John Martin Fischer on the Puzzle of Theological Fatalism

Hide Journal Menu

<< Previous Article

Current Issue

Next Article >>

SRC_4_2_15-26

John Martin Fischer on the Puzzle of Theological Fatalism

David P. Hunt

Department of Philosophy, Whittier College, Whittier, CA 90608 (USA)

Editor | Gregg D. Caruso, Corning Community College, SUNY (USA).

Correspondence | David P. Hunt, Department of Philosophy, Whittier College, Whittier, CA 90608 (USA); Email: [email protected]

Citation | Hunt, D.P. (2017). John Martin Fischer on the Puzzle of Theological Fatalism. Science, Religion and Culture, 4(2): 15-26.

DOI | https://dx.doi.org/10.17582/journal.src/2017/4.2.15.26

The last fifty years have seen a remarkable resurgence of interest among philosophers in the ancient problem of theological fatalism: how divine foreknowledge can be compatible with human freedom. No one has contributed more to the contemporary debate over this problem than John Martin Fischer. The volume under discussion collects eleven of Fischer’s previously published essays on the problem—five earlier papers, and six that appeared just within the last eight years. It also opens with a substantial overview, itself worth the price of admission, written specially for the book.

What is the problem of theological fatalism? Typical formulations begin by positing a presumptively free action—say, Jones’s mowing his lawn tomorrow—and noting that an infallibly omniscient being (e.g., God, as traditionally understood) would have known beforehand—yesterday, last year, before the agent was even born—that Jones will mow. But then it’s hard to see, when tomorrow arrives, how Jones can refrain from mowing; for his refraining would make God’s prior belief false, and that’s impossible. So Jones doesn’t mow his lawn freely after all. Since nothing in this formulation depends on anything distinctive about Jones, or lawn-mowing, or tomorrow, the same considerations count equally against any action. So fate reigns supreme.

I’d like to begin (spoiler alert!) by disclosing Fischer’s general verdict on this argument. I will then unpack the issues for discussion, dividing my comments into three parts.

In the first place, Fischer holds that the argument succeeds. The argument presupposes an understanding of free agency on which the agent must have access to alternative possibilities, and Fischer believes that the argument successfully demonstrates that no one is free in this sense, given divine foreknowledge. Indeed, no one, to my mind, has probed the argument’s vulnerabilities and established its essential soundness more effectively than Fischer. Of course Fischer is hardly alone in his defense of the argument. “Open theists,” for example, are major defenders of the argument, appealing to it as justification for their denial that God knows the contingent future. But revising the argument’s conception of God is not on Fischer’s agenda. While he’s happy to explore revisionary accounts of divine omniscience (one of which we’ll look at later), he never suggests that this is how the problem should be resolved.1

In the second place, however, Fischer holds that the argument fails. It fails because it purports to challenge something we care about, and the freedom the argument succeeds in annihilating, given divine foreknowledge, is a freedom which, on reflection, we can do without. One reason free agency is important is that it makes us suitable subjects of praise and blame: free agency is needed so we can make sense of the moral (and other “reactive”) attitudes which are ineliminable from our self-conception. We might call this “moral freedom,” but that’s too narrow; our interest in free agency is bigger than this. Fischer has argued elsewhere that our most fundamental commitment to free agency comes from our interest in being the authors of our own stories, an interest that encompasses but is not limited to moral responsibility.2 Call this “significant freedom.” The argument fails, then, because it does not finally succeed in showing that divine foreknowledge is incompatible with the freedom worth caring about: significant freedom.

Thus Fischer isn’t a simple compatibilist or incompatibilist about divine foreknowledge and human freedom. He’s an incompatibilist about divine foreknowledge and what we might call “leeway freedom”: the freedom to do otherwise. But he’s a compatibilist about divine foreknowledge and significant freedom. (This hybrid position is obviously coherent only if significant freedom does not require leeway freedom—more on this later!) Fischer’s own term for this position is “semi-compatibilism.” Why not “semi-incompatibilism,” since that seems equally descriptive of the position? Presumably because the freedom with which divine foreknowledge is compatible is more important than the freedom with which it is incompatible. In sum, the argument for theological fatalism fails in a more important way than it succeeds.

As it happens, I’m in substantial agreement with Fischer when it comes to the big picture sketched above. That means that my work as a commentator is cut out for me! Nevertheless, there should be a few points at which I can embellish or delete as we look beyond the big picture.

Fischer’s Semi-compatibilism

Having identified Fischer’s general position relative to theological fatalism, I’d like now to map it in more detail and situate it within the surrounding territory. Along the way, I will explain why (with one important dissent) I think that Fischer is well-positioned on this map.

1. Begin with the Principle of Alternate Possibilities, or PAP:

A person is morally responsible for an action only if he could have done otherwise.

In an important article on PAP, Harry Frankfurt claimed that “[p]ractically no one . . . seems inclined to deny or even to question that the principle of alternate possibilities . . . is true,” adding that “[i]t has generally seemed so overwhelmingly plausible that some philosophers have even characterized it as an a priori truth.”3 But what’s true of moral responsibility would seem to be equally true of the freedom required for moral responsibility; that is,

A person performs an action freely (in the sense required for moral responsibility) only if he could have done otherwise.

In other words, significant freedom requires leeway freedom. When I refer to PAP in what follows, it’s primarily this freedom version of PAP that I will have in mind.

2. Consider now the question whether free agency is compatible with causal determinism. One powerful formulation favoring an incompatibilist answer to this question is the so-called Consequence Argument, which Peter van Inwagen sums up as follows:

If determinism is true, then our acts are the consequences of the laws of nature and events in the remote past. But it is not up to us what went on before we were born, and neither is it up to us what the laws of nature are. Therefore, the consequences of these things (including our present acts) are not up to us.4

Many delicate questions can (and have) been raised about this argument; nevertheless, Fischer accepts the argument’s conclusion that determinism is incompatible with leeway freedom: the freedom to do otherwise. If this is the freedom worth caring about (i.e., if PAP is true), then insofar as determinism might actually be true, this is a very bad result indeed.5

3. Consider next the question whether free agency is compatible with divine foreknowledge. An argument for their incompatibility, involving God’s foreknowledge that Jones will mow his lawn tomorrow, was given above. We’ve already seen that Fischer finds this argument ultimately compelling; in short, he is an incompatibilist about divine foreknowledge and leeway freedom. As with the Consequence Argument, if this is the freedom worth caring about (i.e., if PAP is true), then insofar as a God with infallible foreknowledge might exist, this is a very bad result indeed.

4. Let causal determinism be the thesis that the state of the world at any future time (e.g., tomorrow) is entailed by a fundamental feature of physical reality (the laws of nature) together with some set of facts about the past (e.g., the state of the world 100 years ago). Divine foreknowledge can likewise be understood as the thesis that the state of the world at any future time is entailed by a fundamental feature of theological reality (God’s infallible omniscience) together with some set of facts about the past (e.g., the beliefs God held 100 years ago). The Consequence Argument notes that, since Jones can’t do anything about the laws of nature and the state of the universe 100 years ago, then given causal determinism, he can’t do anything about his mowing tomorrow: he’s going to do it, and he can’t do otherwise. So determinism is incompatible with free agency. Likewise, the argument for theological fatalism notes that, since Jones can’t do anything about God’s existence and the beliefs God held 100 years ago, then given divine foreknowledge, he can’t do anything about his mowing tomorrow: he’s going to do it, and he can’t do otherwise. So divine foreknowledge is incompatible with free agency. Both the theses that are argued to be incompatible with free agency and the arguments used to demonstrate the incompatibility are strikingly similar in structure—a point of some importance to Fischer (1, 52).

5. Indeed, these arguments are so similar, one might suspect that they stand or fall together. Causal determinism and divine foreknowledge are of course very different theses; one could be true and the other false. But the arguments in question don’t assume the truth of the relevant theses; they’re concerned only with the compatibility of these theses (whether true or false) with free agency. Those are the arguments in question, and anyone who thinks that one of them works while the other doesn’t has some explaining to do. As it happens, many theistic critics of the argument (e.g., Alvin Plantinga, William Lane Craig, et al.) are compatibilists about foreknowledge and incompatibilists about determinism. On Fischer’s view, this split verdict is unsustainable.

6. These arguments are truly worrisome, however, only if PAP is true. In that case, our significant freedom would depend on the falsity of causal determinism and the nonexistence of an infallibly omniprescient deity, and that’s to make the freedom we care about a little too precarious for comfort. But there’s reason to think that PAP may be false. Harry Frankfurt, in the very essay from which I quoted earlier when touting PAP’s near-a priori plausibility, goes on to construct a famous counterexample to PAP. Jones decides to mow his lawn, and acts on this decision, under conditions that would ordinarily make him morally responsible for his action. But this isn’t an ordinary case: lurking in the background is Black, equipped with a sci-fi device programmed in such a way that (1) if Jones decides on his own to mow, the device leaves him alone, and (2) if Jones does not decide on his own to mow, the device intervenes to cause him to decide to mow. Suppose now that Jones does decide on his own, and the device is quiescent. Then Jones’s moral responsibility remains intact. But since (2) is also true, it looks like Jones can’t do otherwise than decide to mow.6 PAP, then, appears to be false. Of course Frankfurt’s counterexample and its implications for PAP have stirred up considerable controversy. Fischer has been a major voice in the ensuing discussion, where he’s come down strongly on the Frnkfurtian side.

7. In Book 3 of the Nicomachean Ethics, Aristotle claims that a person is morally responsible for an action only if he (a) was the genuine source of the action and (b) could have done otherwise. Aristotle appears to assume that these are two ways of saying the same thing—or in any case, that the one condition would be satisfied just in case the other was satisfied. One way to understand Frankfurt-counterexamples to PAP is that they are cases in which Aristotle’s two conditions come apart: the person can’t do otherwise, but whatever it is that eliminates the agent’s alternatives manages to leave the agent’s sourcehood intact. These cases are unusual and even contrived, but they reveal something important about the nature of significant freedom. The moral of Frankfurt cases, according to Fischer, is that we should redirect our attention away from untraveled side roads and back to the actual sequence issuing in the action.7 A person is significantly free when his action flows from him, as its source, in the right way.

8. But what exactly is it for an action to issue from an agent, as its source, in the right way? Fischer has a positive proposal here, involving what he calls ‘guidance control’:

An individual exhibits guidance control to the extent that he acts from his own, suitably reasons-responsive mechanism. There are thus two important components of the analysis: ownership of the actual-sequence mechanism, and reasons-responsiveness of that mechanism. (47)

Elsewhere, Fischer has worked out in great detail the conditions under which a mechanism qualifies as “moderately reasons-responsive” and an agent assumes ownership of the mechanism, but those details aren’t directly relevant to the work under discussion.8 What’s important here is that guidance control does not require access to alternative possibilities (which Fischer calls ‘regulative control’), and an agent with guidance control has everything that’s needed for significant freedom.

9. If PAP is false, this has implications for the two arguments considered earlier. The fact that the Consequence Argument demonstrates the incompatibility of determinism and leeway freedom doesn’t show that determinism and significant freedom are incompatible. To reach this conclusion, more than the mere incompatibility of determinism and alternative possibilities would be needed. For the same reason, the mere fact that the argument for theological fatalism demonstrates the incompatibility of divine foreknowledge and leeway freedom, just by itself, doesn’t mean that divine foreknowledge and significant freedom are incompatible.

10. And here’s a further step one might take: In the absence of PAP, there is no good reason to think that divine foreknowledge and significant freedom aren’t compatible. So affirm that they are compatible. Fischer’s identification of significant freedom with guidance control supports this move, since nothing about divine foreknowledge seems incompatible with the agent’s exercise of guidance control.

11. And here’s a yet further step: Absent PAP, there is no good reason to think that causal determinism and significant freedom aren’t compatible. So affirm that they are compatible. Guidance control, in particular, seems as compatible with determinism as it is with foreknowledge. Fischer offers a challenge to fellow PAP-deniers who balk at embracing compatibilism. How exactly does determinism threaten free agency, beyond its denying agents access to alternative possibilities? Unless he has a good answer to this question, the anti-PAPist who agrees that divine foreknowledge and significant freedom are compatible is in an especially poor position to hold that causal determinism and significant freedom aren’t cmpatible (49-52).

This should provide more than enough context for our purposes. I trust that it describes the territory perspicuously and places Fischer on the map accurately. If not, I’m counting on Fischer to correct me!

My only sense of dislocation comes at the last step. If sourcehood is what’s fundamental to free agency, one can derive different implications for significant freedom from divine foreknowledge and causal determinism. Let source compatibilism be the view that (i) significant freedom is grounded in sourcehood rather than alternative possibilities and (ii) the kind of sourcehood in question is compatible with determinism. Source incompatibilism, in contrast, affirms (i) and denies (ii). A source incompatibilist can (and should) agree with Fischer that divine foreknowledge does not jeopardize free agency, but his reasons are the ones articulated by St. Augustine when he writes:

God’s foreknowledge of future events does not compel them to take place. As you remember certain things that you have done and yet have not done all the things that you remember, so God foreknows all the things of which He Himself is the Cause, and yet He is not the Cause of all that He foreknows. (On Free Choice of the Will III.4)

These reasons obviously do not apply to causal determinism.

Fifty years ago, when things were simpler, it was just compatibilists v. incompatibilists. Compatibilists proposed conditional analyses of “could have done otherwise” that were arguably compatible with determinism, and incompatibilists responded that an agent with a single predetermined future couldn’t really do otherwise. Today’s source incompatibilists are philosophers who would have been incompatibilists tout court in those simpler days, before Frankfurt’s critique of PAP identified sourcehood rather than alternative possibilities as the heart of free agency, and they bring to the new problematic those same incompatibilist intuitions. When source compatibilists propose analyses of sourcehood that are arguably compatible with determinism, such as Fischer’s ownership/reasons-responsiveness account, source incompatibilists respond that an agent whose actions can be traced back to causally determining conditions in place before he was even born isn’t the genuine source of his actions. It isn’t surprising that the old battle lines over the compatibility of determinism and alternative possibilities should re-form over the compatibility of determinism and agential sourcehood.

Each side in this dispute sees the other, fairly or unfairly, as insufficiently consistent. From the source-incompatibilist perspective, Fischer is to be commended for maintaining high enough standards for alternative possibilities that they can’t be satisfied in a causally determined world, but admonished for lowering the standards for sourcehood so that they can be satisfied in a deterministic world. I have some sympathy with this diagnosis. Nevertheless, the source incompatibilist has to steer between Scylla and Charybdis here. If the standards can be set too low (and I’ve been pretty vague about where to draw the line, and why), they can also be set too high. Galen Strawson has (in)famously argued that “ultimate responsibility” requires that one be a causa sui, and that this ideal is incoherent.9 This is a challenge for compatibilists and incompatibilists alike, but clearly the higher one’s standards for sourcehood, the harder it is to avoid Strawson’s critique. Then there’s Derk Pereboom, a determinist whose source-incompatibilist intuitions lead him to set the bar high enough that no one can clear it, the result being that we must reconcile ourselves to “living without free will” (as his book’s title announces).10 As a fellow source incompatibilist (and not a determinist), I think that Pereboom is too pessimistic. Still, Fischer can argue that it’s a strength of his position that he’s “lowered” the standards for sourcehood so that they can be satisfied even in a deterministic world.

Fischer, for his part, charges the source incompatibilist—insofar as he holds that divine foreknowledge doesn’t compromise significant freedom while causal determinism does—with treating similar arguments differently. The source incompatibilist’s response to this inconsistency charge is that what’s relevant isn’t the arguments’ similarity with respect to a condition that isn’t necessary for significant freedom (i.e., their both ruling out alternative possibilities), but their dissimilarity with respect to a condition that is necessary (i.e., their differing implications for sourcehood). Divine foreknowledge doesn’t affect the agent’s role as the source of his action; causal determinism does. Fischer isn’t so sure: in divine foreknowledge cases “there is some condition that is entirely external to the agent (God’s belief) that is sufficient for the behavior in question. In this sense then one could say that the ‘source’ of the behavior is entirely outside the agent” (50). This is not, however, the relevant sense of ‘source’ (as Fischer acknowledges). For source incompatibilists, sourcehood is intimately involved in explanation. In Frankfurt’s counterexample, the explanation for Jones’s action isn’t to be found in the alternative-eliminating sci-fi device, but within Jones himself; in divine foreknowledge cases, the explanation isn’t to be found in God’s infallible forebeliefs—“a man does not therefore sin because God foreknew that he would sin,” as Augustine notes (City of God V.10)—but, again, within Jones himself; but in a causally determined world, a sufficient explanation is to be found in conditions external to Jones. That’s the crucial difference that determinism makes.

Free will is almost bottomlessly mysterious, and I’m not sure that Fischer’s semi-compatibilism isn’t, at the end of the day, the right way to go. All I’ve tried to suggest in these brief comments is that there are principled reasons why someone might resist semi-compatibilism in favor of source incompatibilism—reasons, that is, why someone who holds that divine foreknowledge is compatible with significant freedom, despite its being incompatible with leeway freedom, might take a different position when it comes to causal determinism. Fischer is surely right, however, when he notes that “considerations pertaining to God’s foreknowledge (as well as Frankfurt Cases) appear to provide a challenge that at the very least requires the source incompatibilist to sharpen her analysis of sourcehood” (51).

God’s Knowledge of the Future

The discussion so far has highlighted Fischer’s compatibilism: his view that both causal determinism and divine foreknowledge are compatible with significant freedom. But Fischer, as noted, is only a semi-compatibilist, and his writings on divine foreknowledge put him unequivocally on the incompatibilist side of the debate over whether divine foreknowledge is compatible with leeway freedom: the freedom to do otherwise.

The argument for theological fatalism makes the following moves. It (1) takes as inputs divine forebeliefs (e.g., beliefs God held yesterday about what will happen tomorrow), and (2) imputes a certain kind of necessity to them. (Because the beliefs are past, we’re now stuck with them; it’s too late to do anything about the fact that God held those beliefs yesterday.) It then (3) transfers that necessity to the foreknown actions. Finally, it (4) presents those actions as unfree. This last step relies on PAP, and we’ve already seen how Fischer regards this last step as fallacious, since he believes that Frankfurt-type cases have demonstrated the falsity of PAP. But Fischer has been just as important a defender of the argument’s other steps as he’s been a critic of its last step.

Fischer has interesting things to say in his introductory essay in defense of the transfer principle at work in (3), including suggestions for how the argument can succeed without relying on transfer principles at all (2-6). I have little to add to what he says there. I will reserve my comments for (1) and (2), beginning with the latter.

That God’s past beliefs about the future are now necessary has been at the center of contemporary disputes over the argument. Critics don’t typically deny that the past possesses a kind of necessity relevant to human freedom—as when, e.g., I can’t re-call an intemperate email because I already sent it. Call this necessity—the necessity that obtains when it’s too late for things to be otherwise—‘temporal necessity’, to distinguish it from logical necessity, causal necessity, and so on. The critics’ point is that even if the past is temporally necessary, not all facts about the past are temporally necessary. So-called “hard facts” about the past, which are temporally necessary, must be distinguished from “soft facts” about the past, which aren’t. Suppose it’s true that I boarded the plane just two minutes before the doors were closed. One minute after I boarded this was a soft fact about the past; three minutes after I boarded it was a hard fact about the past. Ockhamism (named after its first practitioner, William of Ockham) is the thesis that God’s past beliefs about the contingent future are soft facts about the past. If this thesis is correct, the argument for theological fatalism runs aground at step (2), because God’s forebeliefs won’t qualify as temporally necessary.

A significant prima facie problem for Ockhamism is that someone’s having held a particular belief yesterday seems as much a hard fact about the past as someone’s having held a party, or a grudge, or someone’s hand, yesterday. I may reject or regret some of my past beliefs, but it’s too late for me to do anything about the fact that I held them at the time. It makes no difference if the belief is about the future—e.g., my belief on 11/7/16, one day before the election, that Clinton would win. It’s not clear why it should be any different for divine belief.

One difference between God’s forebeliefs and my forebeliefs is that his, since they’re infallible, entail the future. Since soft facts about the past also entail the future, it was initially hoped that a simple entailment criterion might be enough to sort facts into the hard and the soft, with God’s forebeliefs ending up amongst the latter. Unfortunately, no simple entailment account succeeds. (E.g., that Jones mowed yesterday is surely a hard fact about the past, but it entails that he won’t mow for the first time tomorrow.) The effort to formulate a satisfactory Ockhamist-friendly entailment account exhausted itself in the 1980s. (Fischer was among its leading critics.)

Lately, there have been renewed efforts, grounded in relations other than entailment, to show that God’s forebeliefs aren’t temporally necessary. Fischer examines some Neo-Ockhamist proposals from Trenton Merricks, Storrs McCall, and Jonathan Westphal, in the book’s last three essays.11

Alvin Plantinga had already noted that God’s forebeliefs are counterfactually dependent on the foreknown future. Plantinga concluded that Jones can refrain from mowing, and if he were to refrain, God would have held a different belief in the past. Fischer rightly maintains that this can-claim, as a response to the argument for theological fatalism, simply begs the question (125-6). The Neo-Ockhamists draw attention to the fact that God’s forebeliefs aren’t just counterfactually dependent on the future; they’re also explained by the future. But it’s not clear how this is supposed to confirm or rehabilitate the agent’s ability to do otherwise. Like Plantinga, such critics are too willing to argue from the agent’s presumed ability to do otherwise to the conclusion that God’s forebeliefs are soft, and in the context of an argument challenging the agent’s ability to do otherwise, this seems dialectically inappropriate.

Still, can’t Jones do anything about what God believed 100 years ago? After all, God’s belief depends on what Jones does, not the other way around; so the initiative seems to lie with Jones. It’s some such thoughts as these that make it seem like the Ockhamists are onto something. But it’s important to remember that the dispute here isn’t over whether divine foreknowledge is compatible with significant freedom. Fischer agrees that they are compatible. What Plantinga, Merricks, et al., are arguing is that foreknowledge is compatible with leeway freedom: the freedom to access alternative possibilities. That is, they want to reject the foreknowledge argument’s conclusion while holding onto PAP. But the explanatory priority of Jones’s action, and the explanatory dependence of God’s forebeliefs on that action, only show how Jones can remain the genuine source of his action; they don’t show how God’s forebeliefs can be soft facts about the past, leaving Jones access both to a future in which he mows and a fure in which he doesn’t mow, as required by PAP.

This is especially clear, I believe, if we look at a case involving one of van Inwagen’s “freedom-denying prophetic objects.”12 Suppose that God, knowing that Jones will mow 100 years in the future and wishing (for inscrutable reasons of his own) to make a record of the fact, inscribes on the rocky surface of a distant planet the message, JONES MOWS HIS LAWN ON 4/28/19. Now fast-forward 100 years. Jones mows, and he is the genuine source of his action. Grant the Neo-Ockhamists’ point that the existence of the inscription is explanatorily dependent on Jones’s mowing. “But how exactly,” Fischer asks, “does the dependence point in any way vitiate—or even address—the point about the fixity of the past?” (213). That the inscription exists, that it was caused by God, that it says what it says, are surely hard—temporally necessary—facts about the past, if any facts are. So the only futures open to Jones are ones that include these facts, and in none of these does he refrain from mowing (because otherwise the divine inscription would be false). But what’s true in the inscription scenario seems equally true in the foreknowledge scenario. The Ockhamist idea that Jones can act in such a way that God would have held different beliefs than he in fact held, just because his beliefs are explanatorily dependent on what Jones does, seems to treat God’s past beliefs as less definite and real than a rocky inscription, and that’s a position that classical theists—who regard God as more real than anything else—should be the last to endorse.

The Neo-Ockhamists’ emphasis on explanatory dependence doesn’t support their desired conclusion: that the agent can do otherwise because God’s forebeliefs are soft facts about the past. Instead, it reinforces Fischer’s position that agential sourcehood is unaffected by divine foreknowledge. Since sourcehood is the condition associated with significant freedom, there’s no need to twist ourselves into Ockhamist knots turning God’s forebeliefs into soft facts in order to salvage the alternatives condition.

Let’s turn now to step 1 of the argument. To produce unfree actions as outputs, the argument requires divine forebeliefs as inputs. If there are no divine beliefs about the contingent future, however, the argument is nipped in the bud. But how could that be, since God is omniscient?

For someone X to know a proposition p is (i) for X to believe that p, (ii) for p to be true, and (iii) for X to be in what Fischer calls a knowledge-conferring situation (or KCS) with respect to p. (Different theories of knowledge will provide different accounts of what constitutes a KCS.) One challenge to the possibility of foreknowledge in a contingent universe comes from (ii), the truth-condition; for if there are no future-contingent truths, as many philosophers believe, there is no foreknowledge of them. Fischer, however, focuses on condition (iii). In the typical case, what confers knowledge on a true belief is largely the evidence one has. But if contingent, the future isn’t entailed by any past or present facts—given all of the facts available as evidence, the future could always turn out differently. So how could anyone really know how it will turn out?

Defenders of the argument must explain how God can know the contingent future—divine foreknowledge isn’t something that can just be taken for granted. Fischer has an account to offer, and it rests on two very plausible claims about knowledge. One is fallibilism: knowledge is possible even when certainty isn’t. (If fallibilism is false, we have precious little knowledge.) The other is the idea that knowledge is unified, by which Fischer means that what gives a true belief that p its status as knowledge, whether the knower is man or God, is the same thing: being in a KCS with respect to p. Fischer puts these two ideas together as follows. First, he notes that human beings sometimes know the contingent future. For example, it could be that Jones’s neighbor Smith believes that Jones will mow tomorrow, Jones does mow tomorrow, and Smith’s intimate knowledge of Jones’s habits, etc., puts him in a KCS with respect to Jones’s mowing tomorrow. Second, if Smith can satisfy the conditions for knowing that Jones will mow, so can God. (One theologically awkward consequence of denying this is that humans could end up knowing more than God knows!) In short, knowledge of the contingent future isn’t impossible; even human beings can pull it off, and God does it just the way that we do it: by being in a KCS with respect to a future-contingent proposition.

That’s the short answer to the “how” question, but a somewhat longer answer is needed to show how the fallibilist epistemology that makes such foreknowledge possible can apply to an infallible deity. One can be in a KCS with respect to p, and consequently believe that p, and yet p turn out to be false. How does God avoid this situation? Fischer’s proposal is that God, having acquired his beliefs about future-contingents in the same KCS-based way as humans, “bootstraps” his way to certainty by combining these beliefs with knowledge of his own essential inerrancy—hence the name for Fischer’s approach: the “Bootstrapping View.”

This is in some ways an elegant proposal for how God can know the contingent future. Still, I’m not a convert. (Neither is Fischer, since he acknowledges both of the points I’m about to make!)

First, there’s a problem with how much the Bootstrapping View really explains. Suppose that God is in a KCS with respect to Jones’s mowing tomorrow, and he’s also in a KCS with respect to Jenkins’ mowing tomorrow—a KCS at least equal in its knowledge-conferring power to God’s KCS with respect to Jones’s mowing tomorrow. Suppose further that Jones will mow tomorrow, but Jenkins won’t. God must believe (and therefore know, with certainty) that Jones will mow, on pain of knowing less than human beings (e.g., Jones’s neighbor) might know; but God mustn’t believe that Jenkins will mow, on pain of believing falsely. How does God know that he should believe in the first case and not in the second? The KCSs are on a par. Their truth isn’t on a par, but this difference isn’t available to God, since the Bootstrapping View does not grant to God a “direct apprehension” of the future (32). What is available to God is his certainty that Jones will mow and his lack of certainty that Jenkins will mow. The problem is that God bootstraps to certainty; “God finds himself with the belief that Jones will mow” and combines this belief with knowledge of his own infallibility to reach certainty (44-5). His certainty can’t explain why he forms the one belief, and not the other, in the first place.

“We then have at least a mystery,” Fischer allows (44). This isn’t necessarily a deal-breaker—as Fischer rightly observes, “every major view about God’s knowledge of the future has at least a mystery associated with it” (45). Consider, for example, the idea that God has a built-in crystal ball or “time telescope” providing him direct access to the future. There’s an up-front mystery to time telescopes. The Bootstrapping View eschews mystery at the front end, since God knows the contingent future in the same way that we can know it; but it invokes mystery at the back end, to ensure that God believes a future-contingent proposition with respect to which he’s in a KCS only when the proposition is true. It isn’t easy to compare one mystery with another, since they’re both so . . . mysterious. I’ll come back to this in a moment.

The second problem with the Bootstrapping View is that it doesn’t provide God the complete foreknowledge promised by the traditional understanding of omniscience, since God forebelieves only when he’s in a KCS, and many future-contingent truths aren’t preceded by a KCS (38). In fact, the situation here is even worse than Fischer acknowledges, since KCSs for future-contingent truths become vanishingly rare the further back one goes. God may be in a KCS today with respect to Jones’s mowing tomorrow, but it’s doubtful that a KCS was available to him a year ago, and utterly incredible that he could have been in a KCS with respect to Jones’s mowing 100 years ago. The foreknowledge God acquires on the Bootstrapping View is very limited indeed.

Of course, if it isn’t metaphysically possible for God to foreknow anything more than this, them’s the (theological) breaks. But let’s return to the mystery competition between Bootstrapping and time telescopes. Is it evident that the former is less mysterious than the latter? It isn’t evident to me, anyway. But if these mysteries are roughly on a par, time telescopes win hands down, since they offer more bang (complete rather than partial foreknowledge) for the mystery.

What Is the Problem of Theological Fatalism?

Suppose that Fischer is right in holding that the argument for theological fatalism fails because it relies on PAP. Suppose further that Fischer is right in maintaining that the argument is defensible at every other point. I addressed the first of these in section 1 and the second in section 2, and despite dissenting from some of his conclusions (e.g., semi-compatibilism) and some of his proposals (e.g., bootstrapping), I am in essential agreement with Fischer’s overall assessment of the argument. But what about the argument’s significance? Why care about the problem of theological fatalism?

The argument has an obvious significance for classical theists. If the argument succeeds, theists must either deny human freedom or revise their concept of God. Those are big stakes. In practice, most philosophical theists who accept the argument have sought ways to deny the problematic foreknowledge without violating the requirements of perfect-being theology. Boethians do this by placing God and his knowledge outside time; open theists do it by denying that there are future-contingent truths for God to know, or by holding that knowledge of such truths (should they exist) would be metaphysically impossible. As for theists who reject the argument, they will need to figure out where exactly it goes wrong, and this can lead to serious rethinking of the substantive philosophical assumptions upon which the argument rests—assumptions about the necessity of the past, the nature of belief, power entailment principles, and so on. Whatever they conclude about the argument’s success or failure, theists clearly have skin in this game.

Fischer does not have this stake in the controversy. He is an agnostic—yet he has spent much of his career exploring this argument. This makes perfect sense if there is a non-theological core to the argument. I believe that there is, and I will elaborate on this idea in what follows. Everything I know about Fischer’s work suggests that he would find what I am about to say congenial, but I may press the point further than he would be willing to go—and I might even be mistaken in my congeniality assumption! I look forward to Fischer’s thoughts on this.

There are two questions I’d like to raise about the argument’s larger significance. The first is whether it poses a primarily theological problem. Marilyn Adams, in discussing the problem of evil, makes a useful distinction between taking the problem of evil atheistically—i.e., as offering “a positive disproof of divine existence”—and taking it aporetically—i.e., as “generating a puzzle” for the one who accepts its premises.13 In the latter case, “one may remain confident that the conclusion is false but see the argument as creating a difficulty for anyone who rejects it: that of explaining how the prima facie plausible premisses are not all so acceptable, the inferences not so evident, as they seem.”14 Since Adams cites Zeno’s paradoxes as examples of arguments that ask to be understood aporetically, let’s make this our point of comparison.

In Zeno’s famous Achilles Paradox, we’re presented with an argument, starring a tortoise, renowned for its slowness, and Achilles, Homer’s “fleetest of the Achaeans.” The argument’s conclusion: if Achilles gives the tortoise a head start in a footrace, he can’t pass the tortoise. There’s clearly something wrong with this argument (whatever the details); what’s more, we’re within our epistemic rights in believing that there’s something wrong with the argument even if we don’t know, and perhaps have no idea, what is wrong with it. There’s a puzzle to be worked through here, and doing so might yield important conceptual benefits; but one shouldn’t rely on this argument for any conclusions about Achilles! If we’ve got Achilles (or tortoises) wrong, it will be for some reason other than the one supplied by Zeno’s Achilles Paradox.

Return now to the problem of theological fatalism. An argument is given, starring God, an infallibly omniscient being, and a particular action—say, Jones’s mowing his lawn tomorrow. The argument’s conclusion: Jones doesn’t mow freely. On reflection, one’s response to this argument might parallel one’s response to Zeno’s Achilles Paradox. Here’s an intuition-pump supporting this response. Suppose that Jones’s mowing is, in every other respect, an exemplary instance of free agency by anyone’s standards: it is done intentionally, with full knowledge of the relevant circumstances, not under coercion or duress, in harmony with the agent’s second-order desires, undetermined by prior causal conditions, and so on. Now add one more condition: before Jones was even born, God infallibly believed that Jones would mow tomorrow. How could that additional condition have as a consequence that Jones doesn’t mow freely? There are conditions that clearly would warrant such a reassessment—for example, if it were added that Jones was under the influence of drugs or post-hypnotic suggestion, or controlled by Martians via a chip implanted in his brain. But the idea that the mere presence of an infallible foreknower could make this kind of difference is deeply puzzling. We have good reason to suspect that the argument goes wrong, even if we’re unable to determine exactly where it goes wrong—good reason, in other words, to take the argument aporetically. But then we shouldn’t (pace Boethians and open theists) rely on the argument for any conclusions about God. If we’ve got God wrong, it will be for reasons other than those supplied by the argument for theological fatalism.

I suggest, then, that the problem raised by this argument is not primarily theological in nature. The Achilles Paradox isn’t really about Achilles, and insofar as the problem of theological fatalism is aporetic, it isn’t really about God. Of course, if God exists, he is implicated in this problem—theists do have skin in this game. And by the same token, if Achilles exists, he’s implicated in Zeno’s Paradox. (Could he really have been that slow?) But if Achilles doesn’t exist, we still have the puzzle Zeno bequeathed to us; and if God doesn’t exist, we still have the puzzle of how a mere belief about the future—even one that’s infallible and divine—could undermine Jones’s credentials as a free agent. That’s a puzzle that can fascinate an atheist or agnostic as well as a theist.15

The second question I’d like to raise is this: Assuming that Fischer’s analysis is essentially correct, is the puzzle of theological fatalism simply the passive beneficiary of developments elsewhere in philosophy, or does it have any contribution of its own to make? I’d like to advocate for the latter.

There’s an assumed flow for anyone traveling the territory I mapped out in the first section of my comments. First, one’s commitment to PAP is challenged by an unusual counterexample. Second, one notices that the argument for theological fatalism presupposes PAP. So third, and following from steps one and two, one concludes that the argument for theological fatalism fails. There may be other problems with the argument as well—other reasons one might conclude that the argument fails—but it certainly fails because it relies on PAP, and PAP is false.

In this narrative, the central fact is the refutation of PAP by Frankfurt-style counterexamples. If PAP is false, it can’t be relied on to give the right results in divine foreknowledge (or causal determinism) scenarios, or in cases involving coercion or mind-control, or when a counterfactual intervener like Black is present. Of course this last isn’t just another implication of PAP’s defeat; it’s a counterexample showing us that PAP is false. But what gives it this status? A counterexample to PAP needs to be a case in which it’s clear that the agent acts freely, in the sense required for moral responsibility, while lacking access to relevant alternatives. Frankfurt cases arguably qualify. Coercion and mind-control cases don’t, since it’s not clear in such cases that the agents’ lack of alternatives isn’t what compromises their significant freedom. Deterministic cases don’t qualify either, since the judgment that the agents are significantly free in such cases begs the question unless there’s already good reason to think that PAP is false. But divine foreknowledge cases do qualify as counterexamples to PAP—or so it seems to me.

Why is it intuitively obvious in Frankfurt-type cases that the agent is morally responsible, despite lacking alternatives? Here’s Frankfurt’s own explanation: 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;” for this reason, appealing to Black and his device “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.”16 But this is no less true in a foreknowledge scenario. God’s infallible beliefs about the future rule out alternative possibilities without their making any difference to, or playing any role in, or helping in any way to explain the future. Judging that Jones remains significantly free in a foreknowledge scenario doesn’t require a prior argument against PAP, any more than judging that Jones is significantly free in a Frankfurt scenario requires a prior argument against PAP. Both cases are ones in which we see that PAP is false. Divine foreknowledge is itself a perfectly acceptable counterexample to PAP.

This suggestion is particularly compelling, I believe, if one agrees with Fischer that the argument fails only at the point where it relies on PAP. If it fails at some other point as well, then it fails to demonstrate that agents in a divine foreknowledge scenario lack alternatives; but if they don’t lack alternatives, the fact that we judge them to be significantly free raises no issues with PAP. So the employment of a divine foreknowledge scenario as a counterexample to PAP fits best with a positive assessment of the other steps in the argument for theological fatalism.

One might wonder, however, about the dialectical appropriateness of this proposal. The argument for theological fatalism is essentially the argument that, in a divine foreknowledge scenario, the agent isn’t free. The theological “counterexample” to PAP that I’ve just proposed is essentially that, in a divine foreknowledge scenario, the agent is clearly free. That’s not much of a response! But things aren’t as bad as they seem. Both the argument and the counterexample involve the same divine foreknowledge scenario, but they hold the scenario up to the light in different ways. The argument says, “Look, there are no alternatives here! So the agent isn’t free.” The counterexample says, “Look, divine foreknowledge isn’t in any way interfering with, or even explaining, the person’s action. Despite the individual’s lack of alternatives, this is one free agent!” The latter is more than just a simple denial of the former. If divine foreknowledge provides the scenario in which one sees that PAP is false, as it did for Augustine, it’s not clear why one can’t conclude forthwith that the argument for theological fatalism fails.17

Not only does divine foreknowledge provide its own counterexample to PAP; it arguably provides a better counterexample than do Frankfurt-type cases. One problem with Frankfurt’s original counterexample is that PAP’s defenders keep discovering residual alternatives that Black’s device isn’t equipped to extinguish, and then arguing that these alternatives are sufficient for PAP (or some neighboring principle that upholds the spirit of PAP). PAP’s critics are kept busy tamping down “flickers of freedom,” as Fischer calls them, without any guarantee that another won’t flare up. Another problem, leading to the so-called “dilemma defense” of PAP, stems from the role of “prior signs” as triggers for Black’s intervention; for either the prior sign is causally determinative of Jones’s decision (in which case it begs the question to judge that Jones is free), or it is not (in which case it seems that Jones can always decide otherwise, thereby preserving PAP). But divine foreknowledge scenarios are immune to both objections. Divine foreknowledge eliminates all alternatives, guaranteeing that it got all the relevant ones. And divine foreknowledge doesn’t rely on prior signs, so it avoids that whole can of worms.

In sum, the problem of theological fatalism doesn’t depend for its solution on independent reasons for rejecting PAP; if anything, the case against PAP derives support from theological fatalism. One can confidently reject PAP without having to wait for closure in the debate over Frankfurt cases.

* * *

Given our secular times, especially in the West, it must seem strange to an observer from outside the discipline that the problem of theological fatalism should attract so much serious attention from academic philosophers. In fact, the last fifty years has been a Golden Age for theistic philosophy. (How that came about is another story.) But this doesn’t explain the interest of non-theists like Fischer. What anyone who delves into the problem soon discovers, however, is that divine foreknowledge provides an extreme environment (whether real or fictional) in which some of our most fundamental beliefs about time, possibility, and agency can be tested and refined. That’s what philosophy is all about. One needn’t be a theist to puzzle over the questions raised by divine foreknowledge, any more than one need accept the possibility of time travel to puzzle over the (similar) issues raised in time travel stories, or believe in demons to puzzle over the epistemological questions raised by Descartes’ demonic deceiver.

Endnotes

[1] One exception is the “Geachian” account of foreknowledge, on which the truth about the future can change. This account is completely compatible with free agency, and Fischer gives it serious consideration in one of the volume’s essays, “Engaging with Pike” (pp. 175-8), co-written with Patrick Todd and Neal Tognazzini. I think it’s oversold, but I can’t go into that here.

[2] See his “Stories and the Meaning of Life” in Our Stories: Essays on Life, Death, and Free Will (Oxford, 2009).

[3] “Alternate Possibilities and Moral Responsibility,” Journal of Philosophy 66 (1969), p. 829.

[4] Peter van Inwagen, An Essay on Free Will (Oxford, 1983), p. 16.

[5] Whether quantum indeterminacy leaves enough—and the right kind of—room for free will to establish itself in an otherwise deterministic world is a vexed question.

[6] I’ve substituted our prosaic case of lawn-mowing for Frankfurt’s original case of murder.

[7] Here’s an analogy I find useful. Red spots aren’t the same thing as measles. Measles is the underlying condition; red spots are a symptom that accompanies measles under normal conditions, but could be absent under abnormal conditions (e.g., if the person is immunocompromised). Likewise, access to alternative possibilities isn’t the same thing as free agency. Free agency is the underlying condition; alternative possibilities are a symptom that accompanies free agency under normal conditions, but could be absent under abnormal conditions (like those obtaining in a Frankfurt scenario).

[8] E.g., in John Martin Fischer & Mark Ravizza, Responsibility and Control: A Theory of Moral Responsibility (New York: Cambridge University Press, 1998).

[9] “The Impossibility of Moral Responsibility,” Philosophical Studies 75 (1994), pp. 5-24.

[10] Living without Free Will (Cambridge, 2001).

[11] In his “Truth and Freedom,” Philosophical Review 118 (2009), pp. 29-57, Merricks explicitly denies any connection with Ockhamism, but I think he’s simply wrong about that (and Fischer agrees).