A Middle Knowledge Perspective on Divine Providence

Author: Thomas Rauchenstein, © 1999

Abstract: Theologians and philosophers have long debated how to reconcile the doctrines of divine providence and human free will. Since these doctrines appear to be incompatible with each other, some thinkers have felt it necessary to either reject theism altogether or redefine the traditional concept of God. The following essay attempts to show that neither course of action is required. I argue that the theory of ‘middle knowledge’ can reconcile these doctrines in a manner that is both theologically consistent and philosophically sound.


  1. Providence and Libertarian Freedom
  2. The Molinist Doctrine of Middle Knowledge
  3. From Creaturely-World-Types to Feasible Worlds
  4. The ‘Grounding Objection’: A Reply to Hasker and Adams
  5. Conclusion: the Implications of Molinism for Divine Providence


The biblical tradition conceives of providence as a combination of two characteristics of God: foreknowledge and sovereignty.[1] That God has foreknowledge of all future events means that “there is no event still to occur of which God is ignorant or uncertain. God never has to ‘wait and see’ how things will develop.”[2]  Exhaustive knowledge of the future is an essential attribute of God because if he knows everything, he must also know every truth about what will come to pass (assuming there are such truths), and do so with absolute certainty. Yet it is not enough for God to simply know what will occur; he must also predetermine which events are the ones to occur.[3] For if God only has the power to act once he knows the future, the future will have already been fixed before any decision on his part could be made to change it. It would be akin to knowing everything about a fiction novel, but having no power to alter the plot because it has already been written. Sovereignty requires that God be, first and foremost, the author of the novel so that his knowledge of the plot is a logical outcome of having written it himself. Similarly, only if foreknowledge is a logical consequence of sovereignty can God have genuine control over the world’s events. As Thomas Flint remarks, “If [foreknowledge] were not so dependent – if his decisions made no difference to his foreknowledge – then the notion of God’s being in control of his world would clearly be a sham.”[4]

On the opposite end of the spectrum, human beings are said to possess free will.[5] By free will I mean the “libertarian” principle that not all human actions are determined by prior causal factors.[6] If my choices are determined by antecedent causes, then they are not free because it is causally impossible for me to do otherwise. It is not enough that I could have chosen otherwise had my circumstances been different. Free will implies that I could have done otherwise in the very same circumstances. Libertarians therefore adhere to some version of the following principle: circumstance + free will = undetermined action. Of course, free will may be conditioned upon various causal factors which are either internal or external to the agent performing an action, but none of those factors are sufficient to determine which choice will be made.

Libertarians have generally offered two legs of support for their position (though these are by no means exhaustive). First, they argue that moral responsibility hinges on the ability to choose between alternatives. Human beings cannot be held accountable if all their decisions are causally determined because that would undermine the moral status of their actions. If deeds like giving to charity or helping the poor are the inevitable byproduct of causal forces which are governed by the laws of nature, then it makes no sense to ascribe normative properties like praiseworthiness to them. Likewise, if social and psychological factors determine people to commit rape, then no justification can be given for calling them criminals. They had no choice in the matter. These examples suggest that something is seriously wrong with determinism as a general theory of human action. The fact is, “if I ought to do something, it seems to be necessary to suppose that I can do it.”[7] Second, many libertarians maintain that human reasoning would be a sham without free will. Imagine that you are trying to solve an intellectual puzzle involving two options: X and Y. You deliberate on the matter and finally come to the conclusion that X is the correct answer. But if determinism is true, your belief that X is the right answer was already decided beforehand. The chain of events leading up to your belief was fixed by prior causal factors, and so you could not have believed differently. Consequently, “if the sole reason for my believing or not believing X is that I am causally determined to believe it I have no ground for holding that my judgment is true or false.”[8] Rational judgment is an illusion if human beings cannot freely choose between alternatives. But since we often do make rational judgments, it follows that determinism is false.[9]

By now the reader may already be suspicious of an emerging tension between free will as defined by libertarians and the doctrine that God has complete control over every event that comes to pass. The ability to reject God’s plan for salvation suggests a further capacity to thwart his will, which, as we saw earlier, is not consistent with the concept of providence outlined above. Nelson Pike is a well-known philosopher of religion who argues for this very point. He maintains that one aspect of God’s providence, namely exhaustive knowledge of the future, is logically incompatible with the libertarian view. This position is known as theological fatalism.[10] Traditionally, the argument has been stated as follows:

(1) Necessarily, if God foreknows that I do x, then I do x

(2) God foreknows that I do x

(3) Therefore, necessarily I do x

Premise (1) seems to be true, for it is logically impossible that God’s knowledge be false. That is, the composite state of affairs of God’s foreknowing that I do x, and my doing non-x, does not obtain in any possible world. Premise (2) is also true. The God of traditional theism knows all and only truths, including future tense propositions about creaturely freedom. But if it is necessary that whatever God knows comes to pass, then is it not necessary that I do x? Is it not impossible for me to do otherwise? Unfortunately, this does not follow because the above argument is fallacious. In modal logic, a necessary conclusion cannot follow from an argument that possesses a contingent premise, namely (2). What follows is

(3’) Therefore, I do x.

which is hardly an affront to my freedom! This fatalistic argument shows that I shall do x, not that I must do x.[11] But perhaps (2) can be modified,

(2’) Necessarily, God foreknows that I do x

which logically entails, with (1), that I must do x. But (2’) is false for the theist, because God’s knowledge is not the same in every possible world. It is necessary that He be omniscient, but the content of His foreknowledge is contingent – i.e. it varies from world to world.

Contrary to this reasoning, Nelson Pike argues that there is a sense in which (2’) is true: God’s past knowledge is temporally necessary. It may be the case that God’s foreknowledge is logically contingent, but since his believing that I shall do x is part of the past, it is now impossible for me to change that belief. To change God’s past beliefs is to cause past events – which is patently absurd. So in light of the temporal necessity of God’s beliefs, Pike draws out a new argument for fatalism:

If God existed at t1, and if God believed at t1 that Jones would do X at t2, then if it was within Jones’ power at t2 to refrain from doing X, then (1) it was within Jones’s power at t2 to do something that would have brought it about that God held a false belief at t1, or (2) it was within Jones’ power at t2 to do something which would have brought it about that God did not hold the belief he had at t1, or (3) it was within Jones’ power at t2 to do something that would have brought it about that any person who believed at t1 that Jones would do X at t2 (one of whom was, by hypothesis, God) held a false belief and thus was not God – that is, that God (who by hypothesis existed at t1) did not exist at t1.[12]

He goes on to say that (1) – (3) are unacceptable. (1) entails that Jones can act such that God held a false belief at t1, which obviously flies in the face of traditionalists who want to preserve the infallibility of God’s knowledge. Further (2) – (3) entail that Jones has the power to alter the past. This “power” can mean one of two things: either (a) Jones can act such that God’s past belief at t1 was not a past belief at t1 or (b) Jones has the ability to cause a state of affairs that is past (i.e. the state of God’s non-existence at t1 or His believing that Jones will do non-X instead). But since (a) is logically contradictory and (b) entails the absurdity that one can now cause past events, one must conclude that Jones cannot refrain from performing X. His action is fated by the foreknowledge of God.

In response to Nelson Pike’s argument for theological fatalism, Alvin Plantinga has argued quite persuasively for a type of “power over the past” that does not entail the ability to alter events by retro-causation, what he calls counterfactual power over the past. Consider again the proposition “God foreknows (at t1) that Jones shall do X (at t2).” The libertarian must acknowledge that Jones is free only if he has the power to refrain from doing X at t2. According to Plantinga, Jones can refrain from doing X, but were he to refrain, God would have always foreknown otherwise. It is true that Jones will not exercise his power to do otherwise, but had he done so, God’s past knowledge would have always been different.[13] As William Lane Craig asserts, this type of power “does not mean it is within one’s power to change the past. Rather it is to assert the truth of [various ‘backtracking’] counterfactuals”:

If Jones were to do x at t2, God would have known at t1 that he would do x at t2.


If Jones were to do non-x at t2, God would have known at t1 that he would do non-x at t2.[14]

Thus, even if God’s foreknowledge is part of the past, it does not follow that one’s actions are fated. On the contrary, Plantinga would assert that one has the power to act such that, were one to do so, God would have always known differently. It is this notion of counterfactual power over God’s knowledge that proves fatal to Pikes’ fatalism and enables us to affirm the compatibility of divine foreknowledge and human freedom.[15]

Despite the failure of arguments for theological fatalism, the relationship between sovereignty and freedom remains enigmatic. It is one thing to say that God can have foreknowledge of free actions, but it is quite another thing to say that God predetermines every free action that comes to pass. To understand the difficulty in synthesizing libertarianism with divine sovereignty, ask yourself: “if what I do is up to me, doesn’t it just follow that it is not up to God? God could, presumably, refrain from creating free creatures; but, in creating them, doesn’t he surrender the kind of universal control [that is] essential to the traditional notion of providence?”[16] These questions become especially acute when one recalls that foreknowledge must follow from God’s sovereignty. If God’s exhaustive foreknowledge is not the ultimate consequence of His sovereign determination, then his control over the world is illusory. How then can sovereignty be reconciled with creaturely freedom? A solution to this dilemma was already hinted at in our discussion of theological fatalism – that there are counterfactual truths governing the relationship between foreknowledge and one’s free actions. The affirmation of counterfactual truths, as we shall see, is crucial to the reconciliation of God’s sovereignty with voluntary human action. It is within this context that a full-blown Molinist account of providence can be elucidated.


Molinism is a philosophical view deriving its name from the famous Jesuit priest Luis de Molina. It seeks to explain how God might create a world and knowingly determine the content of its future without violating libertarian freedom.[17] Molinists claim that God possesses “middle knowledge” of all counterfactual statements. Counterfactuals are statements about what would occur if a particular circumstance were to be true, including those truths about creaturely freedom, logically prior to His creative decree to make a world. They also claim that there is a certain order of logical moments in God’s Omniscience that corresponds to, or is symmetric with, the logical order according to which the actual world is instantiated. The exact nature of this logical (or explanatory) order is often left obscure, but a univocal (though somewhat artificial) definition is possible.[18] Suffice it to say that logical priority is not tantamount to temporal priority,[19] in which case, God’s explanatorily progressive knowledge [of different states of affairs that obtain] can be made intelligible within a situation that is not temporally indexed (e.g. God’s “timeless” existence before creation). Consider the chart below:[20]

Logical Moments in Divine Knowledge

In the first moment, God has natural knowledge of all propositions that correspond to logically necessary states of affairs. For instances, at the first moment, God knows that NN: If Clinton were in S, he could freely negotiate with Russia or he could freely refrain from negotiating with Russia. Broadly speaking, God’s natural knowledge allows Him an intuitive grasp of what every logically possible world would be like, including what every free essence could do in any circumstances. Since these propositions are logically necessary, they are neither under God’s control nor dependent on God’s will for their truth.

In the second logical moment of divine omniscience, God has middle knowledge [MK] of all subjunctive (henceforth called “counterfactual”) propositions, including counterfactual propositions about human freedom.[21] At this juncture, God knows what every creaturely essence would do if put in certain circumstances. For example, F: If Peter were in C, he would deny Christ three times,[22] is a true counterfactual proposition about what Peter would do in C; and (F), like all others of its kind, corresponds to counterfactual states of affairs that obtain.[23] God’s intuitive knowledge of (F) is appropriately called “middle” because, as illustrated in the above diagram, it is logically posterior to God’s natural knowledge, but prior to His free knowledge. Characteristic of MK is the fact that (akin to natural knowledge) God possesses it independently of any divine decree to actualize anything. The content of scientia media, therefore, is not under God’s control. But unlike natural knowledge, MK is contingent.

Why should one hold that the content of God’s middle knowledge is neither (a) necessary nor (b) under God’s divine control? If (b) were not the case, then the truth of (F) would not be contingent on what Peter would have done, but on what God decides. As Thomas Flint says, “[t]o suggest that God can decide which counterfactuals are to be true is to abandon the libertarian standpoint essential to Molinism.”[24] So God does not have control over what He knows at the second logical moment of His omniscience. Second, (a) is true because if Peter must deny Christ three times, then it is broadly logically impossible that he refrain from doing so, which destroys the indeterminate causal power we ascribed to him in the first place. So the content of MK is contingent as well.

Finally, we turn to God’s divine decree to actualize a world (which, in logic, occurs between the second and third moments of divine omniscience); at that point, all remaining states of affairs – e.g. future-tense states of affairs – obtain. And at the third moment, which is logically posterior to that decree, God has free knowledge (or foreknowledge) of all these remaining states of affairs in the actual world.[25] Foreknowledge is – unlike MK – dependent on God’s sovereign will because He could have chosen not to make any temporal world whatsoever. And without a temporal world, there would be no foreknowledge. Fundamentally speaking, foreknowledge is an accidental byproduct of God’s combining his pre-volitional knowledge with his divine decree to create. It plays no role in determining what the future will be like, for it “happens” too late in the explanatory chain of knowledge to be of any use to God. As Flint points out, “…on the Molinist picture, God’s foreknowledge is neither the effect nor the cause of our [future] free actions. Foreknowledge follows immediately from God’s conjoining his creative act of will to his prevolitional knowledge.”[26] It is scientia media, not foreknowledge, which furnishes God’s ability to determine the future.


Does God (on a non-Anselmian construal of the divine nature qua logically contingent) have the power to bring about any logically possible world? Let us call any collection of counterfactual propositions of freedom that are true in some possible world a “creaturely-world-type.”[27]  Let CWT1 ….CWTn+1 signify the set of a possible creaturely-world-types. Every member of CWT1 ….CWTn+1 could have been true, but only one member is actually (contingently) true. Finally, suppose that creaturely-world-type CWT1 is the collection of counterfactual propositions that is true. Now this means — in our specific case of interest — that God’s middle knowledge does not afford Him the power to create any possible world containing free creatures. That the truth of a specific world-type precludes some possibilities for world creation is obvious.[28]  Imagine that the following counterfactual (G*) is true, and therefore, contained in CWT1:

G* If the instantiation of G were left free in C, then the instantiation of G would freely do A

The following is technical but crucial: it has already been shown that G* is contingent and is, therefore, false in some possible world (call it W1). In other words, in W1, it is (pre-volitionally) true that the instantiation of G (in C) would have freely refrained from A. But since a different creaturely-world-type is actually true (CWT1), making the pre-volitional structure of W1 false, it follows that God cannot make a world in which G freely does refrain. In attempting to make such a world, God would have to bring it about both that (i) G* is true and that (ii) the instantiation of G* freely refrains from doing A in C. But surely (i) and (ii) are not compossible. Their conjunction is necessarily false. Since it is actually true that G* [and God has no choice in the matter], W1 is not feasible for God to create. For God to “bring it about that” W1, He must cause the instantiation of G to act in a way contrary to what he would freely choose in C. And if God brings it about that G does something other than what he would freely do in a situation, then he didn’t do it freely.[29] Thus, given the truth of the actual creaturely world type CWT1 is follows that at least one (and probably many more) worlds are not feasible for God to instantiate, though they are logically possible worlds. God, then, just cannot actualize any logically possible world containing free creatures.

Let us call the set of all worlds made feasible by true counterfactuals a galaxy of worlds, and let that galaxy be named ‘Alpha-1.’  Only the worlds contained in Alpha-1 are really “possible” for God to make. And so God’s “selection” is restricted by which worlds are proper members of that galaxy. Does Alpha-1 put any constraints on the number (rather than the kind) of worlds God could create? Certainly not. For if there are an infinite number (or at least a potentially infinite number) of creaturely essences that God could choose to instantiate, then it seems eminently reasonable to hold that there are an infinite number of feasible worlds contained in Alpha-1 that are available to Him. But, again, not all logically possible worlds are within God’s power to create.

4. THE “GROUNDING OBJECTION”: A Reply to Hasker and Adams

The central tenet of Molinism is its claim that God has “middle” knowledge of what creaturely essences would do if placed in any circumstances whatsoever. For example, the proposition,

(F) If Peter were in C, he would deny Christ three times

exemplifies, according to Molinists, a true “counterfactual of freedom” about the state of affairs that, if Peter were in such and such a situation, he would deny his Lord three times. Moreover, (F) is true independently of whether Peter exists or not, because the state of affairs to which (F) corresponds obtains pre-volitionally – logically prior to God’s contingent decree to make a world. Molina himself believed that since counterfactual states of affairs pertaining to creaturely freedom obtain at the second moment of God’s omniscience, they are true explanatorily prior to the existence of any categorical facts. Robert Adams is therefore correct in his assessment that (F)’s truth “is independent of the truth or falsity of its antecedent.”[30] In response to the Molinist position, William Hasker asks, “[w]hat, if anything, is the ground of the truth of the counterfactuals of freedom?…” or more bluntly, “[w]hat makes… counterfactuals true?”[31] Hasker offers the following criterion for the truth-grounding of counterfactual propositions:

In order for a (contingent) conditional state of affairs to obtain, its obtaining must be grounded in some categorical state of affairs. More colloquially, truths about ‘what would be the case…if’ must be grounded in truths about what is in fact the case….the truth of causal conditionals, and of their associated counterfactuals, are grounded in the natures, causal powers, inherent tendencies, and the like, of the natural entities described in them.[32]

Hence, counterfactuals (especially those regarding creaturely freedom) can be true or false only if they are grounded in categorical facts about what is already the case in the actual world. But according to the theory of middle knowledge, counterfactuals cannot be grounded in categorical states of affairs because they are true logically prior to any divine decision to make a world. Hence, since the Molinist account cannot satisfy Hasker’s criterion, counterfactuals are neither true nor false. Without true counterfactuals, there is nothing left to guide God’s creative decision, and so the Molinist is finished.

What, then, remains to make (F) true? Perhaps there are facts about Peter’s characteristics that render it certain that the counterfactual in question is true. That is, perhaps (F) is “true by virtue of correspondence with the desires and character” of Peter. Unfortunately, these factors cannot be sufficient to make the counterfactual about Peter true, because if they were sufficient, one would be inclined to say that Peter is psychologically determined to act as he does. But as Robert Adams informs us, if Peter is really a free agent, then he “may act out of character, or change his intentions, or fail to act on them.”[33]  At most, then, antecedent ontological facts about character and psychological inclination provide only probabilistic grounds for affirming (F); they cannot function as grounds for counterfactual truth.

So it appears that adherents of middle knowledge will search in vain to find a categorical ground by which one may judge counterfactuals of freedom as either true or false. Since (F) is not bivalent, God can only know that

(F’) If Peter were in C, he would-probably deny Christ three times.

However, this knowledge is virtually useless to the Molinist who wishes to sustain a traditional account of divine providence.

But does the “grounding” objection formulated by Hasker and Adams inflict damage to the Molinist position? It seems not. Hasker’s conjecture that the truth of counterfactuals is “grounded in the natures, causal powers, inherent tendencies, and the like, of the natural entities described in them” seems false. For one can “entertain counterfactuals about what the world would have been like were different laws of nature or boundary conditions to obtain.”[34] For instance, consider Stephen Hawking’s statement that,

If the rate of expansion one second after the big bang had been smaller even by one part in a hundred thousand million million, the universe would have recollapsed before it ever reached its present size.[35]

This counterfactual truth seems uncontroversial, at least given our present understanding of cosmology. But realize that nothing about what actually is the case makes it true. In reality, the antecedent of this statement is false. And contrary to what some might think, this proposition cannot be grounded in the properties inherent in the numerical values conducive to cosmic collapse, since these values for the Big Bang expansion did not actually obtain. Rather, had the mathematical values mentioned in Hawking’s statement been the case, the universe would have never reached its present size — this is itself another counterfactual state of affairs. Hence, though Hawking’s statement is true, one will search in vain to find any categorical facts that make it true.[36]

But even if Hasker’s criticisms can reasonably show that counterfactuals in science must be grounded in categorical facts about the actual world, the Molinist may point out that counterfactuals of freedom are in no way undermined. As Craig notes, “[t]he demand for a [categorical or essential] ground for volitional counterfactual states of affairs seems misguided. It implicitly presupposes that libertarianism and agent causation are false doctrines.”[37] Consider the past-tense proposition “Peter denied Christ.” One might press the libertarian and ask, what makes it true that Peter acted as he did? The Libertarian will respond that the objector has misconstrued the situation. Nothing about Peter (whether factual or essential) can make it true that he denied Christ; that would be determinism. Rather “Peter denied Christ” is true because the corresponding present-tense proposition “Peter denies Christ” was true. The fact that he freely did it grounds its truth. Now consider the future-tense proposition “Peter will deny Christ.” Again, (on pain of determinism) it would be wrong-headed to ask what makes this statement true; it truth is grounded because its present-tense counterpart “Peter is denying Christ” will be true. The fact that he will do it grounds its truth.[38] Finally, consider the volitional counterfactual “If Peter were in C, he would deny Christ.” As is the case with past-tense and future-tense statements, no categorical facts or essential characteristic about Peter make this counterfactual true. It is true because the present-tense proposition “Peter denies Christ” would be true in the respective circumstances. The fact that he would have done it (had C been the case) grounds its truth. To ask for anything more is to implicitly deny the libertarian sense in which past-tense and future-tense statements about human freedom are grounded.


The theological appeal of the Molinist position, when applied to the Christian doctrine of providence, is obvious. We saw earlier that no incompatibility can be shown to exist between God’s foreknowledge of future contingents and human freedom, and that a plausible solution to the arguments for theological fatalism stemmed from the truth of counterfactual propositions governing the relationship between God’s past beliefs and our future free actions. At that point, the Molinist theory of scientia media was used to explain how God could sovereignly determine the content of the future without violating human volition. According to the theory, God possesses “middle knowledge” of all counterfactual truths about creaturely freedom, and can arrange a world according to such truths – creating persons in various circumstances, and knowing exactly what their free responses would be. All feasible worlds are at God’s disposal to create, and once He decides to make a world, his contingent will “sets in stone” the events that will be. Moreover, the ‘grounding objection’ to counterfactual bivalence was considered, but proved to be unsuccessful. Hasker and Adams have misconstrued the libertarian sense in which counterfactuals of freedom are true. In all, Molinism provides a fruitful account of divine knowledge that preserves human free will and makes sense of the conception of providence that is upheld by traditional theists.


[1] Scripture confirms the idea that foreknowledge and sovereignty are attributes of God. Though a biblical analysis of these two components of providence is beyond the scope of this paper, several passages from the Bible’s prophetical literature can be cited to support this idea. The prophet Isaiah declares, “Behold, the former things have come to pass, and new things I declare; before they spring forth I tell you of them” (42:9). This passage is significant because Isaiah often uses the criterion of foreknowledge as proof of the LORD’s supremacy over the pagan gods of surrounding nations. Since only the God of Israel has certain knowledge of history from beginning to end, the prophet asks, in typical rhetorical style, “Who has performed and done it, calling the generations from the beginning? ‘I, the LORD, am the first; and with the last I am He’” (41:4). Quite remarkable is the manner in which God discerns the very thoughts and decisions of the human heart before they have occurred. This is evidenced by the prediction of Peter’s three-stage denial of Christ in the Gospel accounts[1] and by Kind David’s proclamation in Psalm 139:4, “Before a word is on my tongue, you know it completely.” God’s sovereignty is also attested to by the Scriptures, for as the Psalmist proclaims, nothing happens unless God preordains it. “All the days ordained for me were written in your book before one of them came to be” (139:16). According to the author of Proverbs, even seemly random and contingent events are subject to God’s control: “The lot is cast onto the lap, but its every decision is from the Lord” (16:33).

[2] Thomas Flint, Divine Providence: The Molinist Account (New York: Cornell University Press, 1998), p. 13.

[3] “God exercises sovereignty over [the] world…in the sense that every event, no matter how large or small, is under God’s control and is incorporated into his overall plan for the world.” Ibid, p. 13.

[4] Ibid, p. 36.

[5] The Scriptures teach that human beings are free in the libertarian sense of the word. Throughout the biblical plot-line, God is portrayed as a father who solicits the free response of his creatures for the purpose of mutual friendship and intimacy. The theme of a loving Deity who desires relationship with his creatures is wonderfully portrayed in Old Testament passages like Ezekiel 18: 31-32 where God is described as lamenting the disobedience of His covenant partner Israel, even pleading for her repentance. One even notices a hint of desperation in God’s voice when he asks, “Why will you die, O house of Israel? For I do not take pleasure in the death of anyone, declares the Sovereign LORD. Repent and live.” The New Testament reaffirms the theme of God’s patience in keeping his promise of salvation, “not wanting anyone to perish, but everyone to come to repentance.” (2 Peter 3:9) Paul elaborates on this further when he writes that God “wants all people to be saved and come to the knowledge of the truth”; believers are exhorted to pray for “kings and all those in authority” because of God’s universal salvific intent (1 Tim 2:1-4). What conclusions can we draw? The fact that some persons are unsaved implies that God’s offer of salvation is not intrinsically efficacious. The efficacy of his offer is extrinsic because its application is dependent on one’s free acceptance or rejection of the Gospel message, however involved God may be in helping a person to believe.

[6] For the sake of simplicity, I wish to focus this discussion on one specific position within the libertarian camp – agent causation. For a helpful article on the nature of agent causation, see William L. Rowe, “Two concepts of Freedom,” in Proceedings and Addresses of the American Philosophical Association Supplement to Volume 61 (Sept. 1987), pp. 43-64.

[7] J.P. Moreland, Scaling the Secular City (Grand Rapids: Baker, 1987), p. 89. Here I want to avoid any mention of the rare (and perhaps irrelevant) situations envisaged by some philosophers who hold that, for any indeterminate act, the ability to refrain from that act is not a necessary condition for its being done freely. Imagine, if you will, that a mad neural surgeon has fused an electric apparatus to the lobes of your brain. If, at any time, you were to choose something contrary to his desires, he would force you to do otherwise by sending the required electric pulses into your brain. Suppose further that you freely and deliberately obey every one of the neural surgeon’s requests, such that he never actually has make use of the brain device. Do you still possess libertarian freedom in such a situation, even though you do not possess the ability to refrain from what you choose to do? An answer to that question is beyond the scope of this paper, and I will not attempt to incorporate this novel conundrum into the foregoing discussion. For the purposes of simplicity, then, I will allow the reader to alter my arguments as she sees fit.

[8] H.P. Owen, Christian Theism (Edinburgh: T & T Clark, 1984), p. 118.

[9] Ironically, even if determinism were true, we would not, according to its own strictures, have any rational grounds for believing it! In the end, it is at least prima facie plausible to say that some form of indeterminism is true.

[10] Nelson Pike, “Divine Omniscience and Voluntary Action,” Philosophical Review Volume 74 (1965).

[11] Thomas Aquinas recognizes this point in his Summa Contra Gentiles, bk. I, ch. 67, 10. 10. Another way of understanding this is to make the medieval distinction between necessity in the composed sense (in sensu composite) and necessity in the divided sense (in sensu diviso). In the composite sense, the modal operator applies to the entire conditional – e.g. “Necessarily, if God foreknows that I shall do x, then x.” In the divided sense, the modal operator applies to the consequent of the conditional – e.g. “if God foreknows that I shall do x, then necessarily x.” The theological fatalist wishes to interpret God’s knowing in sensu diviso, but such is a false characterization for the situation. God’s foreknowledge is necessary only in the sense that the composite state of affairs of God’s believing that I shall to do x, and my doing otherwise, never obtains in any possible world. The modality involved in God’s knowledge is compatible with human freedom in sensu composite, but not in sensu diviso.

[12] Pike, p. 33-34.

[13] Alvin Plantinga, God, Freedom, and Evil (Grand Rapids: Eerdmans, 1974), p. 72.

[14] William Lane Craig, “Tachyons, Time Travel, and Divine Omniscience” The Journal of Philosophy Volume 85 (1988), pp. 135-50.  I have adapted Craig’s quotation to fit with Nelson Pike’s example of Jones.

[15] Of course, this point is hotly debated among philosophical theologians. William Hasker has gone to great lengths to show that counterfactual power over the past is really an untenable notion. See his God, Time, and Knowledge (London: Cornell University Press, 1989), pp. 96-115. For a response to this, see William Lane Craig’s “Hasker on Divine Knowledge,” Philosophical Studies Volume 62 (1992), pp. 57-78.

[16] Thomas Flint, Divine Providence, p. 36.

[17] The Molinist account was originally formulated by Luis de Molina in On Divine Foreknowledge Trans Alfred Freddoso (Ithaca: Cornell University Press, 1988). For a very compact and latter-day look at middle knowledge,  see William Lane Craig’s “No Other Name: A Middle Knowledge Perspective on the Exclusivity of Salvation Through Christ,” Faith and Philosophy Volume 6 (1989), pp. 6-9 (internet version)

[18] Thomas Flint, Divine Providence: The Molinist Account (New York: Cornell University Press, 1998), pp. 174-176.

[19] Briefly consider the two propositions below:

(TP) My reading Shakespeare’s Romeo and Juliet is prior to my reading Hamlet.
(LP) My understanding that {[If p → q] • [p]} is prior to my understanding that [q]

(TP) requires a temporal context in which my reading of Romeo and Juliet at t1 comes, in the order of chronology, before my reading of Hamlet at t2. But (LP) need not have any temporal properties. {[If p → q] • [p]} is logically prior to [q] — and this can be shown by the fact that {[If p → q] • [p]}→ [q], but not visa versa. But one need not “delay” or “wait and see” in order to grasp this relationship. The “moments” of understanding indicative of (LP) are logical, not temporal. [At the very least, temporality is an accidental feature of this type of logical reasoning] Hence, logical priority does not entail temporal priority.

[20] This diagram was taken from William Lane Craig’s The Only Wise God (Grand Rapids: Baker House Books, 1987), p. 143.

[21] Jonathan Kvanvig in The Possibility of an All-Knowing God (Palgrave: Macmillan, 1986) correctly distinguishes between “subjunctives of freedom” and “counterfactuals o freedom.” The latter is a bit of a misnomer because conditional propositions about human freedom are not necessarily contrary to fact. Unfortunately, philosophers have perpetuated this mistake, and so the expression “counterfactuals of freedom” will be used throughout.

[22] This subjunctive statement may be symbolized as the following:   F = (p)(Cp → Dp)

where ‘→’ stands for the counterfactual dependency relation; p = the constant “Peter”; Cx = x is in circumstance C; and  Dx = x denies Christ

[23] Counterfactual truths satisfy the same criterion of correspondence as putative truths, that is, correspondence to states of affairs that obtain. The truth that “If I were a bigot, I would kill myself” is true if and only if it is the case that if I were a bigot, I would in fact kill myself. Nothing else makes that proposition true.

[24] Flint, p. 123.

[25] I have deliberately avoided the question of whether counterfactuals of divine freedom are to be construed as part of God’s free knowledge or as part of His pre-volitional knowledge. Counterfactuals of divine freedom are properly denoted as such because they correspond to what God would do in “reaction” to His MK — including what He would do had His MK been different. For fear of being too technical at this point, suffice it to say that divine counterfactuals obviously cannot be known as items of scientia media. For middle knowledge must already be (logically) fixed before God can comprehend how He would react to it.

[26] Flint, p. 44.

[27] I am indebted to Thomas Flint for his helpful terminology here. See his Divine Providence p 51-54. I have found that terms such as “creaturely-world-type” and “galaxy” (this second term will be utilized later) facilitate an accurate understanding of the modality involved in world feasibility.

[28] For an excellent defense of the position that not all logically possible worlds are within God’s power to create, see Flint’s “The Problem of Divine Freedom,” American Philosophical Quarterly Volume 20 (1983), pp. 256-257. Flint’s ideas concisely summarize Alvin Plantinga’s sentiments regarding the concept of feasibility of worlds in God, Freedom, and Evil, pp 34-45. What follows will be a shorter rendition of both their defenses.

[29] Alvin Plantinga’s distinction between strong and weak actualization is, perhaps, a more fruitful way of understanding why God cannot create some possible worlds. For any x, if God strongly actualizes x, then He will directly cause x to be true (e.g. a miracle). With this said, God cannot strongly actualize a free action, for I cannot be caused to freely do something – that would be a contradiction. Rather God weakly actualizes my free actions – i.e. by putting me in the appropriate circumstances, God knows what I would freely do. So God weakly brings about my action, but does not strongly cause or determine my action. This distinction has direct relevance to the notion of feasible worlds. Suppose that a person named Dave is in circumstance (C), and that there are two possible worlds that could result:

|————- Dave helps his mother (W1)
|————- Dave does not help his mother (W2)

Also suppose that if Dave were in C, he would freely help his mother. Can God strongly actualize W2? No, because were he to do so, Dave would not be free with respect to C. Can God weakly actualize W2? Again, he cannot. For if God creates Dave in C, Dave – since he is the author of his own actions – would choose W1 by helping his mother instead. Hence, if God can neither strongly nor weakly actualize W2, then He cannot actualize W2. There is a possible world that God cannot create.

[30] Adams, “Middle Knowledge and the Problem of Evil,” American Philosophical Quarterly Volume 14 (1977), p. 113.

[31] Hasker, God, Time, and Knowledge, p. 29.

[32] Ibid, p. 30.

[33] Adams, p. 111.

[34] Craig, “Hasker on Divine Knowledge,” pp. 57-78.

[35] Stephen Hawking, A Brief History of Time (New York: Bantam Books, 1988), pp. 121-22.

[36] Hasker could plausibly avoid this criticism by claiming that counterfactuals about what the universe would have been like are grounded in the essential nature of physical laws themselves. These counterfactuals are grounded because the mathematical relations characterized by those laws are fixed, no matter how many different constants or values for variables get ‘plugged into’ those laws. However, as will soon become apparent, this point about essential natures will be of no use to the dissenter of counterfactuals of freedom.

[37] Craig, “Hasker on Divine Knowledge,” pp. 57-78. I am indebted to Craig and to Flint for the argumentation that follows.

[38] To my understanding, William Hasker has (in correspondence with Thomas Flint) acknowledged that future-tense statements about creaturely freedom are grounded, though he holds that these truths cannot be known. See Flint’s Divine Providence, p. 130.