The Middle-Knowledge View
3
THE MIDDLE-KNOWLEDGE VIEW
William Lane Craig
he climax of Charles Dickens's wonderful classic A Christmas Carol comes when Scrooge, shaken by the scenes shown him by the Spirit of Christmas Yet to Come, pleads, "Answer me one question. Are these the shadows of the things that Will be, or are they shadows of things that May be, only?"' The ghost does not speak a word in answer to Scrooge.
And with good reason! For had the spirit responded, "These shadows are merely scenes of things that could be," Scrooge might well have breathed a sigh of relief and gone on with his life as before. "After all," he might quite rightly reflect, "almost anything could happen! No need to lose sleep about that!" On the other hand, if the spirit had told him candidly, "No, these shadows are not scenes of things that will be" (as we know to be true from the story's end), then Scrooge might have felt no cause for alarm at all, since none of what he had witnessed would in fact come to pass. In that case, he might not have been led to repent and change his life.
Scrooge's problem was that he was asking the wrong question; he had failed to exhaust the alternatives. For between what could be and what will be lies what would be. What the spirit was revealing to Scrooge was what would happen if Scrooge did not repent and change. The spirit was not exhibiting mere possibilities (it was possible that Scrooge would sell his business and open a flower stand in Covent Garden, but who cares about that?), nor was he showing Scrooge what was in fact going to happen (Dickens assures us that Tiny Tim did not die). Rather the spirit was warning Scrooge that if he did not repent, all these terrible things would come to pass.
In philosophical terminology, the spirit was revealing to Scrooge a bit of counterfactual knowledge. Counterfactuals are conditional statements in the subjunctive mood: for example, "If I were rich, I would buy a Mercedes"; "If Barry Goldwater had been elected president, he would have won the Vietnam War"; and "If you were to ask her, she would say yes." Counterfactuals are so called because the antecedent or consequent clauses are typically contrary to fact: I am not rich, Goldwater was not elected president, and the U.S. did not win the Vietnam War. Nevertheless, sometimes the antecedent and/or consequent is true. For example, your friend wants to ask the girl of his dreams for a date and, emboldened by your reassurance that "If you were to ask her, she would say yes," does ask her and she does say yes.
Counterfactual statements make up an enormous and significant part of our ordinary language and are an indispensable part of our decision making: For example, "If I pulled out into traffic now, I wouldn't make it"; "If I were to ask J. B. for a raise with his mood, he'd tear my head off"; "If we sent the Third Army around the enemy's right flank, we would prevail." Clearly life-and-death decisions are made daily on the basis of the presumed truth of counterfactual statements.
The Doctrine of Middle Knowledge
Christian theologians have typically affirmed that in virtue of his omniscience, God possesses counterfactual knowledge. He knows, for example, what would have happened if he had spared the Canaanites from destruction, what Napoleon would have done had he won the Battle of Waterloo, and how Jones would respond if I were to share the gospel with him. Not until Friedrich Schleiermacher and the advent of modern theology did theologians think to deny God knowledge of true counterfactuals. Everyone who had considered the issue agreed that God has such knowledge.
What theologians did dispute, however, was, so to speak, when God has such counterfactual knowledge. The question here did not have to do with the moment of time at which God acquired his counterfactual knowledge. For whether God is timeless or everlasting throughout time, in neither case are there truths that are unknown to God until some moment at which he discovers them. As an omniscient being, God must know every truth there is and so can never exist in a state of ignorance. Rather the "when" mentioned above refers to the point in the logical order concerning God's creative decree at which God has counterfactual knowledge.
This idea of a logical order with regard to God's decrees is a familiar one to Reformed theologians. For although all God's decrees occur at once rather than sequentially, there is a logical order among the decrees. For example, infralapsarians say that God decreed Christ's death on the cross in order to remedy humanity's fall into sin, so that logically God's decree of the cross comes after his decree of the fall. By contrast, supralapsarians say that God's primary aim for humankind was redemption via the cross, and therefore he decreed the fall in order to have something to redeem humans from. On this scheme, the decree of the cross is logically prior to the decree of the fall. Thus even though it was agreed on all hands that God's decrees occur all at once, theologians debated how they were to be logically arranged.
A similar dispute existed among post-Reformation theologians with respect to the place of God's counterfactual knowledge. Everybody agreed that logically prior to God's decree to create a world, God has knowledge of all necessary truths, including all the possible worlds he might create. This was called God's natural knowledge. It gives him knowledge of what could be. Moreover, everyone agreed that logically subsequent to his decree to create a particular world, God knows all the contingent truths about the actual world, including its past, present and future. This was called God's free knowledge. It involves knowledge of what will be. The disputed question was where one should place God's counterfactual knowledge of what would be. Is it logically prior to or posterior to the divine decree?
Catholic theologians of the Dominican order held that God's counterfactual knowledge is logically subsequent to his decree to create a certain world. They maintained that in decreeing that a particular world exist, God also decreed which counterfactual statements are true. Logically prior to the divine decree, there are no counterfactual truths to be known. All God knows at that logical moment are the necessary truths, including all the various possibilities.
At that logically prior moment God knows, for example, that there is a possible world in which Peter denies Christ three times, and another possible world in which Peter affirms Christ, and yet another world in which it is Matthew who denies Christ three times, and so on. God picks one of these worlds to be actual, and thus subsequent to his decree it is true that Peter will deny Christ three times. Moreover, God knows this truth because he knows which world he has decreed. Not only so, but in decreeing a particular world to be real, God also decrees which counterfactuals are true. Thus he decrees, for example, that if Peter had instead been in such-and-such circumstances, he would have denied Christ two times. God's counterfactual knowledge, like his foreknowledge, is logically posterior to the divine creative decree.
By contrast, Catholic theologians of the Jesuit order inspired by Luis de Molina maintained that God's counterfactual knowledge is logically prior to his creative decree. This difference between the Jesuit Molinists and the Dominicans was no mere matter of theological hair-splitting! The Molinists charged that the Dominicans had in effect obliterated human freedom by making counterfactual truths a consequence of God's decree, for on the Dominican account it is God who determines what each person will do in whatever circumstances he finds himself. By contrast, the Molinists, by placing God's counterfactual knowledge prior to the divine decree, made room for creaturely freedom by exempting counterfactual truths from God's decree. In the same way that necessary truths like 2 + 2 = 4 are prior to and therefore independent of God's decree, so also counterfactual truths about how creatures would freely choose under various circumstances are prior to and independent of God's decree.
Not only does the Molinist view make room for human freedom, but it affords God a means of choosing which world of free creatures to create. For by knowing how persons would freely choose in whatever circumstances they might be in, God can-by decreeing to place just those persons in just those circumstances-bring about his ultimate purposes through free creaturely decisions. Thus, by employing his counterfactual knowledge, God can plan a world down to the last detail and yet do so without annihilating creaturely freedom, since what people would freely do under various circumstances is already factored into the equation by God. Since God's counterfactual knowledge lies logically in between his natural knowledge and his free knowledge, Molinists called it God's middle knowledge.
On the Dominican view, there is one logical moment prior to the divine creative decree at which God knows the range of possible worlds that he might create; then he chooses one of these to be actual. On the Molinist view, there are two logical moments prior to the divine decree: first, the moment at which he has natural knowledge of the range of possible worlds, and second, the moment at which he has knowledge of the proper subset of possible worlds that, given the counterfactuals true at that moment, are feasible for him to create. The counterfactuals which are true at that moment thus serve to delimit the range of possible worlds to worlds feasible for God.
For example, there is a possible world in which Peter affirms Christ in precisely the same circumstances in which he in fact denied him. But given the counterfactual truth that if Peter were in precisely those circumstances he would freely deny Christ, then the possible world in which Peter freely affirms Christ in those circumstances is not feasible for God. God could make Peter affirm Christ in those circumstances, but then his confession would not be free.
Thus on the Molinist scheme, we have the following logical order:
Arguments for Middle Knowledge
Why should one think that the Molinist scheme is correct? Three lines of argument-biblical, theological and philosophical-may be adduced in support of the Molinst position. Let us consider each in turn.
Biblical arguments. Biblically speaking, it is not difficult to show that God possesses counterfactual knowledge. One of the Jesuit theologians' favorite proof-texts was 1 Samuel 23:6-10, which tells of David's using of a divining device (an ephod) to inquire of the Lord whether Saul would attack Keilah, where David was ensconced, and whether the men of Keilah would deliver David over to Saul. In both cases, the device registered an affirmative answer, whereupon David fled the city, so that the predictions did not in fact come true. What the device had mediated to David was not, therefore, simple foreknowledge but counterfactual knowledge. God was letting David know that if he were to remain at Keilah, then Saul would come after him, and that if Saul were to come after David, then the men of Keilah would deliver him over to Saul. The answers given by the divining device were thus correct answers, even though the events did not come to pass, since the answers were indicative of what would happen under certain circumstances.
Although most scriptural prophecy is unconditional, sometimes prophecies are explicitly conditional, like the one David received at Keilah. Consider, for example, Jeremiah's prophecy to King Zedekiah in Jeremiah 38:17-18:
Thus says the LORD, the God of hosts, the God of Israel, If you will only surrender to the officials of the king of Babylon, then your life shall be spared, and this city shall not be burned with fire, and you and your house shall live. But if you do not surrender, ... then this city shall be handed over to the Chaldeans, and they shall burn it with fire, and you yourself shall not escape from their hand.
In his omniscience God knew what would happen whichever course of action Zedekiah chose. Indeed, when we construe certain prophecies as counterfactual warnings, rather than as categorical declarations of simple foreknowledge, we can explain how it is that in Israel the test of a true prophet is the fulfillment of his predictions (Deut 18:22) and yet some predictions given by true prophets do not actually come to pass because the people forewarned responded in an appropriate way (Is 38:1-5; Amos 7:1-6; Jon 3:1-10). In such cases, the prophecy from God was counterfactual knowledge of what would happen under the prevailing circumstances; but were intercessory prayer or repentance to occur, then God would not carry out what had been threatened.
We also find counterfactual knowledge exhibited by Christ. For example, he tells Peter, "Go to the sea and cast a hook; take the first fish that comes up; and when you open its mouth, you will find a coin; take that and give it to them for you and me" (Mt 17:27). This passage is most naturally understood as an expression of Jesus' knowledge that if Peter were to carry out Jesus' instructions, he would find things as the Lord predicted. Again, Jesus commands the disciples, after a futile night of fishing, "Cast the net to the right side of the boat, and you will find some [fish]" On 21:6). The miraculous catch that ensued shows that Jesus knew exactly what would happen if the disciples obeyed his command. Sometimes Jesus makes counterfactual statements himself: "If I had not come and spoken to them, they would not have sin.... If I had not done among them the works that no one else did, they would not have sin" (Jn 15:22, 24). "If my kingdom were of this world, my followers would be fighting to keep me from being handed over to the Jews" (Jn 18:36). "Woe to that one by whom the Son of Man is betrayed! It would have been better for that one to not have been born" (Mt 26:24). These are only a few of many examples.
I think it is plain, then, that the God of the Bible exhibits counterfactual knowledge. Given God's infallibility, it will not do to explain these examples as mere hunches on God's part. If God believes that Saul would besiege Keilah if David were to stay there, then that counterfactual statement is known by God to be true, since it is logically impossible for God to subscribe to false beliefs.
Unfortunately, this does not answer the question of whether God has middle knowledge. For the scriptural passages show only that God possesses counterfactual knowledge, and, as I have said, until modern times all theologians agreed that God possesses counterfactual knowledge. The dispute among them concerned when in the logical order of things this knowledge comes: is it before or after the divine decree? 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.
Theological arguments. The strongest arguments for the Molinist perspective are theological. Once one grasps the concept of middle knowledge, one will find it astonishing in its subtlety and power. Indeed, I would venture to say that it is the single most fruitful theological concept I have ever encountered. In my own work, I have applied it to the issues of Christian particularism, perseverance of the saints and biblical inspiration.2 Thomas Flint has used it to analyze infallibility; and Del Ratzsch has used it to explore evolutionary theory.3 An article begs to be written on a Molinist perspective of quantum indeterminacy and divine sovereignty. With respect to the concerns of this book, middle knowledge provides an illuminating account of divine foreknowledge and providence.
Divine Foreknowledge
The doctrine of divine foreknowledge raises two questions: First, is divine foreknowledge compatible with future contingents? And second, how can God know future contingents? Though I think that acceptable answers to these questions are available to the defender of simple foreknowledge, it is worth laying out a Molinist perspective as well.
The compatibility of divine foreknowledge and future contingents. With respect to the first question, it must be the case that divine foreknowledge and future contingents (in particular, human free acts) are compatible for the simple reason that Scripture teaches both.' When so-called openness theologians dispute this compatibility, their denial is clearly driven not by biblical exegesis but by a philosophical argument derived from ancient Greek fatalism and dressed in theological guise; biblical exegesis is being bent to support a conclusion already determined by philosophical considerations. This is ironic since these same theologians loudly decry the polluting influence of Greek philosophical thought upon the biblical tradition. In fact, it is they themselves who have been seduced by philosophical reasoning of Greek provenance, which was stoutly resisted by the early church fathers. If openness theologians are not convinced by the proffered solutions to fatalism, then it is the better part of intellectual humility simply to confess that one lacks the philosophical insight to solve the problem (cf. Ps 139:6) and hold the biblical doctrines in tension rather than deny the Scripture's clear teaching that God does know the future.5
So what is the argument that allegedly demonstrates the incompatibility of divine foreknowledge and human freedom? Letting x stand for any event, the basic form of the argument is as follows:
(1) Necessarily, if God foreknows x, then x will happen.
(2) God foreknows x.
(3) Therefore, x will necessarily happen.
Since x happens necessarily, it is not a contingent event. In virtue of God's foreknowledge, everything is fated to occur.
The problem with the above form of the argument is that it is just logically fallacious. What is validly implied by (1) and (2) is not (3) but (3'):
(3') Therefore, x will happen.
The fatalist gets things all mixed up here. It is correct that in a valid, deductive argument the premises necessarily imply the conclusion. The conclusion follows necessarily from the premises; that is to say, it is impossible for the premises to be true and the conclusion to be false. But the conclusion itself need not be necessary. The fatalist illicitly transfers the necessity of the inference to the conclusion itself. What necessarily follows from (1) and (2) is just (3'). But the fatalist in his confusion thinks that the conclusion is itself necessarily true and so winds up with (3). In so doing he simply commits a common logical fallacy.
The correct conclusion, (3'), is in no way incompatible with human freedom. From God's knowledge that I shall do x, it does not follow that I must do x but only that I shall do x. That is in no way incompatible with my doing x freely.
Undoubtedly, a major source of the fatalist's confusion is his conflating certainty with necessity. In the writings of contemporary theological fatalists, one frequently finds statements which slide from affirming that something is certainly true to affirming that something is necessarily true. This is sheer confusion. Certainty is a property of persons and has nothing to do with truth, as is evident from the fact that we can be absolutely certain about something that turns out to be false. By contrast, necessity is a property of statements or propositions, indicating that a proposition cannot possibly be false. We can be wholly uncertain about statements that are, unbeknownst to us, necessarily true (for example, some complex mathematical equation or theorem). Thus, when we say that some statement is "certainly true," this is but a manner of speaking indicating that we are certain the statement is true. People are certain; propositions are necessary.
By confusing certainty and necessity, the fatalist makes his logically fallacious argument deceptively appealing. For it is correct that from (1) and (2) we can be absolutely certain that x will come to pass. But it is muddle-headed to think that because x will certainly happen then x will necessarily happen. We can be certain, given God's foreknowledge, that x will not fail to happen, even though it is entirely possible that x fail to happen. Event x could fail to occur, but God knows that it will not. Therefore, we can be sure that it will happenand happen contingently.
Contemporary theological fatalists recognize the fallaciousness of the above form of the argument and therefore try to remedy the defect by making (2) also necessarily true:
(1) Necessarily, if God foreknows x, then x will happen.
(2) Necessarily, God foreknows x.
(3) Therefore, x will necessarily happen.
So formulated, the argument is no longer logically fallacious, and so the question becomes whether the premises are true.
Premise (1) is clearly true. It is perhaps worth noting that this is the case not because of God's essential omniscience or inerrancy but simply in virtue of the definition of "knowledge." Since knowledge entails true belief, anybody's knowing that x will happen necessarily implies that x will happen. Thus, we could replace (1) and (2') with the following:
(1*) Necessarily, if Smith truly believes that x will happen, then x will happen.
(2*) Necessarily, Smith truly believes that x will happen.
And (3) will follow as before. Therefore, if any person ever holds true beliefs about the future (and surely we do, as we smugly remind others when we say, "I told you so!"), then, given the truth of (2), fatalism would also follow from merely human beliefs, a curious conclusion!
Indeed, as ancient Greek fatalists realized, the presence of any agent at all is really superfluous to the argument. All one needs is a true, future-tense statement to get the argument going. Thus, we could replace (1) and (21) with this:
(1**) Necessarily, if it is true that x will happen, then x will happen.
(2**) Necessarily, it is true that x will happen.
In order to avoid the above generalization of their argument to all persons and to mere statements about the future, theological fatalists will deny that the second premise is true with respect to humans or mere statements, as it is for God. They will say that Smith's holding a true belief or some future-tense statement's being true is not necessary in the way that God's holding a belief is necessary.
That raises the question as to whether (2') is true. At face value, (2') is obviously false. Christian theology has always maintained that God's creation of the world is a free act, that God could have created a different world-in which x does not occur-or even no world at all. To say that God necessarily foreknows any event x implies that this is the only world God could have created and thus denies divine freedom.
But theological fatalists have a different sort of necessity in mind when they say that God's foreknowledge is necessary. They are talking about temporal necessity, or the necessity of the past. Often this is expressed by saying that the past is unpreventable or unchangeable. If some event is in the past, then it is now too late to do anything to affect it. It is in that sense necessary. Since God's foreknowledge of future events is now part of the past, it is now fixed and unalterable. Therefore, it is said, (2') is true.
But if (2') is true in that sense, then why are not (2*) and (2**) true as well? The theological fatalist will respond that Smith's beliefs being true or a futuretense statement's being true is not a fact or an event of the past, as is God's holding a belief.
But such an understanding of what constitutes a fact or event seems quite counterintuitive. If Smith believed in 1997 that "Clinton will be impeached," was it not a fact that his belief was true? If Smith held that same belief today, would it not be a fact that his belief is no longer true (since Clinton has left office)? If Smith's belief changes from being true to being false, then surely it was a fact that it was then true and is a fact that it is now false. The same obviously goes for the mere statement "Clinton will he impeached." This statement once had the property of being true and now has the property of being false. In any reasonable sense of "fact," these are past and present facts.
Indeed, a statement's having a truth value is plausibly an event as well. This is most obvious with respect to statements like "Flight 4750 to Paris will depart in five minutes." That statement is false up until it is five minutes prior to departure, it becomes true at five minutes till departure, and then it becomes false again immediately thereafter. Other statements' being true may be more long-lasting events, like "Flight 4750 to Paris will depart within the next hour." Such a statement's being true is clearly an event on any reasonable construal of what constitutes an event.
No theological fatalist whom I have read has even begun to address the question of the nature of facts or events which would make it plausible that Smith's truly believing a future-tense statement and a future-tense statement's being true do not count as past facts or events. But then we see that theological fatalism is not inherently theological at all. If the theological fatalist's reasoning is correct, it can be generalized to show that every time we hold a true belief about the future or every time a statement about the future is true, then the future is fated to occur-surely an incredible inference!
Moreover, we have the best of reasons for thinking that (2') is defective in some way: namely, fatalism posits a constraint on human freedom that is unintelligible. For the fatalist admits that the events God foreknows may be causally indeterminate; indeed, they could theoretically be completely uncaused, spon taneous events. Nevertheless, such events are said to be somehow constrained. But by what? Fate? What is that but a mere name? If my action is causally free, how can it be constrained by God's merely knowing about it?
Sometimes fatalists say that God's foreknowledge places a sort of logical constraint on my action. Even though I am causally free to refrain from my action, there is some sort of logical constraint upon me, rendering it impossible for me to refrain. But insofar as we can make sense of logical constraints, these constraints are not analogous to the sort of necessitation imagined by the theological fatalist. For example, given the fact that I have already played basketball at least once in my life, it is now impossible for me to play basketball for the first time. I am thus not free to go out and play basketball for the first time. But this sort of constraint is not at all analogous to theological fatalism. For in the case we are envisioning, it is within my power to play basketball or not. Whether or not I've played before, I can freely execute the actions of playing basketball. It's just that if I have played before, my actions will not count as playing for the first time. By contrast, the fatalist imagines that if God knows that I shall not play basketball, then even though I am causally free, my actions are mysteriously constrained so that I am literally unable to walk out onto the court, dribble and shoot. But such noncausal determinism is utterly opaque and unintelligible.
The argument for fatalism must therefore be unsound. Since (1) is clearly true, the trouble must lie with (2'). And (2) is notoriously problematic. For the notion of temporal necessity to which the fatalist appeals is so obscure a concept that (2') becomes a veritable mare's nest of philosophical difficulties. For example, since the necessity of (1) is logical necessity and the necessity of (2') is temporal necessity, why think that such mixing of different kinds of modality is valid? If the fatalist answers that logical necessity entails temporal necessity, so that (1) can be construed merely in terms of temporal necessity, then how do we know that such necessity is passed on from the premises to the conclusion in the way that logical necessity is? Indeed, since x is supposed to be a future event, how could it be temporally necessary? And even if x is temporally necessary, how do we know that this sort of necessity is incompatible with an action's being free? So long as a person's choice is causally undetermined, it is a free choice even if that person is unable to choose the opposite of that choice.7 So even if x were temporally necessary, such that not-x cannot occur, it is far from obvious that x is not freely performed or chosen.
All of the above problems arise even if we concede (2') to be true. But why think that this premise is true? What is temporal necessity anyway, and why think that God's past beliefs are now temporally necessary? Theological fatalists have never provided an adequate account of this peculiar modality. I have yet to see an explanation of temporal necessity according to which God's past beliefs are temporally necessary that does not reduce to either the unalterability or the causal closedness of the past.
But interpreting the necessity of the past as its unalterability (or unchanga- bility or unpreventability) is clearly inadequate, since the future, by definition, is just as unalterable as the past. To change the future would be to bring it about that an event which will occur will not occur, which is self-contradictory. It is purely a matter of definition that the past and future cannot be changed, and no fatalistic conclusion follows from this truth. We need not be able to change the future in order to determine the future. If our actions are freely performed, then it lies within our power to determine what the course of future events will be, even if we do not have the power to change the future.
Fatalists will insist that the past is necessary in the sense that we do not have a similar ability to determine the past. Nonfatalists may happily concede the point: backward causation is impossible. But the past's causal closedness does not imply fatalism. For freedom to refrain from doing as God knows one will do does not involve backward causation.
Here we come to the Molinist solution to theological fatalism. The Molinist is quite glad to admit that nothing I can do now will cause or bring about the past. But he will insist that it does lie within my power to freely perform some action a, and if a were to occur, then the past would have been different than it in fact is. Suppose, for example, that God has always believed that in the year 2000 I would accept an invitation to speak at the University of Regensburg. Up until the time arrives I have the ability to accept or refuse the invitation. If I were to refuse the invitation, then God would have held a different belief than the one he in fact held. For if I were to refuse the invitation, then different counterfactual propositions would have been true, and God would have known this via his middle knowledge. Neither the relation between my action and a corresponding counterfactual proposition about it, nor the rela tion between a true counterfactual proposition and God's believing it, is a causal relation. Thus, the causal closedness of the past is irrelevant. If temporal necessity is merely the causal closedness of the past, then it is insufficient to support fatalism.
To my knowledge, no fatalist has explicated a conception of temporal necessity that does not amount to either the unalterability or the causal closedness of the past. Typically, fatalists just appeal gratuitously to some sort of "Fixed Past Principle" to the effect that it is not within my power to act in such a way, and that if I were to do so, then the past would have been differentwhich begs the question. On analyses of temporal necessity that are not reducible to either the unalterability or the causal closedness of the past, God's past beliefs always turn out not to be temporally necessary." The Molinist accounts for this in terms of God's middle knowledge, which infallibly tracks true counterfactual propositions concerning our free choices.
Thus, the argument for theological fatalism is unsound. The doctrine of middle knowledge helps to make the compatibility of divine foreknowledge and human freedom perspicuous.
The basis of divine foreknowledge of future contingents. What, then, about the second question raised by divine foreknowledge, namely, the basis of God's knowledge of future contingents? Detractors of divine foreknowledge sometimes claim that because future events do not exist, they cannot be known by God. The reasoning seems to go as follows:
(1) Only events that actually exist can be known by God.
(2) Future events do not actually exist.
(3) Therefore, future events cannot be known by God.
Now (2) is not uncontroversial. A good many physicists and philosophers of time and space argue that future events do exist. They claim that the difference between past, present and future is merely a subjective matter of human consciousness. For the people in the year 2015, the events of that year are just as real as the events of our present year are for us; and for those people, it is we who have passed away and are unreal. On such a view God transcends the four-dimensional space-time continuum, and thus all events are eternally present to him. It is easy on such a view to understand how God could therefore know events that to us are in the future.
Nevertheless, I do think that such a four-dimensional view of reality faces insuperable philosophical and theological objections, which I have discussed elsewhere.' Therefore, I am inclined to agree with (2) of the above argument. So the question becomes whether there is good reason to think that (1) is true.
In assessing the question of how God knows which events will transpire, it is helpful to distinguish two models of divine cognition: the perceptualist model and the conceptualist model. The perceptualist model construes divine knowledge on the analogy of sense perception. God looks and sees what is there. Such a model is implicitly assumed when people speak of God's "foreseeing" the future or having "foresight" of future events. The perceptualist model of divine cognition does run into real problems when it comes to God's knowledge of the future, for, since future events do not exist, there is nothing there to perceive.
By contrast, on a conceptualist model of divine knowledge, God does not acquire his knowledge of the world by anything like perception. His knowledge of the future is not based on his "looking" ahead and "seeing" what lies in the future (a terribly anthropomorphic notion in any case). Rather God's knowledge is self-contained; it is more like a mind's knowledge of innate ideas. As an omniscient being, God has essentially the property of knowing all truths; there are truths about future events; thus, God knows all truths concerning future events.
Middle knowledge can help us understand how God knows truths about the future. Divine foreknowledge is based on God's middle knowledge of what every creature would freely do under any circumstances and on his knowledge of the divine decree to create certain sets of circumstances and to place certain creatures in them. Given middle knowledge and the divine decree, foreknowledge follows automatically as a result.
Of course, the skeptic may ask how God knows counterfactuals concerning human free choices if those choices do not exist. Molinists could respond either that God knows the individual essence of every possible creature so well that he knows just what each creature would do under any set of circumstances he might place him in, or that God, being omniscient, simply discerns all the truths there are and, prior to the divine decree, there are not only necessary truths but counterfactual truths, and therefore God possesses not only natural knowledge but middle knowledge as well.
So long as we are not seduced into thinking of divine foreknowledge on the model of perception, it is no longer evident why knowledge of future events should be impossible. A conceptualist model along the lines of middle knowledge furnishes a perspicuous basis for God's knowledge of future contingents.
Thus, both with respect to the problem of theological fatalism and the question of the basis of divine foreknowledge, the Molinist doctrine of middle knowledge provides an illuminating account of God's foreknowledge.
Divine Providence
The Molinist account of divine providence is even more stunning than its account of divine foreknowledge. Here its superiority to the doctrine of simple foreknowledge emerges. Consider the following biblical passages:
This man ljesusl, handed over to you according to the definite plan and foreknowledge of God, you crucified and killed by the hands of those outside the law. (Acts 2:23)
For in this city, in fact, both Herod and Pontius Pilate, with the Gentiles and the peoples of Israel gathered together against your holy servant Jesus, whom you anointed, to do whatever your hand and your plan had predestined to take place. (Acts 4:27-28)
Here we have a staggering assertion of divine sovereignty over the affairs of men. The conspiracy to crucify Jesus-which involved not only the Romans and the Jews in Jerusalem at that time, but more particularly Pilate and Herod, who tried Jesus-is said to have happened by God's plan based on his foreknowledge and foreordination. How are we to understand so far-reaching a providence as this?
If we take the term foreknowledge as encompassing middle knowledge, then we can make perfect sense of God's providential control over a world of free agents. For via his middle knowledge, God knew exactly which persons, if members of the Sanhedrin, would freely vote for Jesus' condemnation; which persons, if in Jerusalem, would freely demand Christ's death, favoring the release of Barabbas; what Herod, if king, would freely do in reaction to Jesus and to Pilate's plea to judge him on his own; and what Pilate himself, if holding the prefecture of Palestine in A.D. 27, would freely do under pressure from the Jewish leaders and the crowd. Knowing all the possible circumstances, persons and permutations of these circumstances and persons, God decreed to create just those circumstances and just those people who would freely do what God willed to happen. Thus, the whole scenario, as Luke insists, unfolded according to God's plan. This is truly mind-boggling. When one reflects that the existence of the various circumstances and persons involved was itself the result of a myriad of prior free choices on the part of these and other agents, and these in turn of yet other prior contingencies, and so on, then we see that only an omniscient mind could providentially direct a world of free creatures toward his sovereignly established ends. In fact, Paul reflects that "none of the rulers of this age understood this; for if they had, they would not have crucified the Lord of glory" (1 Cor 2:8). Once one grasps the doctrine of divine middle knowledge, one is led to adoration and praise of God for his breathtaking sovereignty.
Now what account of divine providence can be given in the absence of middle knowledge? Advocates of divine openness freely admit that without middle knowledge, a strong doctrine of divine providence becomes impossible. But such a viewpoint can make no sense whatsoever of scriptural passages such as those cited above. I am bewildered that partisans of this camp can deny divine foreknowledge while claiming to be biblical, when, as seen above, "foreknowledge" is part of the very vocabulary of the New Testament. Nor can it be said that God's plan was hit upon by him late in the game, once he could reasonably guess what the relevant agents would do. For as Paul was wont to emphasize, this was an eternal plan, made from the foundations of the world but hidden for ages in God and now realized in the fullness of time, as God sent forth his Son, manifesting the wisdom of God to the principalities and powers who oppose him (Gal 4:4; Eph 3:9-11; cf. 1 Pet 1:20).
The Augustinian-Calvinist perspective interprets the above passages to mean that foreknowledge is based upon foreordination: God knows what will happen because he makes it happen. Aware of the intentions of his will and his almighty power, God knows that all his purpose shall be accomplished. But this interpretation inevitably makes God the author of sin, since it is he who moved Judas, for example, to betray Christ, a sin that merits the hapless Judas everlasting perdition. But how can a holy God move people to commit moral evil and, moreover, how can these people then be held morally responsible for acts over which they had no control? The Augustinian-Calvinist view seems, in effect, to turn God into the devil.
The proponent of simple foreknowledge can make no good sense of God's providentially planning a world of free creatures. For logically prior to the divine decree, God has natural knowledge of all the possible scenarios, but he does not have knowledge of what would happen under any circumstances. Thus, logically posterior to the divine decree, God must consider himself extraordinarily lucky to find that this world happened to exist. ("What a break!" we can imagine God saying to himself, "Herod and Pilate and all those people each reacted just perfectly!") Actually, the situation is much worse than that, for God had no idea whether Herod or Pilate or the Israelite nation or the Roman Empire would even exist posterior to the divine decree. Indeed, God must be astonished to find himself existing in a world-out of all the possible worlds he could have created-in which mankind falls into sin and he himself enters human history as a substitutionary sacrificial offering to rescue them! Of course, I am speaking anthropomorphically here; but the point remains that without middle knowledge, God cannot know prior to the creative decree what the world would be like. If the defender of simple foreknowledge goes on to say that God's foreordination of future events is based upon his simple foreknowledge, then this trivializes the doctrine of foreordination, making it a fifth wheel that carries no load since, as we have seen, the future by definition cannot be changed. Once God knows that an event really is future, there is nothing more left to do; foreordination becomes a redundancy. Surely, there is more substance to the biblical doctrine of foreordination than the triviality that God decrees that what will happen will happen!
Thus of the options available, the Molinist approach provides by far the most elucidating account of divine providence. It enables us to embrace divine sovereignty and human freedom without mysticism or mental reservation, thereby preserving faithfully the biblical text's affirmation of both these doctrines. We therefore have powerful theological motivation for adopting the Molinist perspective.
Philosophical arguments. Finally, I think that we also have good philosophical grounds for thinking that a doctrine of middle knowledge is correct. We may argue as follows:
(1) If there are true counterfactuals of creaturely 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 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 divine middle knowledge.
Let us say a word in defense of each of the argument's premises. The truth of (1) is required by the definition of omniscience:
(0) For any agent x, x is omniscient = def For every proposition p, if p, then x knows that p and does not believe not-p.
What (0) requires is that any agent is omniscient if and only if he knows all truths and believes no falsehoods. This is the standard definition of omniscience. It entails that if there are counterfactual truths, then an omniscient being must know them.
Opponents of divine foreknowledge have suggested revisionary definitions of omniscience so as to be able to affirm that God is omniscient even as they deny his knowledge of future contingents (and counterfactuals of creaturely freedom)."' William Hasker's revisionist definition is typical:
(O') God is omniscient = def. God knows all propositions which are such that God's knowing them is logically possible.
Revisionists then go on to claim that it is logically impossible to know propositions about future contingents, as shown by the argument for theological fatalism, and so God may count as omniscient despite his ignorance of an infinite number of true propositions.
As it stands, however, (O') is drastically flawed. It does not exclude that God believes false propositions as well as true ones. Worse, (0') actually requires God to know false propositions, which is incoherent as well as theologically unacceptable. For (0') requires that if it is logically possible for God to know some proposition p, then God knows p. But if p is a contingently false proposition, say, There are eight planets in the sun's solar system, then there are possible worlds in which p is true and known by God. Therefore, since it is logically possible for God to know p, he must actually know p, which is absurd. What the revisionist really wants to say is something like this:
(O") God is omniscient = def. God knows only and all true propositions which are such that it is logically possible for God to know them.
Unlike (O'), (0") limits God's knowledge to a certain subset of all true propositions.
The fundamental problem with all such revisionary definitions of omniscience as (0') is that any adequate definition of a concept must accord with our intuitive understanding of the concept. We are not at liberty to "cook" the definition in some desired way without thereby making the definition unacceptably contrived. Definition (0") is guilty of being "cooked" in this way. For intuitively, omniscience involves knowing all truth, yet according to (0") God could conceivably be ignorant of infinite realms of truth and yet still count as omniscient. The only reason why one would prefer (0") to (0) is because one has an ulterior motivation to salvage the attribute of omniscience for a cognitively limited deity rather than to deny outright that God is omniscient. Definition (0") is therefore unacceptably contrived.
A second problem with (0') is that it construes omniscience in modal terms, speaking, not of knowing all truth, but of knowing all truth which is knowable. But omniscience, unlike omnipotence, is not a modal notion. Roughly speaking, omnipotence is the capability of actualizing any logically possible state of affairs. But omniscience is not merely the capability of knowing only and all truths; it is knowing only and all truths. Nor does omniscience mean knowing only and all knowable truths, but knowing only and all truths, period. It is a categorical, not a modal, notion.
Third, the superiority of (0") over (0) depends on there being a difference between a truth and a truth that is logically possible to know. If there is no difference, then (0") collapses back to the general definition (0) and the revisionist has gained nothing. What is a sufficient condition for a proposition to be logically knowable? So far as I can see, the only condition is that the proposition be true. What more is needed? If revisionists think that something more is needed, then we may ask them for an example of a proposition that could be true but logically impossible to know. A proposition like "Nothing exists" or "All agents have ceased to exist" comes to mind; but on traditional theism these propositions are not possibly true, since God is an agent whose nonexistence is impossible. Unless the revisionist can give us some reason to think that a proposition can be true yet unknowable, we have no reason to adopt (0"). It seems that the only intrinsic property that a proposition must possess in order to be logically knowable is truth.
The revisionist will claim at this point that true future contingent propositions are logically impossible for God to know, since if he knows them, then they are not contingently true, as shown by the fatalistic argument." But here the revisionist commits a logical howler. He reasons that for any future-tense proposition p it is impossible that God know p and p be contingently true; therefore, if p is contingently true, it is not possible that God knows p. But such reasoning is logically fallacious. What follows is merely that God does not know p, not that it is impossible that God knows p. Thus, even granted the fatalist's false premise (that it is impossible that God know p and p be contingently true), it does not follow from p's contingent truth that p is such that it is logically impossible for God to know p. Therefore, even on the defective definition (0") proposed by the revisionist, God turns out not to be omniscient, since p is a true proposition that, so far as we can see, is logically possible for God to know, and yet God does not know p. Thus, the theological fatalist must deny divine omniscience and therefore reject God's perfection-a very serious theological consequence, indeed.
So with regard to the first premise of our philosophical argument, omniscience requires that if there are true counterfactuals of creaturely freedom, God must know them.
Premise (2) asserts that there are true counterfactuals of creaturely freedom. This premise does not require us to believe that all counterfactuals about creatures' free acts are either true or false. But it does seem plausible that counterfactuals of the following form are either true or false (letting P be any person, A some action and C any set of circumstances including the whole history of the world up until the point of decision):
(CCF) If P were in C, P would freely do A.
It is counterfactuals of this form that we dignify with the title "counterfactuals of creaturely freedom."
We have every reason to think that there are true counterfactuals of creaturely freedom. In the first place, it is plausible, as I say, that counterfactuals of the form (CCF) are true or false. For once the circumstances are fully specified, any ambiguity which might cause us to doubt that the counterfactual has a truth value is removed. And it is plausible that in many cases P would freely do A in C, just as the counterfactual states. Second, we ourselves often know the truth of such counterfactuals. For example, if I were to offer my wife a plate of chocolate-chip cookies and a plate of liver and onions, I know which one she would choose as surely as I know almost anything! A little reflection reveals how pervasive and indispensable such counterfactual truths are to rational conduct and planning. We base our very lives upon their truth or falsity. Third, as pointed out above, Scripture itself gives examples of such true counterfactuals (think again about Paul's statement in 1 Cor 2:8).
The most common objection urged against the truth of counterfactuals of creaturely freedom is the so-called grounding objection. The basic complaint here is that there is nothing to make such counterfactuals true (since they are supposed to be true logically prior to God's creative decree and even now are usually contrary-to-fact); but without a ground of their truth, they cannot be true.
Thomas Flint, an eminent defender of middle knowledge, has rightly observed that the grounding objection is, in the minds of many philosophers, the principal obstacle to endorsing a Molinist perspective.12 It is therefore all the more remarkable that this objection is virtually never articulated or defended in any depth by its advocates. No anti-Molinist to my knowledge has yet responded to Alvin Plantinga's simple retort: "It seems to me much clearer that some counterfactuals of freedom are at least possibly true than that the truth of propositions must, in general, be grounded in this way.s13 What Plantinga understands-and what the grounding objectors generally do not-is that behind the grounding objection sticks a theory about the relationship of truth and reality that is both subtle and controversial and that needs to be articulated, defended and applied to counterfactuals of creaturely freedom if the grounding objection is to have any force. Anti-Molinists have not even begun to address these issues.
The theory presupposed by the grounding objection is a particular construal of truth as correspondence known among contemporary philosophers as the theory of truth-makers. 14 According to a view of truth as correspondence, a statement is true if and only if reality is as that statement describes. In order to identify the reality corresponding to a true statement, one typically employs the method of disquotation: the statement "Snow is white," for example, is true if and only if snow is white. During the revival of the correspondence theory of truth in the early part of the twentieth century, philosophers such as Ber trand Russell and Ludwig Wittgenstein maintained that there must exist not only truth-bearers (whether these be sentences or thoughts or propositions or what have you) which have the property of being true and so corresponding with reality, but also something in reality in virtue of which the sentences or propositions are true. This interpretation of the correspondence theory was taken up again in the 1980s as the theory of truth-makers.
A truth-maker may be defined as that in virtue of which a sentence orproposi- tion is true. Immediately we see the potentially misleading connotations of the term truth-maker. The word making suggests a causal relation involving some concrete object, but truth-makers are not normally so conceived by their advocates. Instead truth-makers are typically construed to be abstract realities like "facts" or "states of affairs"-more often than not, the fact stated as the truth conditions of a proposition, as disclosed by disquotation. Thus, what makes the statement "Snow is white" true is the fact that snow is white or the state of affairs of snow's being white. Such abstract entities do not stand in causal relations. This invalidates at a single swoop the crude construal of the grounding objection expressed in Robert Adams's demand, "Who or what does cause them [counterfactuals of creaturely freedom] to be true?"15 The question is inept because the relation between a proposition and its truth-maker is not a causal relation.
The grounding objector seems to think that in order to be true, counterfactuals of creaturely freedom must have truth-makers that either are or imply the existence of physical objects. But this assumption seems quite unwarranted, since we can think of other types of possibly true propositions whose truthmakers neither are nor imply physical objects. For example:
1. No physical objects exist.
2. Dinosaurs are extinct today.
3. All ravens are black.
4. Torturing a child is wrong.
5. Napoleon lost the Battle of Waterloo.
6. The U.S. president in 2070 will be a woman.
7. If a rigid rod were placed in uniform motion through the ether, it would suffer a Lorentz-FitzGerald contraction.
Statement 1 could be true and statement 2 is true, yet they preclude truthmakers that imply the relevant physical objects, such as dinosaurs. Statement 3 is a universal statement which does not apply just to any ravens that happen to exist and so cannot he made true just by any existing ravens' being black. Statement 4 is a value judgment which implies neither that children do exist nor that any are actually tortured. Statements 5 and 6 are true tensed statements about persons who no longer or do not yet exist and so cannot have such persons among their truth-makers. Finally, statement 7 is a true counterfactual about the ether of nineteenth-century physics, which does not exist. These statements reveal just how naive an understanding grounding objectors have of the notion of truth-makers. For if these statements have truth-makers, their truth-makers are not physical objects out there in the world but are abstract entities like states of affairs or facts.
Now, as I say, it is a matter of debate whether true propositions do have truth-makers. In a recent critique, Greg Restall demonstrates that given the customary axioms of truth-maker theory, it follows that every true proposition is made true by every truth-maker there is, so that, for example, "Grass is green" is made true by snow's being white. In the understatement of the year, Restall muses, "This is clearly not acceptable for any philosophically discriminating account of truthmakers."16 Truth-maker theorists typically deny the doctrine of truth-maker maximalism, the doctrine that every true statement has a truthmaker. I have yet to encounter an argument for the conclusion that counterfactuals of creaturely freedom cannot be among those types of truths lacking a truth-maker. Indeed, when one reflects on the fact that they are counterfactual, then such statements seem prime candidates for that type of statement which is true without any truth-maker.
If there are, on the other hand, truth-makers for counterfactuals of creaturely freedom, then the most obvious and plausible candidates are the facts or states of affairs disclosed by the disquotation principle. Thus, what makes it true that "If I were rich, I would buy a Mercedes" is the fact that if I were rich I would buy a Mercedes. Just as there are tensed facts that now exist even though the objects and events they are about do not (as illustrated by statements 5 and 6 above), so there are counterfacts that actually exist even though the objects and events they are about do not. If counterfactuals of creaturely freedom have truth-makers, then it is in virtue of these facts or states of affairs that the corresponding propositions are true. And since these counterfacts are not the result of God's decree, they exist even logically prior to God's decree to create any physical objects.
In short, I concur with Plantinga in saying that I am far more confident that there are true counterfactuals of creaturely freedom than I am of the theory which requires that they have truth-makers. And if they do, then no reason has been given why these cannot be the facts or states of affairs that are stated as their truth conditions.
Premise 3 of our philosophical argument for middle knowledge states logically exhaustive alternatives for an omniscient deity and so must be true: Counterfactuals of creaturely freedom are known by God either prior to his decree or only after his decree.
Finally, (4) must be true because if counterfactuals of creaturely freedom were known only after the divine decree, then it is God who determined what every creature would do in every circumstance. Augustinian-Calvinist thinkers bear witness to the truth of this premise in their affirmation of compatibilist theories of creaturely freedom. They thereby testify that God's all-determining decree precludes libertarian freedom, which is the sort of freedom with which we are here concerned. Thus, if God knows counterfactual truths about us only posterior to his decree, then there really are no counterfactuals of creaturely freedom. If there are such counterfactuals, they must be true logically prior to the divine decree.
Given the truth of the premises, the conclusion follows that prior to his creative decree God knows all true counterfactuals of creaturely freedom, which is to say that he has middle knowledge, Q.E.D.
Conclusion
In conclusion, while not explicitly taught by the biblical text, the doctrine of divine middle knowledge is certainly compatible with it, which cannot be said of at least some of its competitors. Middle knowledge redounds to the glory of God and illuminates biblical truth in a dazzling way. Moreover, we have good theological and philosophical grounds for affirming middle knowledge. Theologically, middle knowledge enables us to explain both the compatibility of divine foreknowledge with future contingents as well as the basis of divine foreknowledge and, more importantly, to provide an intelligible account of God's providence over a world of free creatures. Philosophically, omniscience by definition entails knowledge of all truth and, since counterfactuals of creaturely freedom are true logically prior to God's creative decree, they must therefore by known by God at that logical moment. Therefore, we should affirm that God has middle knowledge.
AN OPEN-THEISM RESPONSE
Gregory A. Boyd
Reading William Lane Craig's fine essay reminded me of just how close Molinism is to the open view. Indeed, I shall argue that the view that has come to be labeled open theism could perhaps more accurately be labeled neoMolinism. In essence it differs from the classical Molinist position only in that it expands the content of God's middle knowledge to include "might-counterfactuals." In this response I hope to show that this modification allows the open view to avoid problems which attend to the classical Molinist view while preserving its explanatory power.
The Scope of God's Middle Knowledge
As Craig made clear, defenders of middle knowledge hold that between (viz., "in the middle of") God's knowledge of all logical possibilities, on the one hand, and God's knowledge of what will come to pass, on the other, is God's knowledge of counterfactuals of creaturely freedom: namely, what free agents would do if they were created in other possible worlds. Since God is omni scient, it is argued, he must know the truth value of all propositions. Hence he must know the truth value of all counterfactual propositions, including counterfactuals of creaturely freedom. As Craig argues, the chief value of this position is that it allows God to have significant sovereign control of the world while yet accepting libertarian freedom.
Despite its theological advantages, there are several problems with the classical Molinist position, three of which may be mentioned presently. First, while there is some scriptural warrant for holding that God knows counterfactuals of creaturely freedom, classical Molinism is no better at explaining the "open motif" in Scripture than is any other version of classical theism. It has to take as anthropomorphic all passages in which God changes his mind, expresses regret, experiences surprise or disappointment, speaks and thinks of the future in terms of what may or may not happen, and so on. As I argued in my essay, while Scripture certainly contains anthropomorphisms, there are no good grounds for concluding that all passages which constitute the motif of future openness belong in this category.
Second, classical Molinism is philosophically problematic. In this view, every possible decision any possible free agent might ever make in any possible world is an eternal fact. Every future free decision and every possible future free decision is exhaustively settled in eternity before it takes place. How do we account for this eternal settledness? Craig concedes that it cannot exist because God wills it to exist, for that would constitute determinism. But neither can this eternal settledness exist because agents other than God will it to exist, for created agents are not eternal, and Craig rightly denies retroactive causation. Moreover, agents never will the counterfactuals that are supposedly true about them. They are what the agents would have willed had they been created in a different possible world.
We are left then with the unappealing alternative of denying that anything grounds the eternal settledness of future free acts and of counterfactuals of creaturely freedom. From all eternity the fact of what free agents would do in every possible situation was simply there-without any sufficient reason to account for it. It is simply a metaphysical surd. This position, I submit, is at best strongly counterintuitive. It could also be charged with being dualistic, for the settledness of the world, and of every possible world, is depicted as an uncreated reality that eternally coexists alongside God. Like an attribute of God, it has no explanation outside of itself.
Because of this, it's not at all clear that the middle-knowledge position is consistent with libertarian freedom, and this is the third problem with this position. An agent can be said to possess libertarian freedom if it lies within his or her power to do otherwise, given the exact same set of antecedent conditions. But how can we meaningfully say that agents could have done otherwise if all they shall ever do, and all they would have ever done in any possible world, is an unalterable fact an eternity before they even exist? Stated otherwise, how can agents be said to be self-determining when they don't ground the definiteness of what they shall do and ever would have done in different circumstances? The definiteness unexplainably eternally precedes them, created by nothing. If agents possess self-determining or libertarian freedom, I argue, they must be the ones who ultimately resolve indefinite possibilities into settled facts.
Would-Counterfactuals and Might-Counterfactuals
I am persuaded that Craig is correct in claiming that propositions expressing counterfactuals of creaturely freedom have an eternal truth value and thus that God knows them as true or false. Where Craig and other classical Molinists err, I believe, is in assuming that counterfactuals are exclusively about what agents would or would not do. These do not exhaust the logical possibilities of counterfactual propositions. The logical antithesis of the statement "agent X would do y in situation z" is not the statement "agent X would not do y in situation z." This is a contrary proposition, not a contradictory proposition. The logical antithesis of "agent X would do y in situation z" is rather the statement "agent X might not do y in situation z" This latter statement also has an eternal truth value and hence must be known by God.
If we include might-counterfactuals among God's middle knowledge, we arrive at the following neo-Molinist position: Between God's pre-creational knowledge of all logical possibilities and God's pre-creational knowledge of what will come to pass is God's "middle knowledge" of what free agents might or might not do in certain situations as well as of what free agents would do in other situations. If it is true that agent X might or might not do y in situation z, it is false that agent X would do y in situation z, and vice versa. On the basis of this knowledge, God chooses to have actualized the possible world that best suits his sovereign purpose. The world God chooses to be actualized, however, is more precisely described as a delimited set of possible worlds, any one of which might be actualized, depending on the choices free agents make. Yet because God is infinitely intelligent, he is as perfectly prepared for whatever possible world gets actualized as if it were the only possible world that could be actualized. It's just that precisely because we accept that God is infinitely intelligent, a neo-Molinist doesn't suppose that God must choose to falsify all might-counterfactuals in order to acquire this providential advantage.
The Advantages of Neo-Molinism
As with the classical Molinist view, and over against the simple-foreknowledge and Calvinist views, the neo-Molinist view can ascribe to God significant providential control while affirming libertarian freedom. God's knowledge of "would-counterfactuals" and "might-counterfactuals" allows him to providentially bring about the best possible world, given the reality of libertarian freedom. But the neo-Molinist view can accomplish this while avoiding the three problems mentioned above that attend to the classical Molinist position.
First, the neo-Molinist account does not have to dismiss the entire open motif of Scripture as anthropomorphic. We can now accept at face value Scripture's teaching that God sometimes thinks and speaks of the future in terms of what might or might not come to pass (e.g., Ex 4:7-9; 13:17; Jer 26:3; Ezek 12:3). Moreover, we now have no difficulty accounting for how God can sometimes experience regret (e.g., Gen 6:6; 1 Sam 15:11, 35), disappointment and frustration (e.g., Ezek 22:29-31), for people could have, and should have, made different decisions than they did. We have no difficulty accounting for how an omniscient God can change his mind (e.g., Jer 18:7-11; 26:2-3, 13, 19; Jon 3:10), for a world in which some might-counterfactuals are true is a world in which flexibility on the part of God would be an asset, not a liability. We have no difficulty explaining why God tests people "to know" what they will do (e.g., Gen 22:12; Ex 16:4), for some of their future behavior is only describable by might-counterfactuals. And we can even account for why the Bible sometimes depicts God as experiencing surprise (e.g., Is 5:1-5; Jer 3:6-7, 19-20), for while God perfectly anticipates all possible outcomes, when the improbable occurs, it is by definition not what an omniscient God would expect to occur. He is perfectly prepared for it, to be sure. But this doesn't negate the fact that it was improbable.
Second, if we accept that some might-counterfactuals are eternally true, we no longer have the problem of an ungrounded eternal settledness to possible worlds that include libertarian freedom, and there is no longer any problem accounting for libertarian freedom itself. On the one hand, the determinate aspects of any possible world, including worlds that include true might-counterfactuals, are grounded in God's will. For any possible world God might create, there are things that he decides would come to pass if he were to create them and, thus, things that shall come to pass in the world God decides to create. On the other hand, if some might-counterfactuals are true (viz., the world is not exhaustively definite from all eternity), there is no longer any problem trying to render intelligible the supposed eternal settledness of future free actions or free actions in other possible worlds. In the neo-Molinist view, there simply is no eternal settledness to libertarian free actions. There are only eternal possibilities of what they might or might not do.
To the extent that would-counterfactuals apply to future free agents, they do so because the actions of these agents flow either from the character God has given them (habitus infusus, in classical terminology) or from the character they will acquire if they pursue a certain possible course of action (habitus acquisitus, in classical terminology). In either case the would-counterfactuals are not ungrounded, as in classical Molinism. From all eternity God knows that if he chooses to create free agent X, she will have the basic characteristics of a, b and c. And from all eternity God knows that if agent X freely follows a certain possible life trajectory, she will become the kind of person who would do x in situation z. The would-counterfactuals for which agent X is morally responsible are contingent on the might-counterfactuals for which she is morally responsible.
A SIMPLE-FOREKNOWLEDGE
RESPONSE
David Hunt
Warn Lane Craig's overall position is closest to my own, though we disagree philosophically on some of the details. Let's look first at his response to the classic problem of divine foreknowledge versus human freedom and then at his proposal for augmenting divine foreknowledge with middle knowledge.
Foreknowledge and Fatalism
Craig and I stake out middle positions on the foreknowledge-freewill problem, inasmuch as we both believe that the problem is resolvable without abandoning either exhaustive foreknowledge (Gregory Boyd) or libertarian freedom (Paul Helm). We even share some key intuitions about the problem-for example, that it constitutes a philosophical puzzle (like Zeno's Achilles paradox) rather than a serious challenge to Christian belief, and so requires a philosophical rather than theological solution. But we disagree about the kind of philo sophical solution that is required. Craig is convinced that theological fatalism is just a needlessly complex version of what he calls "Greek fatalism": the view that fatalistic consequences follow from the truth of future-tense statements alone, without any appeal to God. Since the two fatalisms are logically equivalent, Craig avers, they fail for essentially the same reason. I disagree: theological fatalism is much more formidable than Greek fatalism, and its solution involves more than correcting a simple logical fallacy.
The fatalistic thesis, as Craig renders it, is that for any future event x,
(3) Necessarily, x will happen.
The fatalist endeavors to convince us of this conclusion by (1) citing some other fact to which it is necessarily connected, and (2) noting that this fact is itself necessary. Greek fatalism cites as the relevant fact a prior truth about the future, such that
(1") Necessarily, if it was true that x will happen, then x will happen.
(2") Necessarily, it was true that x will happen.
Theological fatalism appeals to a prior divine cognition about the future, such that
(1#) Necessarily, if God believed that x will happen, then x will happen.
(2#) Necessarily, God believed that x will happen!
Both arguments validly imply (3), and the first premise of each is true. How about the second?
Assuming (for the sake of argument) that x will happen, it follows both that it was true that x will happen and also that God believed that x will happen. The only question is whether these are true necessarily, as (2**) and (2#) assert. Now the sole ground for thinking that they are necessary is that the facts they set forth are referenced to the past. In the case of (2**), however, this reference rests on nothing more than grammatical artifice. In virtue of what, was it (at some time in the past) a fact that x will happen? It was a fact simply in virtue of its then being the case that x will happen. This "fact about the past" is wholly parasitic on the future. Contrast this with (2#), the fact appealed to by the theological fatalist. In virtue of what, was it a fact that God believed that x will happen? This was not a fact in virtue of anything that will happen (though it is possible to infer something about the future from this fact about the past); it was a fact simply in virtue of what God was then thinking. This "hard" fact about the past, unlike the "soft" fact cited in (2**), is now accidentally (or, to use Craig's terminology, temporally) necessary.
Rather than concede this point, Craig prefers to challenge the very notion of temporal necessity. But this is an unpromising stratagem. The idea that the past is over and done with is a datum, not something that must (or even could) be argued. Nor is it reducible to considerations of causal closedness, as Craig suggests. A past event whose cause has not yet occurred, as in my tachyonic doorbell example (pp. 85-86), would still be temporally necessary: nothing can now happen such that the buzzer did not sound (given that it did sound). Because the undoing of what is done is "more impossible than the raising of the dead to life, which implies no contradiction, and is called impossible only according to natural power. ,2 Aquinas argues that temporal necessity is binding even on God, who can make and unmake merely causal necessities at will.
Craig is undoubtedly right that we lack an adequate account of this "peculiar modality," but this is true of many of our most fundamental concepts. We lack, for example, a philosophically adequate theory of causation. This does not mean that there are no causes or effects; it just means that we don't really understand what is going on when one thing causes another. Likewise our failure to articulate a philosophically rigorous account of temporal necessity does not mean that it isn't real-our sense of its reality is too basic and deep to require validation by a philosophical theory.
There is, then, an important difference between the argument for Greek fatalism and the argument for theological fatalism: the second premise of the former is false, while the second premise of the latter is true! Instead of resisting the truth of (2#), the antifatalist should welcome it as an opportunity to discover the real flaw in the argument-which, to my mind, is the uncritical assumption (based on the Principle of Alternate Possibilities) that (3) is indeed inconsistent with free agency.
Middle Knowledge
So much for Craig's handling of the problem of divine foreknowledge. I turn now to middle knowledge, where there are both theoretical and practical issues to consider.
The practical question concerns Molinism's usefulness for Christian belief. Craig pronounces it "the single most fruitful theological concept I have ever encountered" (p. 125). If we look just at the vexed issue of divine providence, Craig's enthusiasm is understandable. The key questions to ask of any theory of providence are these: how much control does it offer God without jeopardizing human freedom, and how much freedom does it offer human beings without jeopardizing divine control? I do not know of any remotely plausible account of providence that scores as well by this standard as the doctrine of divine middle knowledge. I'm squarely in Craig's camp on this one.
There are other areas, however, in which middle knowledge may be more a hindrance than a help. One such area is the problem of evil. If Zeus were the supreme being, this problem would hardly arise. (We might instead be baffled by why there is so much excess good in the world!) Evil is a problem for the God of the Bible only because he is understood to be all-good, all-powerful and allknowing. So wouldn't conceptualizing God so that he is even more knowledgeable (by endowing him with middle knowledge) make the problem even worse?
It's hard to answer this question in the abstract, so let's give it a more concrete form.3 Consider what might be called the "postmortem problem of evil." The Bible appears to teach that some (many? most?) human beings will spend eternity in hell. Whatever "eternity in hell" amounts to, it is certainly not the purpose for which God created the world-God does not desire this for anyone (2 Pet 3:9). But if he is equipped with middle knowledge, he knew exactly who would reject him prior to creating anyone; knowing this, he could easily have refrained from creating these people. Why didn't he do so? This is a more difficult question to answer for the Molinist than it is for the open theist (whose God lacks this knowledge) or the defender of simple foreknowledge (whose God knows the actual future but cannot use that knowledge to change the very thing he foreknows). This does not show that there are no reasons why God might create people he "middle-knows" would reject him, but the need to posit and defend such reasons is a cost not borne by the non-Molinist.4
Despite difficulties like this one, there are enough advantages to Molinism that I would welcome it if it were true. But is it true? This is the theoretical question, which seems to me on balance to go against the Molinist.
If there are any true counterfactuals of freedom, an omniscient deity will of course know them. But whether there really are such truths is controversial. Craig defends Molinism from one version of the "grounding objection," but he overlooks objections that seem to me more compelling. Owing to space limitations, I can present only one of these here.
What exactly are we claiming when we say something like the following?
If this water had remained on the burner a full five minutes, it would have boiled.
The first part (or antecedent) of the counterfactual posits a contrary-to-fact scenario in which the water remained on the burner longer than it actually did. The second part (or consequent) then states what would happen in that scenario. Counterfactuals thus have a foot in both the actual and nonactual worlds: the antecedent gives us license to ignore reality (in the stipulated respect); but the consequent, by committing itself to one outcome rather than another, lets us know that not just anything goes. Why is it true that the water would boil under those conditions, rather than true that the water would not boil? The reason is that the actual world works that way. If this quantity of water did not come to a boil after being subjected to this degree of heat for five minutes, physical laws that are true in the actual world would instead be false.
Counterfactuals, in sum, ask us to take the world as it actually is, modifying reality only to the extent that this is required by the contrary-to-fact scenario stipulated in the antecedent. A counterfactual of the form
If P were the case, Q would be the case.
is not about just any world in which P is the case; it's about the closest-or-mostsimilar-world-to-the-actual-world in which P is the case, and what it says about that world is that Q is also the case.
Consider now a counterfactual of freedom, such as the following:
(J) If Jones were offered the job, he would freely accept it.
Given the above analysis, what makes (J) true rather than (J-)?
Q-) If Jones were offered the job, he would freely reject it.
The answer is simply that there is a nonactual world W that meets certain conditions, and in W Jones freely accepts the job. What are these conditions? Simply that Jones is offered the job, and everything else is as much like the actual world as possible. Let C stand for all those features of W in virtue of which it satisfies these conditions. So W is a world in which C obtains and Jones accepts the job, or
W C+ A.
And it's because of what happens in W that (J), and not (J-), is true.
Unfortunately, this is incompatible with the putative freedom of Jones's decision. If Jones accepts the job freely in W then C must leave it causally open for him to reject the job. (This is an assumption that Craig and I share.) So there is a possible world W- that is just like Wwith respect to C, except that in W- Jones rejects the job; or
W-: C + not-A.
But then (J) cannot be true. Because W and W- both include C, they are equally similar (in the relevant respects) to the actual world. So (J-) has just as much title to being true as (J). Rather, neither is true; what is true is that
If Jones were offered the job, he might accept it and he might not accept it.
So much for the impressive resources of Molinism.
Does the above argument "refute" Molinism? That's doubtful. My analysis relies on the standard Lewis-Stalnaker semantics for subjunctive conditionals,5 which Craig (following Alvin Plantinga's response to the "grounding objection") can always dismiss this way: "It seems to me much clearer that some counterfactuals of freedom are at least possibly true than that the Lewis-Stalnaker theory of counterfactuals is correct." 6 The problem with this line is that the Lewis-Stalnaker semantics seems generally right: if this theory were false, the result wouldn't be that (J) now had a chance of being true; rather, I wouldn't even know what I was saying in asserting (J). Until a semantics is developed that can make sense out of counterfactuals of libertarian freedom while retaining what is otherwise plausible about the Lewis-Stalnaker theory, it seems prudent to proceed as though middle knowledge is not available as a theological resource.
AN AUGUSTINIAN-CALVINIST
RESPONSE
Paul Helm
As is clear from my comments on Gregory Boyd's contribution, I largely agree with what William Lane Craig says about the open view, particularly its relation to philosophy. And what he has to say about the nature of God's knowledge, his "conceptualist" approach, seems plausible, as do his strictures on "revisionary" concepts of omniscience. Here I shall make some more reflective and critical remarks on his discussion of counterfactuals, then on fatalism, and then on God's being the author of sin.
Counterfactuals
Craig has an interesting discussion of counterfactuals and, as he implies, the Augustinian view (which is roughly equivalent to the Dominican position) has no difficulty ascribing counterfactual knowledge to God. But Craig's discussion needs a bit of clarification. It is only in respect of the true counterfactuals of the actual world that God's knowledge of them needs to be (for those whom he calls the Dominicans) logically subsequent to his decree to create that world. There are hosts of true counterfactuals of nonactual worlds that God may know logically prior to his decree to create some world. These are, so to speak, double-barreled counterfactuals, of the following form: If God were to decree some world containing Jones and Smith, then God knows (say) that if in that world Jones were to have met Smith, they would have become fast friends.
It's not clear to me how middle knowledge of the Jesuit variety-according to which God's counterfactual knowledge of the actual world is logically prior to his decree to create that world-helps resolve the question of freedom and foreknowledge. Craig supposes that there is a possible world in which Peter affirmed Christ in precisely the same circumstances in which he denied him. But God wants the possible world in which Peter denies Christ to be actual. So what does he do? Can he decree that possible world? How can God be assured that when he decrees that possible world in which, in circumstances C, Peter freely denies Christ, it is that world that is actualized? In addition, simply calling some of what God knows "middle knowledge" does not show us how God can know what Peter would freely do given the actual presence of Peter in circumstances C. For it appears (to me at least) that God cannot, by employing his middle knowledge, decree the world in which Peter denies Christ without ensuring that Peter denies Christ, thus compromising Peter's indeterministic freedom so vital to the Molinist and Arminian scheme of things.
In his philosophical defense of Molinism toward the end of his paper, Craig claims, without argument, that there are true counterfactuals of creaturely freedom; he makes this claim on the grounds that this seems plausible, and that if there are true counterfactuals of creaturely freedom, then God knows them in virtue of his omniscience (p. 139). But it only seems plausible if divine foreknowledge and indeterministic human freedom are compatible. And isn't this the question at issue? It seems to me that either the Molinist scheme begs that question or the problem remains to be faced. Luis de Molina himself sought refuge at this point in the idea that God (alone) "supercomprehends" the propositions that form the content of his middle knowledge-that he alone knows infallibly that, for example, in the actual world in circumstances C, Peter will freely deny Christ.' But how "supercomprehension" works, how it successfully effects the reconciliation between divine foreknowledge and human freedom, remains to this day a well-kept secret.
Fatalism
Craig does not define fatalism, so at the start of his piece it is not at all clear what he has in mind. However, later on "fatalism" becomes something called "theological fatalism," and later on still the "Augustinian-Calvinist perspective." But it is pretty clear that it is inaccurate to tar the Augustinian or Calvinist with this particular brush. For they hold that the divine decree by which all things come to pass is logically contingent. God might have decreed otherwise. Moreover, in bringing his decrees to pass, God employs means-means of which he himself is the primary cause.' So, to take a standard case, when Paul receives divine assurance that both he and his companions onboard ship will be saved (Acts 27:24), they are saved only in virtue of the fact that they did not take to the lifeboats (Acts 27:31). We are entitled to draw the inference that had they taken to the lifeboats, they would not have been saved. So the parallel with ancient Greek fatalists that Craig attempts to draw is far from exact (p. 128). In any case, one of the things shown by the tortuous debates on accidental necessity, the alleged distinction between hard and soft facts, and so on is that the issue of divine foreknowledge and freedom is not equivalent to the ancient problem of fatalism. So even if fatalism is refuted along the lines that Craig suggests, there would still be a problem raised by the existence of an infallible foreknower of the future.
Craig suggests that there is an air of mystery about the prima facie incompatibility of divine foreknowledge and human indeterministic freedom. If there is this incompatibility, then there is "some sort of logical constraint . . . my actions are mysteriously constrained" (p. 130). But there is nothing in the least mysterious about this. If God's foreknowledge is consistent with indeterministic freedom, then we may be indeterministically free. If it is not consistent with indeterministic freedom, then we may be free in some other sense or perhaps not free at all. What is needed at this point are arguments for the compatibility of divine foreknowledge and indeterministic freedom-arguments that, oddly, Craig does not provide, nor does he attempt to refute their incompatibility. Occasionally he suggests the beginnings of an argument, as in his question of whether mixing temporal and logical necessity is valid, or of whether a future event could be temporally necessary, or of whether an action's being temporally necessary is compatible with an action's being temporally free (p. 130). If he thinks that none of these claims is valid, then he should suggest some solid reasons why he thinks this.
At length Craig does offer an argument, one about the causal closedness of the past, which he says is what temporal necessity is or implies (p. 131). Suppose it is. Of course the mere causal closedness of the past does not imply that our actions are determined, but the causal closedness of a knowledge claim or a divine belief about the future might well. And that's what the argument is (or should be) about.
God as the Author of Evil
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 accord ing 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 sin (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.