The Simple-Foreknowledge View
2
THE SIMPLE-FOREKNOWLEDGE
VIEW
David Hunt
Astandard affirmation among Christians is that God is all-knowing, or omniscient. What this means, in short, is that God knows all truths and believes no falsehoods. The doctrine of divine omniscience, like most doctrines, is certainly subject to interpretation, and the consensus in its favor masks considerable disagreement over details. But there are at least two points of interpretation on which the vast majority of believers would surely agree. First, in knowing everything, God doesn't just happen to get things right, so that his omniscience is at bottom a matter of luck; instead, it's his very nature to be all-knowing, so that it's not even possible for him to be mistaken in any respect. Second, when it is said that God is all-knowing, the "all" refers to the future as well as the past and present: God's knowledge contains no gaps that must await future developments before they can be filled in. Because most Christians understand God's omniscience this way, most subscribe to the doctrine of divine. foreknowledge:
(F) God has complete and infallible knowledge of the future.
The present chapter is dedicated to defending this traditional doctrine.
Why does (F) even need to be defended? There are in fact a number of dif ficulties that have been raised against this doctrine. Three of these seem to me to be particularly important, and the bulk of this chapter will be devoted to addressing them. Each comes in the form of an argument designed to show that (F) is incompatible with some other truth to which Christians are deeply committed. The first purports to prove that (F) rules out human beings' possession of genuine free will; the second alleges that (F) is incompatible with the idea that God himself is a personal being capable of intentional action; and the third argues that (F) fails to enhance God's providential position in any way, thus depriving God of what should be the principal theological benefit of foreknowledge. I will refer to these three objections to (F) as the Problem of Human Freedom, the Problem of Divine Agency and the Problem of Divine Providence, respectively.
Of course many who find one or more of these objections persuasive are atheists who see, in the difficulties they raise for (F), yet another reason for rejecting the theistic God. But others are Christians for whom the apparent intractability of one or more of these problems provides a motive for tinkering with (F)-while trying to keep the tinkering within theologically acceptable limits! There are two general directions that such tinkering can take. One is toward curtailing divine foreknowledge to exclude the problematic cognitions (in particular, knowledge of so-called future contingencies).' Call this diminished foreknowledge. The other direction is taken by those who insist that the three problems can be resolved without diminishing divine foreknowledge but who argue that this can be done only if the manner in which God knows future events is elaborated in some special way-say, by grounding his foreknowledge in "middle knowledge" or by removing God from time so that he can view the past, present and future from an atemporal vantage point.' Call this augmented foreknowledge. Advocates of diminished foreknowledge deny (F), while the partisans of augmented foreknowledge accept (F) and then add to it.
It is important to understand that this defense of simple foreknowledge, if successful, will not show that diminished or augmented foreknowledge is wrong; indeed, I have myself a good deal of sympathy for some versions of augmented foreknowledge. What a successful defense would show is that any alternative to simple foreknowledge must be justified on grounds other than its ability to escape the Problem of Human Freedom, the Problem of Divine Agency and the Problem of Divine Providence. These problems do not by themselves provide compelling reasons for embracing a diminished or augmented alternative to simple foreknowledge, since they can all be solved with the resources of simple foreknowledge alone.
The Case for Divine Foreknowledge
Before stating and analyzing these three problems for (F), we need to have a better idea of what is at stake. The mere fact that most believers do accept (F) does not mean that they should accept it. What then can be said on behalf of this doctrine?
There are four points that seem to me to provide powerful initial support for (F). In the first place, there is considerable biblical evidence for divine foreknowledge. God's ability to know what will happen before it happens is integral to who he is; Isaiah even makes it the chief mark by which the true God may be distinguished from false gods (Is 41:22-23). The reported instances in which God exercises this power are so numerous (and familiar) that only a couple of examples will be mentioned here.
The first example illustrates something of the variety of events accessible to divine foreknowledge. In Genesis 40 Joseph recognizes in his fellow prisoners' dreams a divine communication regarding Pharaoh's future decision to restore the chief butler and condemn the chief baker; then in the next chapter Joseph interprets Pharaoh's own dream to foretell seven years of plenty and seven years of famine. In this pair of examples we see God's foresight encompassing both the outcome of Pharaoh's most intimate thought processes and the longterm effects of large-scale climatic forces. The second set of examples, worth including on any short list of "proof-texts" because the knower in this case is God's own incarnate Son, is taken from the first half of Matthew 26. Here we find Jesus predicting, in quick succession, that he will be betrayed on the feast of the Passover (v. 2) by a disciple (v. 21) "who has dipped his hand into the bowl with me" (v. 23); that it is Judas in particular who will betray him (v. 25); that his other disciples will desert him (v. 31); and that Peter will deny him three times before morning (v. 34)-none of which looks like it could be foreknown with certainty in the absence of some supernatural insight into the future.
In addition to dozens of specific examples like these, the Bible contains many summary claims about the incredible range of God's foreknowledge. The psalmist notes how God knows beforehand the least significant things he is going to do ("Even before a word is on my tongue, 0 LORD, you know it completely," Ps 139:4), while another passage from Isaiah speaks of the cosmic sweep of divine foreknowledge ("I am God, and there is no one like me, declaring the end from the beginning and from ancient times things not yet done," Is 46:9-10). In short, the doctrine of divine foreknowledge appears firmly rooted in the plain meaning of scores of biblical texts.
So much for the first reason for accepting (F). The second reason is that the central Christian understanding of God as a supreme and perfect being arguably requires (F). There is at least some dissonance (if not an outright contradiction) in our thinking about God if we suppose him to be unsurpassably great while also attributing to him ignorance of some future event. Unless we affirm (F), we can't say (as we surely want to say) that God is as great as a being can possibly be.
The third point in favor of (F) is that divine sovereignty and providence require God to be supremely knowledgeable as well as supremely powerful. There is, to be sure, considerable theological controversy over how much control God does exercise: the Bible reveals a God who is "in control" in some pretty robust sense but not "controlling" in the manner of a puppeteer or ventriloquist. But we needn't enter this controversy, since even those who minimize God's control over events typically insist that God has the ability to control everything-it's just that he freely refrains from exercising that ability (for familiar reasons). If he has so much as the ability, however, he needs the knowledge, since power without knowledge is blind.
Fourth and finally, the overwhelming consensus of the Church's leading thinkers is that (F) is correct. The first Christian philosopher on record, Justin Martyr, writing in the second century, suggests that God is delaying the Last Judgment because "in His foreknowledge He sees that some will be saved by repentance, some who are, perhaps, not yet in existence. ,4 St. Augustine, sometimes referred to as the "third founder" of Western Christianity (after Christ and Paul), is quite unequivocal: "For to confess that God exists, and at the same time to deny that He has foreknowledge of future things, is the most manifest folly.... For one who is not prescient of all future things is not God. ,5 The same position is taken by all the leading medieval theologians, like Boethius, St. Anselm, St. Thomas Aquinas and so on.6 Martin Luther and John Calvin both subscribe unquestioningly to (F), though their tendency is to mention divine foreknowledge only in connection with God's sovereign power, as when Luther writes, "It is then essentially necessary and wholesome for Christians to know that God ... foresees, purposes and does all things according to His immutable, eternal and infallible will." 7 But (F) is also widely attested by those, like John Wesley, who do not go as far as the Reformers in their denigration of human initiative: "God foreknew those in every nation who would believe, from the beginning of the world to the consummation of all things."" This list could, of course, go on and on. 9 While lacking the authority of Scripture, the near unanimity of such figures surely counts for something, at least among those of us who identify to one degree or another with "the tradition."
It should be admitted, even by critics, that the case for (F) is indeed impressive. But at the same time it is important to note that none of the four reasons just offered on behalf of (F) is absolutely irresistible. Let's take them in order.
The first reason was the biblical evidence supporting (F). But the fact is that the Bible nowhere makes this claim in precise and unambiguous terms. While (F) is certainly a natural inference from the scriptural evidence, it is an inference. This inference begins with the biblical data-particular instances of foreknowledge, like God knew in advance that Pharaoh would pardon the chief butler, and collections of instances, like God knew in advance all the things the psalmist was about to say-and generalizes from them to the universal claim God knows in advance absolutely everything that will ever happen. But such inferences are at least potentially vulnerable to the fallacy of "hasty generalization."
It's also possible that the data on which the inference is based are flawed. At least some cases in which foreknowledge is attributed to God may reflect nothing more than his unexcelled ability to extrapolate from current conditions or his present awareness of what he himself intends to do, rather than reflecting an irreducible power to apprehend the future. So, for example, God might foresee the seven years of plenty and seven years of famine in Egypt not because he espies this event as through a "time telescope," but because his knowledge of meteorological conditions makes him a weather forecaster with out equal or because he has decided to manipulate the weather to this end.
Other predictions might be best understood conditionally, conveying information about what will happen if things keep going in the direction in which they are now headed. A critic of (F) might even suggest that Jesus' prediction of Judas's betrayal really has the following force: "I know you very well, Judas-better even than you know yourself-and you are going to betray me unless you change your present course." (It's not always easy to distinguish a simple prediction from a conditional one, as Jonah found out when God failed to destroy Nineveh as expected.) It doesn't seem very plausible that all the proof-texts for divine foreknowledge can be explained away in such fashion, but the critic of (F) at least has an opening to make the attempt.
The second reason offered on behalf of (F) was the contribution such knowledge makes to divine perfection. But while the Bible does attribute to God a very impressive range of properties, it nowhere sums up these properties into the single concept of a perfect or maximally excellent being, strictly construed. Where we do find this concept featured prominently and explicitly is in Neo-Platonism, a pagan philosophical system. Perhaps then the idea of an unexcellably perfect being is an unscriptural phantom of reason, and we put too much stock in our own intellectual powers when we endorse it and then endeavor to derive from it the doctrine of divine foreknowledge.
The third reason rested on the general usefulness of knowledge. The idea was that additional knowledge gives the knower a comparative advantage when it comes to achieving his ends, and that exhaustive knowledge would therefore give God a superlative advantage (without which he could not qualify as God). What this overlooks is that greater knowledge contributes to greater control only when the knowledge is relevant to the achievement of control. There may, however, be many facts about the future that are irrelevant to God's present purposes, and knowing these facts would endow him with no practical advantage. It appears, then, that our interest in maintaining divine providence and sovereignty might be satisfied with something weaker than M.
The fourth and final reason was simply the weight of the traditional consensus in (F)'s favor. Unlike God, however, the authorities cited earlier are not infallible: maybe they just got this one wrong. Perhaps the whole tradition became implicated in this error when the church fathers fell under the spell of Greek philosophy and began to see the Scriptures through an alien conceptual lens, with the result that (F) became a self-perpetuating mistake that none of these great thinkers had the insight to challenge. That's at least possible, even if we think it's pretty unlikely.'o
So where does this leave us with respect to (F)? We have seen that a strong, broad-based case can be made on behalf of M. We have also seen that each reason making up this case has a point at which it is potentially vulnerable. Of course a merely potential vulnerability would have to be exploited successfully before giving us any grounds for abandoning (F). I leave it to the opponents of (F) to exploit the vulnerabilities we have identified-I won't do their job for them! Notice, though, that even a successful attack on all four of the reasons offered on behalf of (F) wouldn't show that (F) is false; it would only show that (F) is not supported by these particular reasons, leaving it an open question whether it derives support from other reasons (or is true without being backed by reasons at all). The crucial issue, given the prima facie case that has been made on behalf of (F), is whether there are any good reasons for thinking that (F) is false.
This is where the Problem of Human Freedom, the Problem of Divine Agency and the Problem of Divine Providence come in. Let us then turn to a consideration of these three problems. If any of them does succeed in rebutting (F), we might at that point wish (as part of our review of where we went wrong) to return to the potential vulnerabilities in the case for (F), ruefully admitting our errors: "Aha, so we did misinterpret those proof-texts," "We were mistaken in our concept of God and the inferences we made from it," and so on. But until that happens, we are within our rights in regarding (F) as presumptively true.
The Problem of Human Freedom
1. Formulating the problem. The Problem of Human Freedom was recognized as a serious difficulty almost as soon as Christian beliefs began attracting logical scrutiny. Because of its long history and its centrality to the concerns of this book, it will occupy more of our attention than the other two problems.
Let us follow the problem as Augustine formulates it in his dialogue On Free Choice of the Will. It is Augustine's interlocutor, Evodius, who first broaches the subject:
I very much wonder how God can have foreknowledge of everything in the future, and yet we do not sin by necessity. It would be an irreligious and completely insane attack on God's foreknowledge to say that something could happen otherwise than as God foreknew. So suppose that God foreknew that the first human being was going to sin. Anyone who admits, as I do, that God foreknows everything in the future will have to grant me that. Now ... since God foreknew that he was going to sin, his sin necessarily had to happen. How, then, is the will free when such inescapable necessity is found in it?"
In the first sentence of this passage Evodius affirms that divine omniscience extends to the future; in the second sentence he emphatically rejects the idea that God's foreknowledge could ever be mistaken. So Evodius is clearly committed to (F). He then proposes to apply this assumption to a particular example: the primal sin by which Adam and Eve turned their wills from the command of God to the prompting of Satan.
The problem this raises is developed in the last two sentences of the passage, which can be put in argument form as follows:
(1) If God foreknows that Adam will sin, then it's necessary that Adam will sin.
(2) If it's necessary that Adam will sin, then Adam will not sin of his own free will.
If God foreknows that Adam will sin, then Adam will not sin of his own free will.
This conclusion is not explicitly stated in the text, but it's present by implication. (An argument with an implied but unstated conclusion is called an enthymeme.) Nor does Evodius bother to spell out the implications of the argument for agents other than Adam. Since the argument does not turn on any features peculiar to this particular example (something distinctive about Adam, the conditions under which he was acting, etc.), a similar argument can be constructed for any human action. So the argument's full ramifications are these: if (F) is true, no one ever does anything of his own free will.
Why does Evodius, as a Christian, find this conclusion so disturbing? Clearly Evodius would not be troubled by the argument if he thought that God has foreknowledge but humans lack free will or if he thought that humans have free will but God lacks foreknowledge. Instead, Evodius (like most Christians) believes both of the following:
(3) God foreknows that Adam will sin.
(4) Adam will sin of his own free will.
But these are inconsistent with the conclusion of the above argument. Augustine himself recapitulates the problem for Evodius like this: "And so you fear that this argument forces us into one of two positions: either we draw the heretical conclusion that God does not foreknow everything in the future; or, if we cannot accept this conclusion, we must admit that sin happens by necessity and not by will."" If the argument based on (1) and (2) is accepted, we must reject either (3) or (4). But if we want (with Evodius) to keep both (3) and (4), we must find something wrong with the argument. Since the argument is formally valid (it is a "hypothetical syllogism," exemplifying the valid inference form A-*B, B-~C, thereforeA-*C), the argument can be rejected only by rejecting one of its premises, (1) or (2).
Our situation, then, is this: we must find a principled way to deny (1), (2), (3) or (4); more specifically, as Christians, we must find a theologically acceptable way to deny one of these propositions. What has given this problem its hoary status as one of the great philosophical puzzles is that it's just not obvious which of these propositions can be rejected. Premise 3 is not a promising candidate, since it certainly looks like all the reasons that were marshaled in favor of (F) provide equal support for (3). After all, if (F) is true, God surely doesn't have complete and infallible knowledge of the future at some times and not at others; God has such knowledge at all times. Since he has it at all times, he has it the day before he creates Adam, when Adam's sinning is yet future; but then it follows that (3) is (then) true. It's also hard to see how (4) could be denied. Adam's sin (like any sin) is surely blameworthy-if it weren't, it would be difficult indeed to make sense out of the bulk of what the Bible has to say about sin. But we are blameworthy for our actions only to the extent that we do what we do of our own free will. (When my free will is abridged- e.g., because I'm overwhelmed by an external force, I'm acting under duress, I'm suffering a muscle spasm, I'm paralyzed with fear, I don't know what I'm doing, etc.-my responsibility is correspondingly diminished.) So how could (4) fail to be true?
2. Accidental necessity. If God foreknows that Adam will sin, then Adam will sin-that's beyond dispute. But in what sense is it necessary (given God's foreknowledge of it) that Adam will sin? It isn't logical necessity that is in question here: Adam's refraining from sin, unlike his marrying the number 47 or discovering the best way to barbecue a rainbow, does not violate any strictly logical principles. Rather, (1) concerns what the medieval thinkers called accidental necessity. Since the word accidental in this context is roughly synonymous with contingent (i.e., something that is the case but might not have been the case), this phrase may strike some as an oxymoron: how could something that is accidental be necessary or something that is necessary be accidental?
Here's how. Conditions that are not themselves necessary (because they could have turned out differently) may, once those conditions are in place, limit what else is possible. Suppose, for example, that it is not necessary that I pitch this stone over the side of this cliff; I can prevent myself from doing so, and if I nevertheless go ahead and throw the stone, this is still something I could have prevented myself from doing. My launching the stone into space is an "accident"-not, of course, in the ordinary English sense of "unintended" or "fortuitous" (let's assume I threw the stone on purpose) but in the sense that things did not have to go this way. Once they have gone this way, however, certain things that were in my power (before I threw the stone) are no longer in my power.
In the first place, having already thrown the stone, it is no longer possible for me to undo or revoke the fact that I threw the stone; it's too late for that. In general, everything that is past is (now) necessary in this sense. (This idea is captured in such common expressions as "What's done is done," "There's no use crying over spilt milk," and so on.) In the second place, there is nothing I can now do to prevent the consequences of this past event, such as the fact that the stone will soon be lying at the bottom of the canyon. If I realize as it disappears into the chasm that I just threw away a gold nugget, I can only regret this result, which is now, quite literally, out of my hands. At this point in time, between the throwing of the stone and its landing on the rocks below, both the throwing of the stone (a past event) and its arrival at the bottom of the cliff (a future event) are irrevocable.
This is roughly the situation that (1) alleges to obtain with respect to God's antecedent knowledge and Adam's subsequent action. Imagine that Adam has just been formed by God from the dust of the ground. When he comes into existence, God already knows that he will disobediently eat of the tree of the knowledge of good and evil. This is not something God knows as a matter of logical necessity; if something else had been true, God would instead have known that. But given that he knows this and that he has known it from eternity, there is nothing Adam can now do to prevent (or "postvent"!) this fact. He's come on the scene too late for that. But if Adam is stuck with the fact that God already knows he will sin, he's also stuck with the consequences of this fact. A notable consequence of God's foreknowing that Adam will sin is that Adam will sin; so Adam is stuck with this fact as well. God's prior knowledge, which Adam is stuck with because it was in place before he even began to exist, does not leave open the possibility that Adam will act other than God knows he will act. As Adam opens his eyes and gazes upon Eden for the first time, both God's (past) foreknowledge that he will sin and Adam's (future) sinning confront Adam as fixed and irrevocable.
One caveat must be issued before we accept this justification of (1). The only reason offered for thinking that Adam's future sinning is unavoidable is that it is entailed by something else that Adam can't avoid (namely, God's past foreknowledge of his sinful choice), and the only reason offered for thinking that this is unavoidable for Adam is that it belongs to the past. But surprising as it may seem, it is not unqualifiedly true that God's knowing yesterday what will happen tomorrow is a fact about the past. To say that someone knows something is (in part) to say that the person believes something that is true. Both truth and belief are necessary for knowledge. You can believe something that is false, but you can't know something that is false-only truths can be known. But the mere fact that something is true doesn't mean you know it-you must actually believe this truth if knowledge of it is to be credited to you. This has important implications for God's prior knowledge that Adam will sin. To say that God knew that Adam will sin is to say at least this: (i) God believed that Adam will sin and (ii) it is true that Adam will sin. Whereas (i) is about the past, (ii) is about the future; so the fact that God knew that Adam will sin is partly about the past and partly about the future.
Let us say that a "soft fact" about the past is one whose grammar points to the past (e.g., through use of the past tense) but whose content belongs at least in part to the future, whereas a "hard fact" about the past is a fact that is entirely about the past. God knew that Adam will sin is therefore a soft fact about the past. This is significant because a soft fact, unlike a hard fact, is not accidentally necessary just by virtue of being about the past; since a soft fact concerns (at least in part) the future, then unless there is some independent reason to think that the future it points to is inevitable, the fact as a whole might yet turn out false. This undercuts our only reason for thinking that God's prior knowledge of Adam's sin is accidentally necessary at the time Adam comes into existence, and so undercuts our only reason for thinking that the consequences of God's knowledge, including the fact that Adam will sin, are fixed and inevitable.
Accidental necessity is by no means a straightforward notion, but once we understand what is being said in (1), it is very hard to see how it could be denied. We can now treat (2) more briefly, since we have already done the work required to understand what the first part of this premise is asserting. If it is necessary that Adam will sin-if, that is, it is now accidentally necessary (irrevocable, unavoidable, not-able-to-be-other-than) that Adam will sin-then Adam will not sin of his own free will. The idea, surely, is this: "free choice" is a sham if there is really only one course of action open to an agent. But this is exactly the situation facing Adam: from the moment he first drew breath, it was accidentally necessary that he would eat of the tree of knowledge. The Fall is "locked in" by God's prior knowledge, and Adam is stuck with the results. So Adam does not sin of his own free will, given that God's beliefs before Adam even came into existence precluded any genuine possibility that Adam might not sin.
Well, there we have it: (1) and (2), no less than (3) and (4), appear to be rock solid. But they are mutually contradictory; hence at least one of them must somehow be false! Which one could it possibly he?
3. The Boethian and Calvinist solutions. It may surprise the reader to learn this late in the game that every one of these propositions, which we have been at such pains to justify, has in fact been the subject of serious criticism. A moment ago we were despairing over how to select one of these seemingly undeniable propositions for rejection; now we find ourselves being overwhelmed by an excess of candidates! Let's begin our review of how we could have misread the situation so badly with some comments on (3) and (4).
Not only are (3) and (4) more problematic than they appear, but most of the figures cited earlier as part of the traditional consensus favoring (F) dissent from one or both of these propositions. Ironically, even Evodius's mentor, Augustine, withheld his full endorsement from (3)-the very thing he tells Evodius it would be heretical to deny!-and later came to withhold it from (4) as well. How is this possible? Was Evodius simply mistaken in thinking that he as a Christian needed to affirm both (3) and (4)?
With respect to (3), those who reject (F) in favor of diminished foreknowledge would of course regard Evodius as mistaken. But denying (3) because one denies (F) is obviously unhelpful to the business of defending (F) against the Problem of Human Freedom! Is there another ground for denying (3) that might contribute toward the defense of (F) rather than presupposing its falsity? Well, suppose that God knows the future without foreknowing it. This would be the case if God (and his knowledge) exists outside of time. On this conception of God he could still know everything that is yet future relative to us; but since he is not himself in time, these things would not be future relative to him, and so he would not know them as future. This was in fact Augustine's view of God, as he makes clear elsewhere.15 But since it is the Late Roman philosopher and theologian Boethius who gave this position its classic formulation, I shall refer to it as the Boethian view.
And how exactly does the Boethian move help matters when it comes to the Problem of Human Freedom? As Boethius himself explains it, an atemporal deity, "embracing all the infinite spaces of the future and the past, considers them in his simple act of knowledge as though they were now going on." Since God's knowledge of Adam's sin does not precede the sin, it's not part of a fixed past dictating how Adam's future decisions and actions must go. "Why then do you require those things to be made necessary which are scanned by the light of God's sight, when not even men make necessary those things they see? After all, your looking at them does not confer any necessity on those things you presently see. ,16 God sees everything at once from a vantage point outside time and history, and this is no more deleterious to the freedom of the actions God observes from this vantage point than is your sitting on a hill observing the picnickers, kite-flyers, baseball players and so on at the nearby park deleterious to their freedom. The Boethian solution, in short, responds to the Problem of Human Freedom by denying (3), yet it does so in such a way that (F) is preserved.
As for (4), most of those cited in support of (F) would also resist this proposition to one degree or another, if not deny it outright. On Free Choice of the Will, ostensibly written in defense of the will, is as much about its bondage as it is about its freedom, with its final chapters devoted to the postlapsarian "ignorance and difficulty" by which our wills are hampered. Augustine's doubts about human autonomy hardened further in the course of his involvement in the Pelagian controversy, until he could write at the end of his career: "On the solution to this question I tried hard to maintain the free decision of the human will, but the grace of God was victorious."" Since Calvin is popularly associated with antipathy toward (4), let us call this the Calvinist solution.
The Calvinist position is rooted in our utter dependence on God and our utter incapacity to do anything good without the aid of divine grace. If God is truly sovereign, what room is there for "little gods" exercising their own autonomous agency over against God? Few proponents of this position are as intemperate in their rejection of (4) as is Luther, who ends The Bondage of the Will by declaring that "free will lies vanquished and prostrate."18 But even where free will is frankly asserted, as in Aquinas, it is often so subsumed under divine causality and hedged about with "enabling" and "effective" grace that it's hard to know how (4) can continue to be affirmed in any straightforward sense.'
Should we jump at the chance to evade the Problem of Human Freedom that is offered us by the apparent deniability of (3) and (4)? I don't believe we should. My reticence on this score is not, however, rooted in the assumption that (3) and (4) are both true. Just for the record, I rather like the atemporal view of God, for various reasons I won't go into here; so I am inclined to join Augustine and Boethius (not to mention Anselm, Aquinas, et al.) in denying (3) while affirming (F). (I am not inclined to deny or seriously qualify premise (4) just because I don't understand free will clearly enough to see that other things to which I am committed are really incompatible with it.) But denying (3)-or (4), for that matter-does not yield a satisfactory solution to the Problem of Human Freedom, even if the proposition denied is in fact false. Let me explain.
Sometimes we have good reason to believe that there must be something wrong with an argument before we are in a position to see what is wrong with it. Consider, for example, the most famous of Zeno's paradoxes of motion, in which it is argued that Achilles, renowned for his speed, could not win a race against a tortoise if the tortoise were given a head start, no matter how small. This is because Achilles, to pass the tortoise, would first have to catch it, and this would involve his traversing an infinite number of distances: first, the distance to the tortoise's initial position x; then the distance to the new position x + j the tortoise reached while Achilles was getting to x; then the distance to the new position x + j + k the tortoise reached while Achilles was getting to x + j and so on. One way that Zeno's problem can be formulated is this:
(a) If Achilles can overtake the tortoise, then it's possible for someone to finish traversing each of an infinite number of intervals.
(b) If it's possible for someone to finish traversing each of an infinite number of intervals, then it's possible to perform an infinite number of tasks in a finite amount of time.
(c) It's not possible to perform an infinite number of tasks in a finite amount of time.
Achilles cannot overtake the tortoise.
And what's wrong with that? Well, we might be pretty sure (at least prior to encountering Zeno's argument!) that
(d) Achilles can overtake the tortoise.
Since the denial of (d) follows from (a)-(c), one of the propositions (a)-(d) must be false.
Suppose now that someone were to propose that the problem can be solved by rejecting (d). Wouldn't this be utterly preposterous? Of course it would; but be careful why you regard it as preposterous. Perhaps it's because you are so sure that (d) is true: it's preposterous to address Zeno's argument by denying (d) because it's preposterous to deny W. But suppose archeologists digging near Mycenae unearth a tablet, dated circa 700 B.C., bearing the title (in archaic Greek) "The True Story of Achilles' Heel." It turns out that Homer's references to the "fleet-footed Achilles" were just a joke: the real Achilles had an enormous bone spur on one heel, making it too painful for him to walk unaided, let alone run. So (d) is false after all. Does this solve Zeno's paradox? Of course not. What is paradoxical is that you could begin with someone believed to be quite fast-faster than a tortoise, at any rate-and then demonstrate, in the way Zeno does, that this person could not outrun a tortoise after all. Denying (d) simply removes Achilles from complicity in this paradox; it does nothing to resolve it. Whether or not (d) is (also) false, we can be confident that something is seriously wrong with (a)-(c), even if we're not sure just what it is.
It seems to me that we face exactly the same situation when it comes to the argument for the incompatibility of divine foreknowledge and human freedom that Augustine puts in the mouth of Evodius. If God had attached tiny wires to Adam so that he could manipulate him as one would a marionette or if he had caused a fog to settle over Adam's mind so that he didn't know what he was doing, then we could readily see how this might change Adam's circumstances so that he did not eat the apple of his own free will. But there is something fishy about the idea that Adam's action, while in every other respect satisfying the most exacting requirements for free will, might nevertheless count as unfree simply because God foreknew what he would do.
I am therefore in hearty agreement with the way one of my coauthors, William Lane Craig, put the point in an earlier book:
Denying (3) or (4) does not get to the nub of the problem raised by Evodius's argument any more than denying (d) in the case of Zeno's argument doeseven if (3), (4) and (d) are all false. Because Zeno's argument is so dubious, it cannot provide a reason for denying (d); even if there is some independent reason for denying (d) (e.g., the unearthed tablet), we know that there is trouble somewhere in (a)-(c) as well. Likewise, because Evodius's argument is so dubious, it cannot provide a reason for denying (3) or (4); even if there is some independent reason for denying (3) or (4)-of the sort canvassed by Augustine and friends-we have good reason to suspect trouble somewhere in (1) and (2) as well.'
I believe it is precisely because Augustine understood this that he never drew on his own doubts about (3) and (4) when formulating a response to the Problem of Human Freedom. We will return to Augustine's analysis of the problem in a moment. But first, let's reexamine the prospects for rejecting (1), since the irrelevance of (3) and (4) means that the fallacy must lie either here or in (2).
4. The Ockbamist solution. The defense of (1) offered in section two depended crucially on Adam's future sinning's being entailed by a hard fact about the past, a fact that Adam was stuck with from the moment he first drew breath. We saw that while God's foreknowing that Adam will sin does not qualify as a hard fact, this fact includes God's forebelieving that Adam will sin; and this latter fact does look like a hard fact that entails Adam's future sinning. The fourteenth-century philosopher and logician William Ockham nevertheless maintained that appearances here are deceiving.22 The fact that God believes that Adam will sin, he averred, is also a soft fact, remaining "soft" until Adam actually sins; and if this fact is unavailable to Evodius's argument, we can safely conclude that God's foreknowing that Adam will sin does not involve any hard fact which both entails and also predates Adam's sinning. So the argument breaks down at (1).
This Ockhamist position has developed into what is probably the major strategy used by contributors to the contemporary philosophical debate in their attempt to evade the Problem of Human Freedom.'3 But given our earlier reasons for thinking that God believes that Adam will sin is indeed a hard fact, on what grounds do Ockham and his modern followers assert the contrary? There are three principal ways that the Ockhamist position might be defended.
Strategy 1. One way to argue that God believes that Adam will sin is a soft fact is to do so in the same way we argued that God knows that Adam will sin is a soft fact, namely, by first unpacking what it is for God to hold a belief about the future and then showing that this fact is constituted not just by what is now the case but also (in part) by what will be the case later. This direct approach can succeed, I believe, only if it can be shown how one and the same state of the divine mind-the one that, as matters actually stand, constitutes God's believing that Adam will sin-would instead count as a different belief (namely, the belief that Adam will not sin) if Adam does not sin.
While this may seem pretty dubious, there is at least some theoretical space for this maneuver. Linda Zagzebski has formulated a position she calls "Thomistic Ockhamism" under which God's cognitive act is part of his very essence; John Fischer has suggested applying Hilary Putnam's notion of "wide content" to divine foreknowledge; and I have laid out a "dispositional omniscience scenario" in which God knows everything without actively thinking about it all the time.24 These thumbnail descriptions will convey very little to the reader unfamiliar with the relevant literature; I offer them only as evidence that such approaches exist. But if any succeeds, the Ockhamist is home free; for if God's belief is constituted as the belief that Adam will sin only retroactively, once Adam actually sins, then nothing about God's prior belief would appear to be inconsistent with either Adam's freedom to sin or his freedom not to sin.
It is noteworthy that all three of the suggestions for implementing this strategy mentioned in the preceding paragraph represent attempts by critics of Ockhamism to understand how God's beliefs about the future could possibly be "soft."25 None of the leading defenders of Ockhamism has invested in this strategy.26 Perhaps they regard it, with reason, as a poisoned pawn, attractive for its offer to make the Ockhamist thesis intelligible but fatal in its theological implications: after all, if Jesus' prior belief about what Judas will do is constituted as the belief that Judas will betray me only retroactively, once Judas actually betrays him, then it's pretty baffling how Jesus could make the prediction reported in Matthew 26:25 prior to Judas's betraying him. But the Ockhamists' lack of interest in this first strategy may also reflect nothing more than their confidence in the remaining two strategies. These other strategies derive from two criteria proposed by Ockham himself, under which a fact is soft in virtue of how it is related to certain other facts: "that proposition that is about the present in such a way that (i) it is nevertheless equivalent to one about the future and (ii) its truth depends on the truth of the one about the future does not have [corresponding to it] a necessary proposition about the past."27 Let's see how (i) and (ii) can be parlayed into critiques of (1).
Strategy 2. Some Ockhamists have argued that, however difficult it is to understand how God's forebeliefs could be soft, we can be confident that they are soft, since there is a simple argument by which this can be demonstrated. If this is so, then the fact that we can't understand what divine belief could possibly be if it is to count as soft is irrelevant-there's a lot we don't and indeed can't understand about God, and this is one of those things.
But can it be shown that God's forebeliefs are soft? Here's the argument. God's infallibility ensures that
God believed that Adam will sin -> Adam will sin while his omniscience guarantees that
Adam will sin - God believed that Adam will sin It follows that
God believed that Adam will sin H Adam will sin
I do not find this argument at all convincing. If God believed that Adam will sin and Adam will sin were the same fact, it would be easy to see why they would both have to be soft (if either were soft) or both hard (if either were hard). But they are obviously different facts: the first is about what God believed and the second is about what Adam will do. The right conclusion to draw, it seems to me, is that a hard fact can be logically equivalent to a soft fact when these are not the same fact, as in the present case. I therefore reject this second defense of Ockhamism.29
But this argument is again unconvincing. Let us agree that were Adam to do otherwise, God would have believed otherwise; the problem is that the foregoing argument provides no reason to think that Adam can do otherwise. Suppose I fail to save a child from a burning building because I am knocked unconscious by a falling beam. Then it's true that, had I saved the child, I would not have been knocked unconscious; but this does nothing to alter the fact that, as matters actually stand, I could not save the child. Nor is the result any different when the past is dependent on the future, as in the foreknowledge case.
Imagine that tachyons (particles that travel backward in time) have been adapted for household use and that my friend the tycoon has installed one of the new tachyonic doorbells at his mansion. Arriving for a dinner party, I press the button outside the front door; this causes a buzzer inside the house to sound five minutes earlier, giving the butler a headstart in getting to the door. Now let's go back to my arrival at the door. Is it possible for me to refrain from ringing the bell, given that it has already sounded? Suppose you say yes, justifying your answer as follows. Since the buzzer's having sounded depends on my pressing the button, if I were to refrain from pressing the button, the buzzer would not have sounded. My alternatives, then, are to press the button (in which case the buzzer sounded) or to refrain from pressing the button (in which case the buzzer did not sound): both possibilities are open to me. What this justification overlooks, however, is that the buzzer has already sounded; so the state of affairs that consists of my refraining from pressing the button (and the buzzer's not having sounded), a state that was unproblematic enough before the buzzer sounded, is now impossible. Indeed, insofar as I am able to refrain from pressing the button, this is only to the extent that the buzzer's having sounded does not depend on my pressing the button. If I had backed away from the door, deciding at the last moment not to attend the dinner, the buzzer that sounded five minutes earlier would have to be explained in some other way (e.g., the mechanism malfunctioned, there was a disturbance in the "tachyonic field," another dinner guest arrives and rings the bell instead). But whatever its cause, the sounding of the buzzer is a hard fact that I cannot now revoke.
The same is true for God's belief five minutes ago that I will ring my friend's doorbell. Does it make any difference that God's belief depends on what I do and that it would have been different if I were to act differently? No, for he has already held this belief, and my action must be consistent with this fact about the past. But whereas the sounding of the buzzer leaves me with alternatives to pressing the button, since it might sound without my causing it to ring, God's believing that I will press the button does not leave me any alternatives to pressing the button, since he can't hold that belief and I fail to press the button. This defense of Ockhamism, like the first two, is a dead end.
Despite its popularity among critics of the argument, I cannot see how the Ockhamist solution is even remotely plausible. Fortunately there is one more solution to consider.
5. The Augustinian solution. If we are right to accept (1) and also right to think of the argument in which it figures as the theological equivalent of Zeno's paradoxes (so that we cannot look to premise 3 or 4 for a solution), then the problem must lie in (2). At the end of section two, I stated briefly a reason for accepting (2). Since we must now reassess this reason, it will be useful to formulate and name it. Let the "Principle of Alternate Possibilities" (or PAP) be the name and this the formulation:
(PAP) A person X does an action A of his own free will only if X could have done otherwise.
If it is accidentally necessary before X is even born that X will do A, then X never has it in his power to do other than A; but then it follows from (PAP) that X does not do A of his own free will. So (2) must be true. The only way to deny this premise is to deny (PAP).
This is the route Augustine himself takes. While rejecting (3) because he endorses the atemporal version of (F) and qualifying (4) in the interests of divine grace, Augustine sees that denying either or both of these propositions does not get to the heart of the problem Evodius is posing for him. So rather than opting for an easy accommodation to the argument's conclusion, Augustine undertakes to refute the argument, and he does so by disputing (PAP). What he concludes, in brief, is that there is a sense in which (PAP) is true and a sense in which (PAP) is false, and the sense in which it is true does not support (2). Without that support, there is absolutely no reason to think that (2) is true; so Evodius's argument fails.
One way it might come about that an agent cannot do otherwise is if prior states of the universe (e.g., ones that obtained even before the agent was born) causally determine what the person does. Some philosophers who believe (or fear) that scientific naturalism commits them to regarding all of our actions as causally determined, but who are also unwilling to dismiss free will and moral responsibility as an illusion, have attempted to recast these notions in such a way that they are compatible with the absence of real alternatives. It is fair to say that this "compatibilist" understanding of free will is the predominant one among secular philosophers today.
It is important to understand that this is not the position from which Augustine challenges (PAP). Among the ideas discussed by Augustine and Evodius in On Free Choice of the Will is the possibility that human actions are the result of "natural necessity," where a paradigm of natural necessity is the falling of a stone.31 For a scientific naturalist, human beings are just as much a part of nature as stones, and our behavior is no less a product of "natural necessity," only much more complex; so this is just the kind of necessity that must be shown compatible with free will if the scientific naturalist is to preserve our ordinary concept of ourselves as free and responsible agents. But Augustine emphatically rejects such "compatibilism": it is because the stone falls by "natural necessity" that it does not fall by free choice and is not blameworthy for falling. If Adam's fall, like the stone's fall, were determined by "natural necessity," Adam too would be neither free nor blameworthy for sinning. So at least in the case where natural necessity or causal determinism is responsible for the fact that an agent cannot do otherwise, Augustine would not dispute (PAP).
The argument based on divine foreknowledge raises a completely different set of considerations. Augustine points out that the necessity that attaches to our actions in virtue of God's prior knowledge and infallibility is not at all like the necessity that attaches to the stone's fall in virtue of its prior suspension in midair and its subjection to the relevant natural laws.
It seems to me that a number of things recommend Augustine's position, of which I will mention two. First, Augustine's solution speaks directly to our suspicions about argument (1)-(2). Divine foreknowledge deprives Adam of alternatives, but we just can't believe that it deprives him of free will. Augustine's advice, in effect, is to trust our intuitions about this case. We think there is a problem here because the normal case in which an agent lacks alternatives (that he would otherwise possess) is one in which he is the victim of outside interference; the absence of alternatives is therefore a generally reliable indicator that an agent's free will has been abrogated. But foreknowledge cases show us that this is not universally true: because divine foreknowledge does not cause or compel what it knows, it is an exception to the general rule. (Divine power might cause or compel the future, but that's another story!)34
Second, Augustine's solution is consistent with recent critiques of (PAP) in the philosophical literature. This is important because it shows that the Problem of Human Freedom is not uniquely theological and that Augustine's solution is of general application-it's not a piece of special pleading whose only virtue is that it gets the traditional theist off the hook. Here's a nontheological story, first told by the philosopher Harry Frankfurt, in which (PAP) also yields the wrong result: Black wishes to see Smith dead, he knows that Jones shares this wish, and he thinks it likely that Jones will act on this wish. Unwilling to take any chances, Black deploys a science-fiction device capable of monitoring or controlling a person's thoughts and programs it to monitor Jones's intentions with respect to murdering Smith. If the device indicates that Jones is not going to murder Smith on his own, the device then interferes with Jones's mental processes to ensure that he murders Smith after all; but if the device indicates that Jones is going to murder Smith on his own, it simply monitors the course of events without interference. As it turns out, Jones goes ahead and murders Smith on his own, and nothing happens to trigger the device's intervention.
What are we to make of this case? It sure looks like Jones cannot do otherwise than kill Smith-either he will do it on his own or he will do it because the device forces him. But we are also strongly inclined to hold that Jones kills Smith of his own free will. Frankfurt justifies this judgment on the following grounds: Black's device "played no role at all in leading (Jones] to act as he did"; indeed, "everything happened just as it would have happened without Black's presence in the situation and without his readiness to intrude into it"; so the device is "irrelevant to the problem of accounting for a person's action" and "does not help in any way to understand either what made [Jones] act as he did or what, in other circumstances, he might have done."35 But what is here true of Black's device is equally true of God's foreknowledge that Jones will kill Smith. (As Augustine puts it in The City of God, "a man does not therefore sin because God foreknew that he would sin. ,) 36 Though both Black's device and God's foreknowledge eliminate all alternatives to Jones's action, neither does so in such a way as to jeopardize Jones's free agency.37
Ockham, as it turns out, was right about the relevance to the Problem of Human Freedom of the fact that God's believing that Adam will sin depends on Adam's future sinning and not the other way around. But he was wrong to think that the dependence of God's past belief on Adam's future action contributes to a solution to this problem by showing how Adam might possibly do otherwise despite God's prior belief that he would sin. Instead it shows how one form of accidental necessity-the one generated by divine foreknowledge-has no adverse consequences for free will, despite its elimination of alternative possibilities.
6. Conclusion. So (F) survives this first attack. I think we can be pretty confident of this even before we identify an actual flaw in the argument, just as we can be confident that the reality of motion survives Zeno's "paradox" without resting our confidence on a sophisticated analysis of where Zeno goes wrong. It's just implausible on its face that mere advance knowledge, even of the infallible and divine variety, should be able to transform an action that would in other circumstances be free into one that is not free. But the argument poses an interesting philosophical problem that is well worth grappling with. The result of my own grappling with the argument is the rejection of (2), and I recommend this result to the reader (for the reasons given). But it does seem to me that someone might be well within their rights to reject the argument without knowing quite where the fallacy lies. In any case, the Problem of Human Freedom provides no grounds for abandoning commitment to simple foreknowledge.
The Problem of Divine Agency
This brings us to the second argument against (F). This argument, as well as the third one we will look at below, raises problems that are at least implicit in much traditional discussion of God's providential governance of the world; but I don't know of any classic theological text where it is given so pithy a formulation as the one Augustine puts in the mouth of Evodius. I will instead introduce the Problem of Divine Agency by way of a story.
Suppose that your quirky uncle Clarence-the one who invented the time machine when everyone said it was impossible-shows up at your seventeenth-birthday party with an unusual gift. "I know you've been anxious about some upcoming decisions," he tells you. "So on my last trip into the future, I took along a video camera and recorded some of the key moments in the rest of your life." He hands you a brightly wrapped box. "This will tell you, for example, whether you get into Harvard and whether you accept. Happy viewing!" You thank him, lock the present in a drawer and return to the party.
Later that night you unwrap the videocassette and put it in the VCR. It begins with a trivial scene Uncle Clarence apparently included just to prove the tape's authenticity. There you are at the refreshment table, an hour or so after he gave you the cassette, about to upset the punchbowl . . . and there it goes-all over the floor, your pants and Mrs. Flint's dress. Just before the scene ends, you notice Uncle Clarence's reflection caught in a mirror at the end of the room, camera in hand, recording the incident. You decide to erase that part of the tape. But what about Harvard? That question is soon answered. The next scene shows you opening the letter from the admissions committee. You've been accepted! What a relief. But wait, you are now opening another letter, this one from Cornell. You've been accepted there as well, and Cornell is including a full scholarship. You had always dreamed of attending Harvard, but Harvard offered no money. What will you decide to do? The question is answered as the camera shows you towing a U-Haul trailer into Ithaca, New York, with one eye on the road and one eye on a campus map showing the location of the freshman residence hall to which you have been assigned at Cornell. And that's just the beginning of the tape. The next scene shows a wedding .. .
Let's interrupt the story at this point, because we've followed it far enough to introduce the problem on which we want to focus. You must now live the scenes that you have watched on tape. What will that be like? In ordinary life you deliberate among alternative courses of action, reach decisions that resolve your uncertainty and then act on your newly acquired intentions. That's what it is to be an active participant in your life rather than a passive spectator or an actor reading a script. But it looks like your viewing of the tape has condemned you to the latter category, at least with respect to the episodes filmed by Uncle Clarence. How can you engage in deliberation when you already know how it will turn out? How can you resolve uncertainty with a decision when there is no uncertainty to resolve? How can you acquire an intention to act in a particular way if your prior knowledge makes intention-acquisition pointless? By showing you what the future holds, the video short-circuits the route by which you usually approach your own actions, namely, through deliberation, decision-making and intentional agency. This leaves deeply problematic how you can even be an agent with respect to those future actions.
What should we think of this problem? It's certainly hard to escape the feeling that I would approach my future actions rather differently if I already knew what I am going to do; what is less clear is whether I should approach them differently and whether what may be true of me can be extended to God. I believe that the answer to both questions is no, making the Problem of Divine Agency doubly defective. But to see this, we need to put the worries underlying this problem in argument form.
Let's assume for present purposes that (F) is true. Then since God foreknows everything, his own future actions are among the things he foreknows. So
(1) If God is an intentional agent, then it is possible for him to acquire the intention to perform an action while already believing that he will perform that action.
(2) It's impossible for anyone to acquire the intention to perform an action while already believing that he will perform that action.
God is not an intentional agent.
If this is an unacceptable conclusion, we must reject the assumption that God foreknows his own exercises of intentional agency, and this means in turn rejecting (F). Right?
But if God does not acquire intentions in the way that we do, doesn't that at least make it difficult to understand how he can be an intentional agent in anything like the way that we understand ourselves to be? Not at all. In normal cases in which a human agent X performs an intentional action A, the following conditions are satisfied:
(a) X holds a belief B that it's up to her whether or not she does A.
(b) X endorses a judgment J about the advisability of doing A instead of something else.
(c) X does A with the intention Ito do A.
(d) Belief B and judgment J help explain why X has intention I.
(e) X's having intention I is preceded by a time at which she does not have intention I.
Now it seems to me that if God satisfies all these conditions except (e), it is perverse to suggest that he fails to qualify as an intentional agent or that the sense in which he is an intentional agent is nothing like the sense in which we are intentional agents. I would add that even in the human case, where (a)-(e) are all true, it is only (a)-(d) that make an action intentional; (e) is simply a byproduct of the fact that human beings are temporally limited and do not start out life knowing everything, so that they come to acquire intentions to act in certain ways only as the relevant information becomes available to them in time. That's an accident of the human situation, not an essential constituent of intentional agency per se.
In case this blow to (1) isn't enough to put the argument down for the count, let's make it a one-two punch. Premise (2) is also objectionable-so even if God does acquire (any of) his intentions, this is still compatible with his already having known what he will do.
To see what is wrong with this assessment, let us return to Uncle Clarence's videotape. After the scene in which you are granted tenure at Purdue, there is just one more episode on the tape, this one apparently coming some years later. You are living in what looks like a Stone Age village somewhere in the jungle. The naked tribespeople seem accustomed to your presence; you are astonished to hear yourself communicating with them in what strikes your younger self-the one viewing the videotape-as an unintelligible system of clicks and grunts. It isn't until you observe yourself returning to your hut, opening a Bible and beginning to write laboriously in an unknown language that you realize what has happened: you have joined Wycliffe as a Bible translator! Your shock and consternation know no bounds. What ever could have possessed you to give up the comfortable academic position you had striven so long to attain? That must have been one doozy of a midlife crisis! But this is evidently what the future holds in store for you.
What's your situation now, having viewed the tape to the end? You know that sometime around age forty-five you are going to face a defining moment in your life as two very different visions of the good life contend for your allegiance. You also know the ultimate outcome of that struggle. There is a sense, then, in which the question whether the mission field will prevail over the ivied walls of academe is already settled for you. It is settled in the sense that you are now prepared to assent to the truth of a certain proposition: I will choose Wycliffe over Purdue. But you are most emphatically not prepared to assent to the choice itself. Your seventeen-year-old self is absolutely horrified at how your forty-five-year-old self is going to mess things up. A missionary is the very last thing you want to be! It is clear that your will is far from settled, even if your belief is settled by the irrefutable evidence of Uncle Clarence's videotape.
We can now sum up the problem with (2). While foreknowledge and intention-acquisition both resolve unsettled beliefs, what one comes to believe as a result of foreknowledge is a propositional belief about what will happen, whereas what one comes to believe as a result of intention-acquisition is a practical belief about what to do. The former does not entail the latter; so even if the propositional belief is acquired first, it may still be necessary to go through the actual process of intention-acquisition (including protracted deliberation, if that is what is required) in order to achieve the practical belief. It's clear, for example, that all the work of acquiring the practical belief to join Wycliffe lies ahead of your seventeen-year-old self and that there is considerable point to your doing so, even though you already possess the propositional belief. There is no need in the case of (2) to appeal to unique features of the divine situation, since this premise is false for human agents as well. What makes God or anyone an agent is the power to decide what to do. God's omni science does not deprive him of this power, since it determines only his propositional beliefs, leaving his practical beliefs-the ones relevant to agencywide open.42
So (2), like (1), must be rejected. I conclude that (F) easily survives the challenge posed by the Problem of Divine Agency.
The Problem of Divine Providence
We now come to the last of our three problems: the Problem of Divine Providence. By "providence" I mean activity God undertakes with a view toward achieving his intended ends, whether in individual lives or on a cosmic scale. And what exactly is the problem with providence? Well, suppose the Problem of Divine Agency has been neutralized, so that we are no longer troubled by the mistaken notion that (F) might interfere with God's intentional action. But it isn't enough that divine foreknowledge doesn't worsen God's agential position-shouldn't it improve his position? Shouldn't the vast amounts of additional information with which his foreknowledge provides him make some difference to what he can do, making him more competent in some way than he would be if he lacked knowledge of the future? One would certainly think so; yet some critics of (F) have argued that foreknowledge is utterly useless to God.
One such critic is William Hasker. "The doctrine of divine foreknowledge, in its most widely held form," Hasker argues, "is of no importance whatever for the religiously significant concerns about prayer, providence, and prophecy."43 And why is this? Hasker considers a case in which a young woman prays to God for guidance regarding two potential husbands. Can God's knowledge of the future contribute anything to the advice he gives her? Apparently not:
Hasker isn't denying here that God can bring about or prevent future events, nor is he denying that at least some of the things God knows might prove useful as he decides whether (and if so, how best) to influence events; what he is denying is that anything God knows about the future can contribute to God's decision. Could knowledge of what will happen help God act so as to prevent what he sees will happen? That doesn't seem right, since his intervention would then be undermining the very knowledge on which it is based.45 Or could knowledge of what will happen help God act so as to bring about what he sees will happen? Here God's intervention reinforces rather than undermines what he foreknows, but in doing so it raises a problem of circularity: his intervention is based on his knowledge, which is based on the future, which is based on his intervention. It is because the future is the result of whatever has or has not been done to influence it that knowledge of the future comes to God "too late" (in the logical rather than temporal order) for God to draw on it in his own efforts to influence the future.
The basic objection seems to be that (i) God can't put his foreknowledge to use without generating an explanatory circle and (ii) such circles are impossible. Neither part of the objection is immune to challenge. The philosopher David Lewis, in an important article on the similar loops spawned by timetravel stories, argues that while such loops are inexplicable they are nevertheless possible." These are deep waters, and I'm not at all confident that I can see to the bottom of them; so while I think it's far from obvious that Hasker is right about (ii), I propose to leave it alone. On the other hand, (i) is readily refutable. In fact, I'm inclined to think that God can get quite a lot of (noncircular) use out of his knowledge of future contingencies. But the defense of this position, which I have undertaken elsewhere, can get pretty complicated.47 So let me adopt here the much simpler and more modest project of developing a single case in which foreknowledge would be useful to God. Having shown this, I will leave it to the reader to judge whether reality might support many or few such cases.
Suppose that an eccentric billionaire named Bill receives a fundraising letter from Wycliffe and responds with the following proposal. He is prepared to transfer $1,000,000,000 to Wycliffe's account if a representative of Wycliffe can guess which number between 1 and 100 he is going to choose. The staff of Wycliffe's home office begin a marathon prayer session, in the course of which they become certain that Bill will choose the number 47. This certainty comes from God, who foresees that Bill will indeed choose the number 47. The next day Wycliffe's fundraising director, Charity, arrives at Bill's office, where a team of lawyers has drawn up a legally binding transfer of the promised sum contingent on her guessing the right number. She hands over a sealed envelope containing a slip of paper on which the number 47 has been written. Bill then thinks for a few seconds before writing a number on a legal pad. He holds it up so everyone can see: it is the number 47. Charity's envelope is then opened to reveal the match. The forces of Satan howl in dismay as Bill duly transfers the billion dollars to Wycliffe.
It is hard to see how Hasker's objections have any purchase on this case. Information that is available to God through simple foreknowledge but that is not available through diminished foreknowledge is being used to give God a "providential edge." God's use of this information does not guarantee that he will get what he wants-Wycliffe's staff might always resist God's leading-but if it were unavailable to him there would be no leading for them to resist. At the very least, he raises the odds (1/100 without intervention, somewhat better if he offers advice based on intimate knowledge of Bill's psychology) that he will get the outcome he desires. And he puts this knowledge to use without involving himself in any explanatory loops. Charity's submitting the number 47 to Bill's lawyers presupposes God's guidance at the staff prayer meeting, which presupposes God's foreknowledge that Bill will choose the number 47, which presupposes Bill's choosing the number 47, which presupposes ... what? Who knows why Bill chose that particular number? Whatever the answer, it's not because Charity handed his lawyers an envelope with the number 47 written inside. There is no circularity here with which we need to be concerned.
Among the various objections that opponents might wish to raise about this example, I would guess that the following two would be the most common and important.
Objection 1. If God's foreknowledge informs a divine intervention that occurs after the foreknown event takes place, he could have secured the same result by waiting for the event to occur and then cognizing it with his present knowledge; there is nothing foreknowledge helps him do that he couldn't have done without it. But if his foreknowledge informs an intervention that occurs before the foreknown event takes place, as it does in our example, the intervention might have causal effects on the foreknown event or in some other way close the explanatory chain so that we are facing an invidious loop after all. Suppose, for example, that Bill has a spy in the Wycliffe organization who jimmies open the locked drawer where the sealed envelope was placed overnight for safekeeping and then steams it open. Now consider various ways things might go at this point:
(A) Learning from the spy that Wycliffe has selected the number 47, and wishing all along to donate the money to Wycliffe, Bill therefore chooses the number 47. Here's the explanatory circle we wanted to avoid.
(B) Wishing to keep his money, Bill chooses a different number. But of course he can't, since God's prescience would then be mistaken. This scenario is incoherent.
(C) The spy-a Frenchman accustomed to European script-misreads the 47 as 41. The next day Bill is about to choose the number 41 when he receives an urgent message from the spy: Wycliffe has chosen 41! Bill quickly casts about for a different number, and 47 is the first one that pops into his head. This restores coherence, but only by reinstating the explanatory circle.
Reply to objection 1. My task was to describe an unproblematic scenario for God's providential use of foreknowledge; the fact that there might be other scenarios that are problematic is irrelevant to the question of whether my original scenario (sans spy) is acceptable. But even if Bill has a spy at Wycliffe, God's foreknowledge would still give him the same providential edge. If (A)(C) are metaphysically impossible, we can be sure that reality won't include such situations. The spy's mission will therefore fail. (Perhaps it even fails because God intervenes to that end. If you wanted to help Wycliffe get the $1,000,000,000 and you knew about the spy, wouldn't you try to thwart him if you could?) Because it is metaphysically impossible for the spy to succeed, God's foreknowledge will have the same providential benefits it had in the original case.
Objection 2. If (F) is true, God foreknows not only Bill's choice of number but also the Wycliffe staffs responsiveness to his guidance, Charity's delivering the number to Bill's office, Bill's decision not to renege on his agreement and so on. So even if Bill's choosing the number 47 does not depend on God's intervention, God knows other things that depend on his intervention. We overlook this problem, Hasker explains, because we
think, albeit unconsciously, of God's foreknowledge along the lines of the limited foreknowledge we ourselves sometimes have, when we see certain events com ing that are not contingent upon anything we may choose to do or to refrain from doing. In such cases the actions that we take in view of our foresight lead to no paradox. But if we could foresee everything, then for us, as for God, it would be too late to do anything about it.48
In the divine case, it would appear, the circle is ineliminable.
Reply to objection 2. The situation as Hasker describes it is prima facie odd. He allows that God might gain some providential advantage if his knowledge of the past and present were augmented with limited foreknowledge-say, if Bill's choosing the number 47 were the only future contingency in his epistemic repertoire; but he insists that God would lose this advantage if he had unlimited foreknowledge, as (F) claims.
But why should the acquisition of even more knowledge make God's providential position worse? Mere possession of information about the future is harmless (as was shown in our discussion of the Problem of Divine Agency); only its use can generate explanatory loops. But why suppose that God must use this problematic extra knowledge? David Basinger has an answer: 11f] God purposely avoids the consideration of relevant data to which he has access, it is ... possible that he will make decisions that fail to actualize his desired ends to the extent they could have been actualized if he had considered all of the relevant data available." Among such data is "foreknowledge of what he will do in fact and what will follow in fact. ,49 Since a God with unlimited foreknowledge knows not only that Bill will choose the number 47 but also that Wycliffe will get the billion dollars in the end and since the latter is relevant information that God would use and not merely possess, we've got an invidious loop on our hands.
I'm just not convinced that this is "relevant data," however. The fact that you want to get into Harvard and that submitting an application is a necessary means to this end is a reason for submitting an application; the fact that you will get into Harvard (because you learn this from Uncle Clarence's video) is not a reason for applying. (Nor is this fact a reason not to apply-"I don't have to; I'm going to get in anyway!") In the Wycliffe example, either God's basing his intervention at the staff meeting on knowledge of what will result from his intervention does generate an invidious explanatory circle or it doesn't. If it's the latter, then Basinger hasn't raised any problems for (F); but if it's the former, this is itself good reason to think that God would (could) not use this item of foreknowledge in the way Basinger claims. By providing God with complete information about the future, unlimited foreknowledge gives God the flexibility to use what is useful and ignore the rest. This is an advantage not enjoyed by a God with knowledge of the past and present alone .50
This problem is not quite as consequential as the first two: God's inability to employ his foreknowledge providentially would not show that (F) is false, it would only show that this doctrine's importance lies elsewhere (e.g., in its contribution to God's metaphysical excellence). But there are nevertheless a couple of reasons for taking the problem seriously. First, it is much tougher than the other two. While I'm pretty confident in what I have said about the Problem of Human Freedom and the Problem of Divine Agency, the Problem of Divine Providence presents a scenario that is every bit as tangled as the most complex time-travel story, and I'm not at all sure that I have it right. At the very least there is much more that needs to be said about this one.
Second, it is worth defending simple foreknowledge on this score because its alleged failure to provide God with any providential advantage has led critics to squeeze it from both sides. Want more providential control than is afforded by diminished foreknowledge? Then you'll need to augment foreknowledge with "middle knowledge." Think that middle knowledge is incoherent? Then there's nothing to be gained by insisting that God has more than diminished foreknowledge. It is important to see that this is a false dilemma. While simple foreknowledge clearly offers less providential control than middle knowledge, it does offer more than past-and-present knowledge. This means that there is a middle ground to stand on should the problems with middle knowledge prove insuperable.
Conclusion
In defending free agency (both human and divine) from various threats posed by (F), I have said very little about what it is that I am defending. This was intentional. I have tried to get by with an unelaborated, pretheoretical understanding of free agency-simple free agency, one might say-in much the way I tried to operate with a similar understanding of (F) as simple foreknowledge. One virtue of this approach is that I don't need to pretend more knowledge than I actually possess. Perhaps some of my coauthors know what free agency is and how God knows the future-I'm quite sure that I don't! But I also hope to gain some insight into these questions by wrestling with the three problems raised against M. This brings me to a second virtue of this minimalist approach, which is that I am relatively free to be instructed by these three problems. How must I think about divine foreknowledge and free agency if these are to be compatible with each other? Rather than starting off with a rich account of foreknowledge and agency and applying these to the defense of (F), I see my Christian commitment to (F) as providing me with an opportunity to develop a (richer) account of foreknowledge and agency. The best way to do this is to presuppose as little as possible and see where the argument leads.
The most general conclusion to which the argument has led, in my estimation, is that it is the explanatory rather than temporal order that is important to agency. We overlook this because the two orders normally coincide: when A explains or causes B, A is typically earlier than B. Commitment to (F) forces us to consider what happens when these two orders come apart. We have seen that God's beliefs about the future may settle the question of what will happen, thereby eliminating alternatives to what he believes; but we have also seen that this need not disrupt the actual exercise of agency (whether human or divine). Thus Clark Pinnock is partly right and partly wrong when he writes, "Total knowledge of the future would imply a fixity of events. Nothing in the future would need to be decided. It also would imply that human freedom is an illusion, that we make no difference and are not responsible. ,5' He is right in the first sentence but wrong (I believe) in the implications he derives from it in the second and third sentences.
The conclusion that free agency is compatible with a "closed" future is not open-ended; it extends only as far as the cause or explanation of the future is to be found in the agent (and not in something prior to the agent). Thus our examination of these three problems has given us no grounds for holding that free agency is compatible with causal determinism (which posits sufficient prior conditions for all an agent's actions). Nor does it support compatibilism with respect to divine determinism. A book on divine sovereignty and human freedom might force us to entertain that prospect! But because divine foreknowledge neither causes nor explains what it knows, it does not require us to adopt an attenuated understanding of free agency under which an agent may be "free" even when his actions are causally determined by a force outside himself, whether God or the universe.
I began this chapter with the ordinary believer's commitment to M. The bottom line is that there is no good reason, at least among the three problems we examined, for the believer to jettison this commitment. That's a happy result from a theological standpoint. The fact that this result has, as I believe, interesting implications for free agency provides a nice philosophical bonus.
AN OPEN-THEISM RESPONSE
Gregory A. Boyd
here is much in David Hunt's carefully argued essay with which I agree. There are a few crucial areas where I find myself in disagreement, however. I shall first review two areas of agreement and then discuss three areas where I disagree.
The Bible and God's Foreknowledge
I agree with Hunt (as well as with William Lane Craig and Paul Helm) that the Bible contains examples of God's foreknowing what shall come pass. Hunt "infers" from these examples that God possesses exhaustively definite foreknowledge about all that shall come to pass.' Yet he candidly acknowledges that "the Bible nowhere makes this claim" and therefore concedes that the claim is "at least potentially vulnerable to the fallacy of `hasty generalization' "
(p. 70). I appreciate Hunt's candor.
I myself believe that this is precisely the fallacy committed by those who hold to the classical view of foreknowledge. The teaching and examples in Scripture certainly warrant the conclusion that some of the future is settled in God's mind. Indeed, I believe the Bible warrants the conclusion that God can settle as much of the future ahead of time as he sovereignly sees fit. But, as I attempted to show in my original essay, the Bible with just as much force also warrants the conclusion that God has sovereignly decided to not settle all of the future. Some of the future is open, as is evidenced by the fact that God speaks and thinks about it in subjunctive terms; that he changes his intentions in response to changing circumstances; that he expresses regret, disappointment, surprise and frustration regarding what comes to pass; and so on.
Invalidity of Some Standard Arguments
With noteworthy patience and philosophical care, Hunt refuted a number of arguments that are sometimes used in an attempt to demonstrate the compatibility of exhaustively definite foreknowledge and human freedom. I found myself in agreement with most of Hunt's refutations. In particular, his refutation of the argument based on the distinction between "hard" and "soft" facts was insightful. (Helm effectively dealt with this as well.) It could be argued that God's knowledge that, for instance, Adam would sin constitutes a "soft" fact. But it cannot be cogently argued that God's belief that Adam would sin constitutes a "soft" fact. This is as "hard" a fact about the past (viz., a past mental state of God) as anything could be. And since the beliefs of an omniscient being are by definition infallibly correct, we have to conclude that by "accidental necessity," Adam's future sinning was "locked in" before he was created, in which case he could not be free.
I turn now to three areas of disagreement.
A "Diminished" View of Foreknowledge?
Hunt labels the open view's concept of foreknowledge "diminished foreknowledge" (p. 66). He further suggests that there is "some dissonance (if not outright contradiction) in our thinking about God if we suppose him to be unsurpassably great while also attributing to him ignorance of some future event" (p. 69).
Clearly, if God were ignorant of anything, we could not call him "unsurpassably great." We would have to conclude that his knowledge was indeed "diminished," and hence we would have to deny that he is omniscient. But the open view of the future vehemently denies that God is ignorant of anything. If a future event was in fact settled and God was ignorant of this, that would be "diminished foreknowledge." But if a future event is possible, not yet settled, we hardly attribute ignorance to God by saying he knows it as possible. For we are simply saying, with all Christians, that God knows all of reality exactly like it is.
The mistake Hunt makes-and the one repeated by Craig, Helm and many other critics of the open view-is confusing a difference of opinion about the content of reality with a difference of opinion about the scope of God's knowledge. On this mistaken logic, open theists could just as easily argue that classical theists "diminish" the scope of God's knowledge because in their view God is "ignorant" of the reality of genuine possibilities!
Foreknowledge and Providential Control
This first objection is closely related to a second objection. Hunt believes that although the simple-foreknowledge view gives God less control than either the middle-knowledge perspective or the Reformed perspective, it gives God more control than the "diminished foreknowledge" (viz., the open-theists') perspective. I respectfully disagree. Two points may be made.
First, open theists agree with the historic orthodox understanding that God possesses unlimited or "infinite" intelligence. If thought through consistently, this entails that God's intelligence doesn't get "thinned out," as it were, by having to attend to any number of possibilities. He may attend to and anticipate a trillion possible story lines as though each was the only possible story line he had to attend to. This means that God loses no providential advantage by virtue of the fact that he faces a future partly composed (by his own design!) of possibilities. Whatever story line comes to pass, God perfectly anticipated itas though this was the only possible story line that could be actualized. When the story line of history is completed, it will look to us finite beings as though God eternally foreknew just this one story line.
An open theist can thus celebrate, as much as can any classical theist, the wonder of God's incomprehensible foreknowledge. We can celebrate the unsurpassable sovereign advantage this foreknowledge affords God. It's just that we also celebrate that God is so intelligent, he didn't need to have only one story line to pay attention to in order to perfectly anticipate it. In the open view, God doesn't under-know the future, he over-knows it. The criticism that God loses providential control by having more than one possible story line to attend to is really based on an anthropomorphic presupposition that God's intelligence is not infinite. Humans lose control to the extent that we must anticipate possibilities rather than simply know certainties, for we have only so much intelligence to go around. But for a God of infinite intelligence, there is virtually no advantage in anticipating settled story lines versus possible story lines.
Second, precisely because some of the future is open in our view, God has more, not less, providential control than in the simple-foreknowledge position. In the simple-foreknowledge view, God cannot alter what is going to come to pass, for it is eternally settled. In the open view, God can "alter the future," for "the future" is not exhaustively settled. Because the future is partly composed of possibilities, God can warn us (as in biblical prophecy) about future circumstances and can lead us to alter what otherwise would come to pass. We "trust that God holds the future" not because he is certain of what will transpire. What confidence does that bring? Perhaps the future of which God is certain has one of our children being kidnapped! We rather "trust that God holds the future" because he perfectly knows and anticipates all that may transpire. The "may" gives space for God to intervene and lead us in a different direction.
The Frankfurt Analogy
My final objection to Hunt's essay concerns his endorsement of the Frankfurt analogy. As said above, Hunt effectively refutes a number of ways people have attempted to render exhaustively definite foreknowledge with free agency. Yet he believes that exhaustively definite foreknowledge and free agency are compatible, and he pins his hope for a plausible philosophical defense of this compatibility on one argument. Those who defend libertarian freedom often assume that "A person X does an action A of his own free will only if X could have done otherwise." Hunt agrees that if this principle is valid, then God's foreknowledge that X will do A means that X did not do A freely. For God's foreknowledge "locks in" the fact that X will do A, thus ruling out the possibility of X's doing -A. But Hunt rejects this principle. Person X may do A freely, even though it is not possible for X do to A. Hunt's case for the compatibility of exhaustively definite foreknowledge and free agency rests on the cogency of his case against this principle.
The only support Hunt offers in his essay for this rejection is an analogy, first submitted by Harry Frankfurt. Suppose one person (Black) wants another person (Smith) dead. He knows that a third person (Jones) wants to kill Smith. Black therefore deploys a science-fiction device that is capable of monitoring and, if need be, controlling Jones's mental processes. Hence, even if Jones had second thoughts about killing Smith, the device would ensure that he would do so. As it turns out, Jones freely follows through on his intention to kill Smith. Hunt argues that this analogy demonstrates that it is possible to do something freely, even though one couldn't have done otherwise, for Jones would have killed Smith in any case. As Hunt puts it, "Though both Black's device and God's foreknowledge eliminate all alternatives to Jones's action, neither does so in such a way as to jeopardize Jones's agency" (p. 90).
In my estimation, Frankfurt's analogy is fatally flawed. The central problem is that it defines a free decision in terms of behavior rather than in terms of mental processes. It is true that in terms of external behavior, Jones had no alternative to murdering Smith. But it is not true that Jones had no alternative in terms of his internal mental (and spiritual) processes. Though he didn't know it, Jones's choice was either to murder Smith of his own volition or to not. Had he listened to his conscience and decided not to murder Smith, the science-fiction device would have made him kill Smith anyway, but in this case it wouldn't be a murder Jones freely committed. A perfectly fair and informed jury would never hold Jones morally responsible for doing what the device made him do.
I therefore conclude that Hunt has failed in his attempt to demonstrate the compatibility of exhaustively definite foreknowledge and free agency. And since I largely agree with Hunt's own refutation of other possible defenses, I remain under the conviction that exhaustively definite foreknowledge is not logically compatible with free agency. Person X is free (and thus morally responsible) regarding an action A if and only if it is genuinely possible for him to do A. Hence, in perfectly knowing X, God would know that both A and -A are possible. The future is open to X's doing A or A.
A MIDDLE-KNOWLEDGE
RESPONSE
William Lane Craig
avid Hunt's provocative position is that fatalism is true-but not to worry, for fatalism is compatible with libertarian freedom!
In examining Hunt's position, the first question we need to ask is, why should we think that fatalism is true? Has Hunt explicated a sense of temporal necessity that does not reduce to either the unalterability or the causal closedness of the past?
He has not. Hunt merely appeals to examples to evoke in us the feeling of the past's necessity. When Hunt concludes, "In general, everything that is past is (now) necessary in this sense" (p. 75), we are left wondering, "In what sense is that?" Let ACT stand for "act in a certain way and, if one were to act in that way, some past fact would not have been a fact." If Hunt means that it is not possible for me to ACT, then Hunt's generalization from a single example is hasty.
We can envision instances in which I could ACT. If, for example, I have a time machine that enables me to travel back in time, then I can ACT. Now in the absence of a time machine, it is doubtless not within my power to ACT with respect to a broad range of past facts. But that does not imply that there are no past facts relative to which I can ACT. Unless we are logical fatalists, we must agree that at least the past fact of a future contingent proposition's being true is something now past that is still such that I can ACT with respect to it. But then I can also ACT with respect to God's past beliefs. For, necessarily, if I were to refrain from some action A, then a true future contingent proposition p stating that I shall do A would have been false. And, necessarily, if p had been false, then God would not have believed p. It therefore follows that were I to refrain from A, God would not have believed p.
Now Hunt is quite willing to admit the truth of counterfactuals of the form
1. If I were to refrain from A, then God would not have believed p.
and presumably of the form
2. If I were to refrain from A, then p would not have possessed the truth value True.
But he insists that I do not, in fact, have the power to refrain from A. For both God's belief that p and p's possession of the truth value True are "hard" facts about the past, and it is not within my power to ACT with respect to hard facts about the past.
What is a "hard fact"? Hunt says that it is "a fact that is entirely about the past" (p. 76). What does that mean? Hunt does not explain. He seems to be striving to articulate what Alfred Freddoso has called temporally indifferent propositions. In Freddoso's analysis, a temporally indifferent propositions is one that could be true at a first, an intermediate, or a last moment of time.' Hence, a temporally indifferent proposition that is true at a time t will be entirely about t.
If, with Hunt, we identify hard facts with the facts described by temporally indifferent propositions true at a past moment of time, then p's possessing the value True is not a hard fact. Accordingly, it may be the case that it is within my power to ACT with respect to that fact. But then I also have the power to ACT with respect to God's belief that p. For the statement "God believes p" is not temporally indifferent either, since it could not be true at a last moment of; time. Therefore, "God believed p" is not a hard fact about the past.
Furthermore, why we should accept the equation between a fact's being "entirely about the past" and its being temporally necessary? If time travel is possible, then facts that are entirely about the past are not beyond the power of the time traveler to affect. Analogously, temporally indifferent propositions based on divine foreknowledge like Isaiah prophesies the destruction ofJerusa- lem may be such that it is within the power of later agents to ACT with respect to them.2 Accordingly, this fact is not temporally necessary before the relevant agents freely destroy Jerusalem. This is admittedly weird; but then time travel and divine foreknowledge are exotic notions, and we should not be surprised if they lead to startling implications.
The crucial difference between these cases and Hunt's alleged parallel, If I had saved the child, then I would not have been knocked unconscious, is that his example gratuitously deviates from the standard counterfactual semantics. The type of counterfactual at issue here is called a backtracking counterfactual because the truth of the antecedent requires that the past be different than it was. Now such backtracking counterfactuals are deviant from the standard semantics for counterfactuals. On the standard semantics, the past is held constant, and one asks what would then happen in the possible worlds that are most similar to ours in which the counterfactual's antecedent is true. This is called "the standard resolution of vagueness" among possible worlds in which the antecedent is true. On the standard semantics, true counterfactuals having the antecedent Hunt imagines have consequents like "then he would have survived the fire." In order to claim that a backtracking counterfactual is true, one needs some justification for adopting a special resolution of vagueness. Otherwise we get counterintuitive results from our semantics.3 It is a matter of considerable controversy when a special resolution of vagueness is justified. I have elsewhere argued that it is warranted only in cases in which the event described in the antecedent is a condition of which the event in the consequent is the consequence. In Hunt's example, my not being knocked unconscious is not a consequence of my saving the child; indeed quite the reverse is the case.4 In the absence of any justification of a special resolution of vagueness, we have no reason to think that Hunt's counterfactual is true. By contrast, in the case of divine foreknowledge, a special resolution must be used, for given God's essential omniscience, there is no possible world in which the antecedent is true and he does not believe what the antecedent states.
As for Hunt's second example of the tachyonic doorbell, he just assumes that no special resolution of vagueness is in order. But in the case of time-traveling signals, backtracking counterfactuals can be justified.5 As for God's belief that I shall press the button, what is impossible is the conjunction of God's belief and my failing to press the button. But my failing to press the button is not impossible. If I were to have refrained, God would have believed differently.
Hunt also takes for granted that temporal necessity is closed under entailment, that is to say, that the necessity of the premises is passed on to the conclusion of his argument. But we have the very best of reasons for thinking that temporal necessity as Hunt imagines it is not closed under entailmentnamely, fatalism posits a constraint on our actions that is completely unintelligible. It is curious that Hunt actually cites me approvingly on this point but then turns around and says that "divine foreknowledge deprives Adam of alternatives" (p. 88), though not of free will. If divine foreknowledge leaves Adam causally free, what mysterious force deprives him of alternatives? In fact there can be no such mysterious force, for that would be to posit some causal constraint on Adam. But if there is no such mysterious force, what deprives Adam of the power to act other than as he will? It seems, then, that even if Hunt could enunciate some sense in which God's believing p is temporally necessary, temporal necessity must not be closed under entailment, so that the events described by p do not occur necessarily.
So much for fatalism. What about Hunt's solution to it? That solution is predicated on a Frankfurtian conception of libertarian freedom that rejects the Principle of Alternative Possibilities. Since I am not causally constrained to do some action A, I do A freely, even if it is not within my power to freely refrain from doing A. Thus, my doing A necessarily is not incompatible with my doing A freely.
The problem with Hunt's solution is that it appears to be inconsistent with Frankfurtian freedom itself. For as Thomas Flint observes, Harry Frankfurt's examples leave intact the libertarian intuition that if I do A freely, then I could have failed to do A freely. 6 But Hunt's fatalism does not allow this. For given God's foreknowledge, I cannot fail to do A freely. I must do A, and I must do A freely-that is, without causal constraint-since all the historical antecedents of A are also foreknown by God and, hence, are necessary. Thus, Hunt's solution is inconsistent with the very concept of freedom it presupposes. Frankfurt's examples involve my either doing A freely or doing A under causal constraint. But these options are not open under Hunt's fatalism. All he can affirm is that every free action of mine is done without causal determination. But the fatalist has already conceded that; that point is not in dispute. The issue is, what sense of freedom remains if I cannot freely refrain from A? Frankfurt's answer to that question is one that is not open to the theological fatalist.
AN AUGUSTINIAN-CALVINIST
RESPONSE
Paul Helm
avid Hunt initially concentrates on defending the consistency of the following propositions:
(3) God foreknows that Adam will sin.
(4) Adam will sin of his own free will.
But he adopts an interesting dialectical attitude toward them. He says, "Sometimes we have good reason to believe that there must be something wrong with an argument before we are in a position to see what is wrong with it" (p. 80), and he then offers Zeno's paradox of Achilles and the tortoise as a case in point. Even though we may not be able to see where the fallacy lies, we are convinced that there must be a fallacy since we are convinced that Achilles is faster than the tortoise.
Hunt claims that we face exactly the same situation when it comes to the argument for the incompatibility of divine foreknowledge and human freedom. He says that "there is something fishy about the idea that Adam's action, while in every other respect satisfying the most exacting requirements for free will, might nevertheless count as unfree simply because God foreknew what he would do" (p. 81). His argument is outlined here:
(i) God foreknows that Adam will sin (assumption).
(ii) The supposition that God's knowing some act in advance adversely affects the (incompatibilistic) freedom of that act is fishy and unintelligible.
(iii) Therefore (i) is consistent with Adam's sinning of his own free will.
As far as I can see, Hunt offers no argument for (ii) beyond calling the claim that God's foreknowledge of Adam's sin renders it unfree (i.e., not an indeterministic act) "fishy."
This does not seem to be a very persuasive argument. In fact, it looks remarkably like a case of begging the question. The question is, Is divine foreknowledge incompatible with human freedom? To say that a positive answer to this question is fishy (or, following William Lane Craig, is "unintelligible") and then to dismiss it without argument is a plain begging of the question.
The parallel with Zeno's paradox of Achilles and the tortoise is not apt. There is nothing in the least paradoxical about the claim that divine foreknowledge is inconsistent with human indeterministic action, as is testified by the generations of discussion of the issue and by the diverse viewpoints represented in this book. It is a serious problem meriting serious discussion.
Further, there is an inherent weakness in the case of the appeal to the Achilles paradox. For once this particular ball starts to roll, it is difficult to stop it. Take any metaphysical question: Is time unreal? Is the human self a simple entity? Do material things have essences? Does God exist? Do I have counterparts in other possible worlds? All these questions can, it seems, be given the Achilles treatment. Thus it may be claimed that the idea that time is unreal is unintelligible, fishy. Therefore whatever the arguments for that view, there must be flaws in them even though we cannot at present find them. Therefore it is reasonable to believe that time is real. And so on.
If Hunt's appeal to the precedent of Zeno's paradox of Achilles and the tortoise works in the case of (3) and (4), should it not also do to rule (1) out of court, since what (1) says-that if God foreknows that Adam will sin, then Adam cannot sin of his own free will but instead his action of sinning is necessitated-looks equally fishy and unintelligible. How could God's foreknowl edge of an action necessitate that action? In fact Hunt comes close to adopting this attitude when he says, "Divine foreknowledge deprives Adam of alternatives, but we just can't believe that it deprives him of free will" (p. 88; see also the conclusion to his discussion of "The Problem of Human Freedom"). So it is a puzzle why we need to devote philosophical attention to (1) if it is sufficient to appeal to the Achilles paradox in the case of (3) and (4). Why doesn't Achilles also do the job on (1)?
Despite this, Hunt does give attention to (1), and I agree with him wholeheartedly in his able rejection of the Ockhamist strategy as an attempt to disable the consequences that the necessity of the past has for human freedom. Rejecting Ockhamism, Hunt by this stage accepts (1), (3) and (4). So the problem must lie with (2)-that if it's necessary that Adam will sin, then Adam will not sin of his own free will. At this point, in his attack on (2), Hunt, following Harry Frankfurt, rejects the "Principle of Alternate Possibilities" (PAP). Adam sins of necessity if God foreknows what he will do, and his sinning by necessity is inconsistent with (PAP). But (PAP) is not necessary for a satisfactory account of freedom. And so Adam may sin of necessity while retaining his freedom.
If (PAP) is rejected, then Hunt may justifiably claim that "divine foreknowledge deprives Adam of alternatives, but we just can't believe that it deprives him of free will" (p. 88). Let us suppose that Hunt is correct in this. It seems that one who upholds the Augustinian-Calvinist perspective could co-opt Hunt as an ally at this point, by arguing in the following way:
(1) Human freedom is compatible with the denial of (PAP). (Hunt's conclusion)
(2) The denial of (PAP) is consistent with causal determinism, since (PAP) is inconsistent with causal determinism.
(3) Therefore, freedom is consistent with causal determinism.'
Whether or not Hunt would welcome this argument is hard to say since throughout his (rather misleading) account of Augustine's attitude to compatibilism, he opposes compatibilism. What Hunt says about Augustine is misleading because throughout his account of what goes on in Augustine's early treatise On Free Choice of the Will, Hunt claims that Augustine rejects (PAP) but not on causally determinist grounds (p. 8). It is true that Augustine rejects the idea that human actions are the result of natural necessity, but of course natural necessity does not exhaust the varieties of determinism or of compatibilism, not even among modern secular naturalists-not, at least, if the paradigm of natural necessity is the falling of a stone. The view that human actions are the causal outcome of beliefs and desires, say, gives us a very unstone-like version of determinism and (for a compatibilist) of compatibilism. And certainly in his later writings Augustine denies (PAP) on the grounds that it is inconsistent both with that version of compatibilism he favors and with the sufficiency and efficacy of divine grace, as Hunt himself notes (p. 93 n. 40).
So if Hunt denies (PAP), he is committed to some version or other of determinism. His argument shows that divine foreknowledge is consistent both with determinism and with Frankfurtian freedom. So is Hunt among those who share the Augustinian-Calvinist perspective?
Fatalism Again
Finally, since the word fatalism occurs more than once in Hunt's characterization of Augustinianism, it is worth saying something about this in addition to the remarks I will make in response to Craig. It is obvious that not every doctrine that denies indeterministic free will is a case of fatalism, since determinism denies it and determinism is distinct from fatalism. Fatalism is the doctrine that whatever will be will be, holding (in one prominent version) that the denial of free will follows from the laws of logic alone. It holds that some future event X will occur no matter what. But determinism denies this. (As we have seen, the Frankfurtianism that Hunt favors also entails the denial of indeterministic free will. So if this denial is sufficient to make a doctrine fatalistic, then Hunt will find himself in the serried ranks of the theological fatalists.) This suggests that the phrase theological fatalism is not sufficiently discriminating. What Hunt and Craig call theological fatalism would he better characterized as one of at least three distinct views-theological determinism, theological Frankfurtianism or theological decretalism. Though these last two views entail determinism, they need not entail theological determinism.
Divine Providence
The other two problems for simple foreknowledge with which Hunt deals concern the problem of divine agency and the problem of divine providence. As I could not make much of the problem of divine agency, I shall round off these comments by a brief remark on divine providence.
Hunt attempts a refutation of William Hasker's claim that the doctrine of divine foreknowledge is of no importance whatever for prayer, providence and prophecy. Hunt offers an ingenious story to show that what God knows about the future might offer practical help on the basis of what he foresees. But this example does not meet Hasker's point, which is that foreknowledge as a corollary of divine omniscience-"foreknowledge that embraces all actual free choices, including those that are yet to be made"-is of no practical impor- tance.Z Selective foreknowledge, of the sort that God might make known to others so that they can bring about changes not yet within the ken of God, is not within Hasker's ken either.