’s knowledge of contingencies from his necessary knowledge of himself.” This seems to suggest a conflation of necessary and middle knowledge. Perhaps Frame, in the end, tips his hand when he notes that “Arminians say that God’s foreordination is based on his foreknowledge. The Calvinist need not deny that this is the case. But he should go on, then, and point out to the Arminian that that foreknowledge itself is in turn based upon foreordination!” (see John M. Frame, The Doctrine of God: A Theology of Lordship [Phillipsburg, N.J.: Presbyterian & Reformed, 2002], 151; 505; 150). This last comment would suggest that Ware has correctly interpreted Frame, i.e., Frame has embraced the “Calvinist-Middle-Knowledge-View.” In response to Frame’s comments, we are compelled to make two important points. First, the Arminian/Molinist need not deny that foreknowledge is based on foreordination once the distinction between middle and free knowledge is grasped. For on this model, foreordination is based on foreknowledge (under the rubric “middle knowledge”), and foreknowl¬edge (under the rubric “free knowledge”) is based on foreordination. Hence, if we follow Frame’s line of reasoning, the Calvinist can contribute nothing to the discussion since the Arminian already has in place the very qualification Frame wants to secure! Second, and more importantly, it seems that no Presbyterian (e.g., Frame) should accept the doctrine of middle knowledge, for in the Westminster Confession of Faith we read: “Although God knows whatsoever may or can come to pass upon all supposed conditions, yet hath He not decreed any thing because He foresaw it as future, or as that which would come to pass upon such conditions” (3.2). Since the Confession denies that God’s decree is based upon his prescience in any sense, one must resist Frame’s suggestion that God’s foreordination is grounded in his foreknowledge.
45. Craig himself would agree with this, writing: “If one adopts the view that our good works merit salvation, then Molinism would, indeed, seem to lead to works-righteousness on the part of man, since it is not God who does the works, but we ourselves, at God’s instigation. On the Thomist view, the works are viewed as works of God to which He causally determines us, and therefore salvation by human works does not result. But then it is difficult to see either how man is genuinely free with regard to such works or how the merit wrought by such works can be attributed to man rather than to God. The conclusion follows, then, that the Molinist who wishes to remain a Molinist ought to reject the Catholic doctrine of justification in favor of a Protestant understanding of salvation as a wholly unmerited and freely accorded gift of God’s grace” (DFHF, 272). It is interesting to note that the Roman Catholic Church has not officially declared which view, Thomism or Molinism, is the orthodox understanding of grace. But as the Council of Orange decreed, and as Thomas Aquinas confirmed, we are made members “of Christ through grace alone” (Summa Theol. IIIa.62.1; cf. Eph 2:8–9; Rom 11:6). Cf. Ludwig Ott, The Fundamentals of Catholic Dogma (4th ed.; trans. Patrick Lynch; ed. James Canon Bastible; Rockford, Ill.: Tan Books & Publishers, 1974), 42–43. Note also how close some Catholics come to affirming, with Molina, election according to human merit (thus showing the plausible historical [if not logical] connection between rejecting sola fide and rejecting sola gratia [cf. Rom 3:28; 4:1–5; 5:1]); cf. Robert A. Sungenis, Not By Faith Alone: The Biblical Evidence for the Catholic Doctrine of Justification (Santa Barbara, Calif.: Queenship Publishing, 1997), 670. Thus, Craig seems right to say that, in order to preserve the orthodox doctrine of sola gratia, “a Christian would therefore seem compelled to choose either Thomism or Protestantism” (DFHF, 332 n. 51).
46. Cf. Ware, God’s Greater Glory, 85ff. Suarez would seem to offer a libertarian twist on Luther’s (compatibilist) dictum that we are free in the “things below” but our will is bound to sin in the “things above”—i.e., we have libertarian freedom to choose, say, what we will wear each day, but we do not have such freedom to choose to trust in Christ. Hence, middle knowledge is helpful in understanding divine providence in general, but in order to rightly apprehend salvation we must turn to Augustine’s doctrine(s) of unconditional election and effectual calling. To his credit, Craig is at least open to Suarez’s suggestion (cf. DFHF, 271).
47. To quote Arminius: “This also is a horrible affirmation, ‘Some among men have been created unto life eternal, and others unto death eternal”‘ (James Arminius, Certain Articles to Be Diligently Examined and Weighed, in The Works of James Arminius [trans. James Nichols and William Nichols; 3 vols.; Grand Rapids: Baker, 1986], 2:710).
48. Cf. Clark H. Pinnock and Robert C. Brow, Unbounded Love: A Good News Theology for the 21st Century (Downers Grove, Ill.: InterVarsity Press, 1994), passim.
49. Note the following words of Arminius: The Schoolmen therefore denominate the First [type of knowledge] “natural,” and the Second “free knowledge.” The knowledge by which God knows anything [si hoc sit] if it be or exist, is [media] intermediate between the two [kinds of knowledge] described. .. [above]: In fact it precedes the free act of the will with regard to intelligence. But it knows something future according to vision, only through its hypothesis. Free knowledge, or that of vision, which is also called “Prescience,” is not the cause of things: But the knowledge which is practical and of simple intelligence, and which is denominated “natural” or “necessary,” is the cause of all things by the mode of prescribing and directing, to which is added the action of the will and of the capability. The middle or intermediate kind of knowledge ought to intervene in things which depend on the liberty of created [arbitrii] choice or pleasure. (Private Disputations, in Works, 2:342 [italics in original])
50. Richard A. Muller, “Grace, Election and Contingent Choice: Arminius’s Gambit and the Reformed Response,” in The Grace of God, The Bondage of the Will (ed. Thomas R. Schreiner and Bruce Ware; Grand Rapids: Baker, 1995), 265–66. Cf. Muller, God, Creation, and Providence in the Thought of Jacob Arminius (Grand Rapids: Baker, 1991); and Eef Dekker, “Was Arminius a Molinist?,” Sixteenth Century Journal 28 (1996): 337-52. In a recent lecture on the doctrine of middle knowledge, Craig has conceded this point; to obtain a copy of his lecture “What Does God Know?,” go to www.campuscrusade.com/william_craig.htm.
51. Which entails “(7) Therefore, God knows true counterfactuals of creaturely freedom logically prior to the divine creative decree.”
52. For an apocryphal reference used by Molina, see Wis 4:10–14. For Molina’s use of these texts, see Campbell, “Beautiful Mind,” 257ff.
53. A. A. Hodge, Outlines of Theology (Carlisle, Pa.: Banner of Truth, 1972), 148 (italics in original).
54. Craig, “Middle-Knowledge View,” 125.
55. Ware’s comments are strange, to be sure, since he (inconsistently) affirms the self-existence of God elsewhere (cf. God’s Greater Glory, 44–51).
56. Again, our critique of (2)’ above has rendered this Molinist contention unlikely.
57. William Lane Craig, “ ‘No Other Name’: A Middle Knowledge Perspective on the Exclusivity of Salvation through Christ,” Faith and Philosophy 6 (April 1989): 179. Thus, Craig later writes: “Hence, there will be an infinite number of worlds known to God by His natural knowledge which are not realizable by Him because the counterfactuals of creaturely freedom which must be true in order for Him to weakly actualize such worlds are in fact false. His middle knowledge serves to delimit, so to speak, the range of logically possible worlds to those which are feasible for Him to actualize. This might be thought to impugn divine omnipotence, but in fact such a restriction poses no non-logical limit to God’s power” (ibid., 180). Notice that middle knowledge only recognizes which worlds are feasible for God to make. The scientia media is not sovereign over these worlds. Hence, the creature, not God, determines—at least in some sense—the content of the divine mind. Notice that these feasible worlds are purely hypothetical. Thus, we have the purely hypothetical determining, so to speak, what God can and cannot do. Cornelius Van Til was, therefore, right to say that the Arminian system(s) of theology is/are built on the assumption that “bare possibility precedes actuality.” But, he continues, “The biblical position holds God to be the source of all possibility” (A Survey of Christian Epistemology [vol. 2 of In Defense of the Faith; Phillipsburg, N.J.: Presbyterian & Reformed, n.d.], 88).
58. Reginald Garrigou-Lagrange, God: His Existence and Nature: A Thomistic Solution to Certain Agnostic Antinomies (trans. Dom Bede Rose; 2 vols.; 5th ed.; St. Louis, Mo.: Herder, 1946), 2:557–58 (italics in original). A Reformed theologian makes a similar point: on the theory of middle knowledge “God. .. becomes dependent on the world, derives knowledge from the world that he did not have and could not obtain from himself, and hence, in his knowledge, ceases to be one, simple, and independent—that is, God” (Herman Bavinck, God and Creation [vol. 2 of Reformed Dogmatics; ed. John Bolt; trans. John Vriend; Grand Rapids: Baker Academic, 2004], 201).
59. Craig, DFHF, 272–73.
60. Commenting on these verses F. F. Bruce, The Book of Acts (rev. ed.; NICNT; Grand Rapids: Eerdmans, 1988), 337, writes: “The God who created all could not be envisaged as requiring anything from his creatures. If he is pleased to accept their service, it is not because he lacks something which they can supply. Here again parallels to Paul’s argument can be adduced from Greek literature and philosophy. But the great prophets of Israel also had to refute the false notion that God is somehow dependent on his people’s worship and service, when they saw how many of their fellow Israelites were devoted to it. How can the Lord of heaven need anything that his creatures can give him?. .. [cf. Ps 50:9–12]. Far from being able to supply any need of his, it is he who supplies every need of theirs: to them all he gives ‘life, breath, and everything.’” Also, Joseph Fitzmyer provides a similar insight in The Acts of the Apostles (AB 31; New York: Doubleday, 1998), 608: “The God of whom Paul speaks is not just the creator of all but lacks nothing that human beings can supply. Paul echoes a motif common to the OT (Ps 50:9–12; Amos 5:12–23; 2 Macc 14:35) and to Greek philosophy (Aristobulus, frg. 4; cf. Eusebius, Praep. ev. 13.12.3); Euripides, Herc. fur. 1345–46: ‘For God, if indeed God he be, is in need of nothing.’.. . Though God needs nothing and takes nothing from human beings, he is the one who sustains all life, human, animal, and plant. God is the creator of the human race; human beings are not autochthonous, and Athenians are not sprung from Attic soil.” Not only does this passage of Scripture speak of the divine aseity, but it has at least one important methodological implication, namely, the fact that an idea inspired by Greek philosophy in no way suggests that it is unbiblical! Thus, the common open theistic argument that classical theism gets most of its ideas from Greek philosophy, even if true, is irrelevant to the issue at hand (i.e., is God wholly a se?).
61. Stephen Charnock, Discourses upon the Existence and Attributes of God (2 vols.; Grand Rapids: Baker, 1979), 1:456–58.
62. John N. Oswalt, The Book of Isaiah: Chapters 40–66 (NICOT; Grand Rapids: Eerdmans, 1998), 60.
63. Arminius, Private Disputations, in Works of Arminius, 2:341. It would seem, then, that the Arminian must make a choice, namely, reject his (biblical) intuition that God is wholly self-existent or reject middle knowledge. Arminius’s insight is also devastating to open theism, since the open-view of God (which rejects unlimited prescience) presupposes that the divine cognition is dependent upon the creature. For a full critique of open theism, see Campbell, “Beautiful Mind,” esp. chs. 1–6.
64. See Campbell, “Beautiful Mind,” ch. 6.
65. Paul Helm, “An Augustinian-Calvinist Response,” in Divine Foreknowledge: Four Views, 158–59 (responding to Craig’s contribution, “The Middle-Knowledge View,” in the same volume).
66. If the Molinist or Suarezian insists that, at the moment of choice, Peter could, if he had wanted to, chosen otherwise, we respond that all compatibilists agree. In fact, this sort of response vindicates the compatibilist view of freedom!
67. Garrigou-Lagrange, God: His Existence and Nature, 2:484–85 (italics in original).
68. Ibid., 486. For Garrigou-Lagrange’s full critique of the divine scientia media, see ibid., 68-92, 151-58, Appendix 4; and Garrigou-Lagrange, The One God: A Commentary on the First Part of St. Thomas’ Theological Summa (trans. Dom Bede Rose; St. Louis, Mo.: Herder, 1946), 460-73; and Garrigou-Lagrange, Predestination: The Meaning of Predestination in Scripture and the Church (trans. Dom Bede Rose; St. Louis, Mo.: Herder, 1939; repr., Rockford, Ill.: Tan Books & Publishers, 1998), 126-67.
69. Francis Turretin, First Through Tenth Topics (vol. 1 of Institutes of Elenctic Theology; ed. James T. Dennison, Jr.; trans. George Musgrave Giger; Phillipsburg, N.J.: Presbyterian & Reformed, 1992), 215–16.
70. Ibid., 214. Cf. A. A. Hodge, Outlines of Theology, 148—on this page Hodge gives a succinct and helpful list of the standard Reformed criticisms of the doctrine of middle knowledge, which are presented more thoroughly by Turretin, Institutes, 1:212-18. For other helpful criticisms of the scientia media from a Reformed perspective, see Bavinck, Reformed Dogmatics, 2:201-2; and Charles Hodge, Systematic Theology (3 vols.; Grand Rapids: Eerdmans, n.d.), 1:398-400.
71. The author wishes to thank Scott Oliphint, Carl Trueman, Oliver Crisp, Becca Shingleton, and Matthew Jordan for their comments on earlier drafts of this article.