The Independence Solution to the Problem of Theological Fatalism

The Independence Solution to the Problem of Theological Fatalism Ryan Wasserman Western Washington University Origen of Alexandria famously responded to the threat of theological fatalism by focusing on the explanatory relationship between future events and God’s foreknowledge. In his commentary on Romans he writes that …even if we should conceive of foreknowledge according to the popular understanding, it will not be because God knows that an event will occur that it happens; but, because something is going to take place it is known by God before it happens. (c.246/2002, p.90) Much has been written about this idea recently, but the literature has focused almost exclusively on the second half of Origen’s thesis and the idea that some of God’s past beliefs depend on our future actions.1 I believe that this is a mistake. Even if God’s beliefs do depend (in some sense) on our actions, the more important point is that our actions do not depend (in the right way) on God’s beliefs. This, I will argue, is the key to reconciling human freedom and divine foreknowledge.2 I begin by reviewing the standard argument for fatalism and by highlighting an important ambiguity in one of its guiding principles (section 1). I then argue against the traditional interpretation of that principle and in favor of an alternative reading (section 2). Finally, I show how this new reading leads to a novel solution to the problem of theological fatalism (section 3). 1 For recent discussions of this approach, see Rota (2010), Merricks (2011), McCall (2011), Westphal (2011), Fischer and Tognazzini (2014), Byerly (2015), Todd and Fischer (2015), Swenson (2016, 2017), Reichenbach (2017), van Riel (2017), Cyr and Law (2020), and Wasserman (Forthcoming). Much of this literature puts things in terms of dependence, although Origen’s primary concern is with explanation. In my view, it is important to keep these things distinct, since there can be explanation without dependence (as in cases of overdetermination) and dependence without explanation (at least for some forms of dependence). See Wasserman (Forthcoming) for further discussion. 2 This, of course, is not a novel suggestion. The focus on the first part of Origen’s thesis arguably goes back to Augustine, although his response differs importantly from my own. 1 1. Two Kinds of Fixity There are many ways of stating the traditional argument for fatalism, but one natural starting point is the thought our freedom is constrained by what has already taken place. Here is one way of stating this idea, due to John Martin Fisher: Fixity of the Past (FP): For any action Y, agent S, and time t, if it is true that if S were to do Y at t, some fact about the past relative to t would not have been a fact, then S cannot at t do Y at t. (1994, p.78) This principle may require some qualifications (see below), but the basic idea seems plausible: none of us can perform any action that would require the past to be different from how it actually is.3 From here, it is a short step to the doctrine of theological fatalism. For suppose God exists necessarily, and that he is both essentially omniscient and essentially eternal. Suppose further that God’s eternality involves existing at every time and that his omniscience requires believing every truth (as well as disbelieving every falsehood). Finally, let us suppose that a certain individual—Jones—is about to sit down at time t. Presumably, if it is now true that Jones will sit at t, then this proposition was also true a thousand years in the past. Moreover, given God’s eternality, he existed at that time and, given his omniscience, he believed (then) that Jones would sit at t. This is a fact about the past, relative to t. Moreover, given God’s essential features, this would not have been a fact, had it not been the case that Jones was going to sit at t. After all, in that case it would have been false that Jones was going to sit at t, in which case God’s omniscience would have prevented him from believing that proposition. Given FP, it follows that Jones cannot avoid sitting at t. Indeed, it follows that Jones cannot do anything other than what he actually does. Since the same thing would apply to each of us, we can conclude that everything we do is unavoidable. In other words, the doctrine of fatalism is correct.4 3 It is also plausible to think that the pastness of the relevant events is what explains our inability in these cases: none of us can perform any action that would require the past to be different because that would require the past to be different. For more on FP, see Fischer (2011). 4 For the classic statement of this argument, see Pike (1965). 2 There are many different responses to this argument, but I am convinced that a proper analysis begins with an important distinction. This distinction can be brought out by a simple case of time travel.5 Suppose that Smith is a time traveler who departs from the future (at t+) and arrives just before Jones sits down (at t). After observing this event, Smith travels even further into the past where, upon arrival, she sits down to record the events of her journey (at t–). There are two very different way this recording might go. First, if Smith is simply writing things down for her own recollection, then it would be natural for her to record those events in the order of her experience: I departed in my time machine at t+. I observed Jones sit at t. I appeared in my time machine at t–. On the other hand, if Smith is recording these events as part of a history book for non-time travelers, then it might make sense for her to record the events in the order of their occurrence: A time traveler will appear in her time machine at t–. A time traveler will observe Jones sit at t. A time traveler will depart in her time machine at t+. These different orderings correspond to what David Lewis (1976, p.146) calls “personal time” and “external time”. External time, for Lewis, is “time itself”, which we can think of as the ordering of events that would be given by an Objective World Historian. 6 Personal time, in contrast, is the “time of a particular time traveler: roughly, that which is measured by his wristwatch.” (ibid) Less roughly, we can think of personal time as the ordering of events that would match the underlying causal relations in the life of a particular person.7 To explain in more detail: it is widely assumed that identity-over-time requires causation, so that a person at one time is only identical to a person at another time if (for example) the psychological state of the one is caused, in the right sort of way, 5 The idea that time travel might teach us something important about fatalism is suggested by Craig (1987, Chapter 8), Murray and Rea (2008, pp.52-54), and Swenson (2016, pp.664-665), among others. 6 Given contemporary physics, it is unclear whether there is any such thing as external time in Lewis’s sense. See Wasserman (2018, pp.67-69). I will set this complication aside in what follows. 7 See Wasserman (2018, pp.7-8). We can also extend this notion of “personal” time to things that are not persons (watches, time machines, etc.). 3 by the psychological state of the other.8 In most cases, this kind of causation will follow the external order of time. For example, my current psychological state “follows after” yesterday’s psychological state in both the order of causation and the order of time. However, these things will come apart in cases of backward time travel. For example, Smith’s t– state comes before her t+ state in time, but it comes after her t+ state in the order of causation (what she believes, etc. at t– is caused, in part, by her experiences at t+). This notion of personal time can then be extended outward to include other events that play a role in the life of an individual. For example, Jones’ sitting at t is a cause of what Smith sees at that time, and what she sees at that time is a cause of what she writes down at t–. So, we can say that the sitting comes before the writing, according to Smith’s personal time.9 We can even extend this notion of personal time beyond the limits of a person’s life. For example, if Smith’s parents first met in the distant future (long after t+), there would still be a sense in which that event is a part of her past when she sits down to write in her journal at t–. After all, her parents’ meeting is a cause of her very existence, and is thus a (partial) cause of everything that she does at t–. So, at a minimum, we can say that the causal past of an agent includes all of those events that plays a causal role in her experiences and behavior.10 Various details will need to be filled in but, for now, the main point is that there are two different ways of thinking about time and, hence, two different ways of understanding FP. Since external time concerns real time and personal time concerns causation, we can state these principles as follows: Fixity of the Temporal Past (FTP): For any action Y, agent S, and time t, if it is true that if S were to do Y at t, some fact about the external past relative to t would not have been a fact, then S cannot at t do Y at t. Fixity of the Causal Past (FCP): For any action Y, agent S, and time t, if it is true that if S were to do Y at t, some fact about S’s causal past relative to t would not have been a fact, then S cannot at t do Y at t. 8 I assume the psychological condition for illustrative purposes only. The important point is that identity-over-time requires a causal connection of some kind or other. 9 As Lewis (1976) notes, we must be careful about how far we “extend” this notion of personal time. See also Wasserman (2018, p.5). 10 The notion of the causal past and its relation to issues of freedom is more complex than it may at first appear. See, for example, the argument in Rea (2015) and the discussion in Wasserman (Unpublished). 4 The first of these principles says that freedom is constrained by the external past— that an agent is only free to do those things that are compatible with all of the temporally earlier facts. FCP, on the other hand, says that an agent’s freedom is constrained by her personal past—that she can only do those things that are compatible with the facts about her own causal history.11 Given this distinction, there are two different ways of understanding the traditional argument for theological fatalism. So, in order to assess that argument, we must first adjudicate between these readings. 2. Against the Temporal Reading Many philosophers seem to have had something like FTP in mind when discussing the topic of fatalism. For example, in his classic paper, Nelson Pike specifically refers to times and temporal order when stating his version of FP: “No action performed at a given time can alter the fact that a given person held a certain belief at a time prior to the time in question.” (1965, p.33) However, there are at least four different reasons to think that this approach is mistaken, and that fixity should be understood in terms of causation, rather than time. First, FCP can explain everything that FTP can, and it can do so in a more satisfying manner. To see this, first consider the fact that I spilt my milk this morning at breakfast. This may have been upsetting, but there’s no use crying over it now. After all, that event is in the past—both temporally and causally—so FTP and FCP will agree that there’s nothing I can do about it now. The same thing will be true for all other events in my causal history, since all of those events are also earlier in time. Now consider an event that is in my temporal past, but which is not a part of my causal history—perhaps Jones’ sitting down at t. Since that event is in the temporal past, FTP correctly says that there’s nothing that I can do about it now. However, if we stipulate that the event is not a part of my causal history, then FCP will remain silent on the matter. This might seem like a problem for FCP, but in fact the opposite is true. After all, if I were in possession of a time machine, then there would be something that I could do about Jones’ sitting at t—namely, go back in time and stop him. So, while my inability to stop that event is partly due to the fact that it is in the (temporal) past, it is also partly due to the fact that I lack access to 11 For a closely related idea, see Rea (2015, p.272). Fischer (1994, pp.78-79) also distinguishes between “causal” and “non-causal” interpretations of FP, but his distinction is very different from the one that Rea and I have in mind. 5 the past.12 FTP ignores the second part of this explanation, and is for that reason incomplete. A second reason to prefer FCP is that it can help to explain things that FTP cannot. Suppose, for example, that Smith travels back in time to when her grandfather was a baby. There are many different things that Smith might do at that time, but one thing we know for sure is that she will not kill her grandfather. After all, if Grandfather were killed, then Father would not be born, in which case Smith would not be born, in which case she would not be able to travel back in time and kill Grandfather. FCP gives us a straightforward explanation of what is going on in this case: Grandfather’s not being killed is a part of Smith’s causal history, even when she is visiting the (temporal) past. Moreover, Smith’s killing Grandfather would require this fact about her causal history to be different than it is. Hence, FCP tells us that Smith cannot kill her grandfather at that time.13 The same thing is not true for FTP. To see this, suppose for the moment that God does not exist14 and that the laws of nature are indeterministic (in the sense that the temporal past, together with the laws, does not causally necessitate the future). Given the first assumption, Smith’s killing Grandfather would not require any of God’s past beliefs to be different and, given the second, it would not require any of the physical facts about the past to have been different. In other words, there are other possible worlds that intrinsically duplicate the world in question up until the relevant time, but where Smith successfully kills (baby) Grandfather.15 But, in that case, FTP cannot explain why Smith lacks the relevant ability. Of course, FTP is only a necessary condition on ability, so there is no reason to think that it must provide such an explanation. But the point is that FCP does provide such an explanation, and that is a mark in its favor. 12 Compare: I cannot visit Disneyland in the next few hours since Disneyland is far away in California. This explanation is true, but incomplete. My inability to visit Disneyland in the next few hours is partly due to the fact that it is far away and partly due to the fact that I do not have access to an airplane on such short notice. 13 Note that this reasoning would not apply to other people at that time, since Grandfather’s not being killed would presumably not be a part of their causal pasts. 14 One might think that this is out of place, since we have been assuming that God exists necessarily. However, the ordinary practice of philosophy suggests that hypothetical reasoning is not limited by metaphysical possibility. After all, one can reason about the implications of an essentially omniscient being, even if one believes that such a being is metaphysically impossible. 15 For example, there are worlds in which Smith kills Grandfather, but where his reproductive material is saved so as to produce Father (via in vitro fertilization), so that Father can go on to sire Smith (thus allowing her to develop time-travel technology and go back in time to kill Grandfather). For other examples of this kind, see Carrol (2016). 6 A third reason for preferring FCP is that FTP is simply incorrect. This much is old news.16 Consider, for example, the proposition that Jones would sit at t. This proposition was true at t–, and this is a fact about the past, relative to t. Moreover, this fact would not have been a fact, had Jones not sat at t. So, given FTP, it follows that Jones is unable to refrain from sitting down at that time. The problem is that most people—including those who defend the argument for theological fatalism—will reject this line of reasoning.17 The mere fact that it was true that Jones was going to sit at t is not enough to undermine his ability to do otherwise. One way to avoid this result would be to restrict FTP to “hard” facts about the past.18 According to this approach, the fact that God believed, in advance, that Jones was going to sit at t is a hard fact about the past, whereas the mere fact that it was true, in advance, that Jones was going to sit at t is a “soft” fact about the past. If this much is correct, then the defender of theological fatalism can uphold a modified version of her original argument without endorsing the reasoning behind logical fatalism. However, the problem with this move is familiar. In order to provide a satisfactory response to the original worry, the theological fatalist owes us an explanation of what it is for a fact to be hard (as opposed to soft), where this explanation draws the distinction in the way required by her argument (putting future-tensed truths on one side of the divide and divine beliefs on the other). This has proven to be an exceedingly difficult task.19 In any case, the important point for our purposes is that FCP completely avoids these issues. After all, it is not as if the truth of the relevant proposition (that Jones was going to sit at t) is a cause of his sitting at that time. Nor is it causally relevant to anything else in Jones’ life. But, in that case, it is irrelevant to FCP, and thus provides no reason to doubt Jones’ ability to do otherwise. In this way, FCP completely sidesteps the issues of hard facts, while simultaneously exposing the error in the standard argument for logical fatalism. A fourth and final reason for preferring FCP is that FTP is ultimately unsalvageable. Even if the principle can be amended so as to avoid the preceding problem, there are other examples that strike at the very heart of the view. Consider, for example, the original case of Jones’ sitting at t. In our earlier version of the story, Smith is merely an observer. But suppose that, before traveling back in time, she considers her other options. In particular, she considers the question of whether or 16 17 18 19 See Adams (1976), Hoffman (1979), and the overview in Fischer (1994, Chapter 3). See, for example, Pike (1965, pp.35-36), Hasker (1989, pp.76-77), and Zagzebski (1991, p.14). See, for example, Adams (1976). See, for example, the survey of failed attempts in Fischer (1994, Chapter 6). 7 not she can stop Jones from sitting at t. Intuitively, the answer to this question is yes. After all, Smith has a time machine that will take her back to Jones’s exact location and, once there, it would be a trivial matter to delay his sitting. This, however, would directly contradict FTP since the relevant event is in the temporal past, relative to t+. Hence, FTP incorrectly limits the abilities of those who travel in time.20 Indeed, the results suggested by FTP seem downright bizarre. To illustrate, suppose that Smith ends up travelling back to a time just before t, and finds herself standing a few feet away from Jones as he prepares to sit down. If we continue to assume the falsity of causal determinism (and temporarily bracket the issue of theological fatalism), then everyone should agree that Smith can, at that time, stop Jones from sitting at t. Certainly, nothing about FTP rules this out, since Jones’ sitting is no longer in the temporal past once Smith emerges from her machine. Moreover, we can suppose that Smith has whatever other features are required for action in this case (she has control over her body, there is nothing to stop her from pulling the chair out from under Jones, etc.) But, in that case, it seems absurd to think that Smith lacked the relevant ability before getting into her time machine. After all, the following principle seems highly plausible: For any agent S, actions x and y, and times t1, t2, and t3, if (i) S can, at t1, perform x at t2 and (ii) S’s performing x at t2 would make it the case that S can perform y at t3, then (iii) S can, at t1, perform y at t3 (in the sense that it is among her options going forward). For example: it is true right now that I can depart on an airplane for LAX tonight. Moreover, if I were to depart on an airplane for LAX tonight, I would then be in a position to visit Disneyland tomorrow. So it is true, right now, that I have the ability to visit Disneyland tomorrow. The same thing should be true for Smith when it comes to stopping Jones. At the beginning of the story, Smith can depart in her time machine at t+. Moreover, if she departs in her time machine at t+, she will then be in a position to stop Jones from sitting at t. So, at the beginning of the story, Smith is able to stop Jones from sitting at t. But this result directly contradicts FTP, since Jones’ sitting at t is already a part of the temporal past at that point. For this reason, I conclude that FTP is not only unmotivated, incomplete, and incorrect, but ultimately unsalvageable. 20 As Hud Hudson pointed out to me, one could insist that the relevant fact about Jones is soft, so that the restricted version of FTP would have no application. However, it is difficult to come up with a definition of a soft fact would deliver this result. 8 3. The Argument Reconsidered I have argued that the Fixity of the Past principle is ambiguous, and that the traditional understanding of this view is mistaken. I will now argue that these conclusions have important implications for the debate over theological fatalism. To begin, recall the original argument about Jones and his sitting. If it is true that Jones is going to sit at t then this proposition was already true in the past (at time t–). Moreover, given God’s eternality and omniscience, he both existed at t– and believed, at that time, that Jones would sit at t. However, given his essential nature, God would not have had this belief, if Jones was going to avoid sitting at t. (After all, if Jones was to avoid sitting, then the relevant proposition would have been false, which would have required God to reject it.) So, given FP, it follows that Jones cannot avoid sitting at t. If we let F be the fact that God believed at t– that Jones was going to sit at t, then we can formalize this argument as follows: 1. F is a fact about the past, relative to t. 2. If Jones were to avoid sitting at t, then F would not have been a fact. 3. For any action Y, agent S, and time t, if it is true that if S were to do Y at t, some fact about the past relative to t would not have been a fact, then S cannot at t do Y at t. [So] 4. Jones cannot avoid sitting at t. Since the same form of argument would apply to every human action, it follows that everything we do is unavoidable. The problem with this argument should by now be obvious. If “the past” refers to the temporal past, then the first premise of the argument is correct. God’s belief does come before Jones’ action in the order of external time. However, on this interpretation, the third premise of the argument will be mistaken, since it will be equivalent to FTP—a principle that we have just shown to be false. On the other hand, if we take “the past” to refer to the causal past, then the third premise will be equivalent to FCP—a principle that has so far proven plausible. However, if we understand “the past” in terms of causation, then the fatalist’s argument will go wrong at the very first step. There might be some unusual cases in which God’s foreknowledge ends up causing the very acts that he foreknows, but those would be exceptions to the general rule. In the present case, there’s no reason to think that God’s belief is a cause of Jones’ sitting, so there’s no reason to think that it is a part of his causal past, relative to t. Thus, on the first interpretation of the argument, the third premise 9 is false, and, on the second interpretation, the first premise is false. Either way, the argument is unsound. Since this response hinges on the claim that our actions are causally independent of God’s foreknowledge, I will refer to it as the independence solution to the problem of theological fatalism. The independence solution can be helpfully contrasted with the “dependence” response, which is also inspired by the earlier quote from Origen: [I]t will not be because God knows that an event will occur that it happens; but, because something is going to take place it is known by God before it happens. (c.246/2002, p.90, italics mine) According to the dependence response, the key part of this claim is that God’s foreknowledge depends, in some way, on future events. In particular, God’s past belief that Jones would sit at t depends on Jones’ future sitting. To this, the dependency theorist adds the plausible-sounding claim that what depends upon our actions should not be held fixed when determining what we can do. Here is one way of formalizing this idea: Openness of the Dependent (OD): For any fact F, action Y, agent S, and time t, if F’s being a fact depends on S’s doing Y at t, then the fact that F would not have been a fact, had S not done Y at t, is irrelevant to the question of whether or not S can at t do Y at t.21 To illustrate, suppose that I strike a match at t and the match lights a split-second later (at t+). Suppose further that the striking causes the lighting, and that the lighting would not have occurred in the absence of the striking (so that the lighting depends, causally speaking, on the striking of the match).22 In that case OD tells us that the fact that the match lit at t+ is not relevant to the question of whether or not I could have avoided striking the match at t. To put it another way, my avoiding the striking does not need to be compatible with the fact that the match lit in order for it to be the case that I could have done otherwise. (After all, that fact depends on what I choose to do.23) To take another case, recall the earlier example of the 21 This way of putting the principle is suggested by Cyr and Law (2020, p.146). See also Wasserman (Forthcoming). 22 I assume that it is possible for there to be overdetermining causation, so causation is not, on its own, sufficient for counterfactual dependence. In general, I take causal dependence to be counterfactual dependence that is explained by the causal facts. See Wasserman (2018, section 5). 23 The same point applies when the effect is in the past. If Smith travels back from t+ to t, then it is a fact about the past (relative to t+) that Smith appears at that time. But we should not 10 future-tensed proposition that Jones will sit at t. This proposition was true in the past (relative to t), but the fact that it was true depended on Jones’ future action. After all, that fact about the past would not have been a fact if Jones was not going to sit at t, and his sitting at t is what makes the relevant proposition true.24 So, the general lesson seems to be that dependent facts must be left open when determining what an agent can (or cannot) do. This, however, directly contradicts the Fixity Principle if some of God’s past beliefs depend on our future actions. After all, FP says that all of the facts about the past—including those about God’s beliefs—must be held fixed when determining what an agent can do. The dependency theorist rejects this claim and thus concludes that FP—at least in its temporal form—is mistaken. Thus stated, the dependency response appears to be independent of the independence solution. OD, for example, says that the causal effects of our actions (and the other things that depend on our actions) should not be held fixed when assessing our abilities, whereas FCP says that the causal precursors of our actions should be held fixed. These are different claims about different things. But both principles will agree about the case of divine foreknowledge, since God’s past beliefs depend (in some way) on our actions without our actions depending (causally) on his beliefs. Thus, the opponent of theological fatalism may conclude that both replies are successful, and that FTP is doubly doomed. However, there are at least three reasons to prefer the independence solution to the dependency response. The first is dialectical. As it stands, the dependency response is importantly ambiguous since there are many different kinds of dependency. On some readings, the central principle—OD—will turn out to be mistaken. For example, if causal determinism is true, then the past or the laws will depend, counterfactually, on each action that we perform (since the closest world in which we act differently is one in which the past or the laws would have been different). But that does not mean that all of the facts about the past and the laws are irrelevant to what we can do. Hence, counterfactual dependence is not enough for the dependency response.25 However, as soon as we move to more substantial forms of dependence, it becomes less clear whether God’s past beliefs depend on our actions in the relevant sense. For example, knowledge is constituted, in part, by truth, and the truth of the relevant proposition (that Jones will sit at t) depends on Jones’ sitting hold that fact fixed when considering whether or not Smith could have done otherwise (and not traveled back to t) since this past fact depends on Smith’s future action. 24 In this case, the “making” is presumably metaphysical, rather than causal. 25 For further criticisms of the counterfactual approach, see Wasserman (Forthcoming, section 2). 11 at t. Hence, God’s knowledge that Jones will sit at t depends in some “metaphysical” sense on Jones’ future action. But the same thing is not true for God’s belief that Jones will sit at t, since belief is not constituted—even in part—by truth.26 For this reason, God’s past beliefs do not depend, in this sense, on our future actions. Thus, the challenge is for the advocate of the dependence response to provide a sense of dependency on which (i) OD is true, and (ii) God’s beliefs depend on our actions in that sense. I believe that this challenge can be met,27 but the important point for our purposes is that the problem does not even arise for the independence solution. After all, the independence theorist has already specified that the relevant dependence is causal, and our actions do not typically depend on God’s beliefs in this sense. Indeed, if we take the first half of Origen’s thesis as our starting point, then the dialectical situation between the fatalist and her opponent is, in effect, reversed. If it is the dependence of our actions on God’s beliefs that would undermine freedom, then the theological fatalist must identify a form of dependence such that (i) our actions depend on God’s belief in that sense and (ii) that kind of dependence is relevant to what we can or cannot do. On the face of it, this challenge is at least as difficult as the one facing the dependency theorist. A second reason to prefer the independence solution—due to Andrew Law—is that it avoids one of the most prominent objections facing the dependency response. This objection—from Todd (2013) and Fischer and Todd (2013)—goes as follows. Suppose that Jones has made a promise that he will not sit down at t, but that God has foreseen that this promise will be broken. Rather than waiting until t, God decides to “pre-punish” Jones by arranging for him to spend ten hours in the local prison. (We can imagine that this is done surreptitiously, so that Jones remains unaware of God’s involvement.) Upon being released, Jones goes on to break his promise and sit down at t in exactly the way that God foresaw. In this case, Jones’ prepunishment will “depend” on his future sitting in the same way that God’s past belief depends on that sitting. So, (OD) will tell us that we should not hold this punishment fixed when determining whether or not Jones can avoid sitting at t. Indeed, since the dependency theorist will want to treat these cases as analogous, they will have to say that Jones can avoid sitting at t. But Todd and Fischer take this to be the wrong result: Jones cannot avoid sitting at t if God has already pre-punished him for that act. After all, God is by nature morally perfect and could not, therefore, unjustly punish someone for a crime that they will not commit. 26 27 For further development of this worry, see Wasserman (Forthcoming, section 3). See Wasserman (Forthcoming, section 5). 12 It is unclear how much dialectical force this objection will have against the dependency theorist. But the important point is that the independence solution completely avoids the issue. After all, Jones’ earlier imprisonment will obviously be a part of his causal history when he later decides to sit down at t. FCP will therefore tell us that that fact must be held fixed when deciding whether or not he could have done otherwise. Indeed, as Law (Forthcoming a) points out, our intuitions about this case are strongest if we suppose that Jones’ imprisonment plays a role in his later decision to sit down at t (perhaps it puts Jones in a sour mood and he decides to break his promise partly out of spite).28 In that case, the imprisonment will not just be a part of Jones’ personal history—it will also play a direct role in the relevant behavior.29 So, given the truth of a principle like FCP, it will follow that Jones cannot avoid sitting down at t. In this way, the independence theorist can draw a plausible distinction between the case of divine foreknowledge and the case of divine pre-punishment. A third and final reason to prefer the independence solution is that the dependence response delivers incorrect results in certain cases of time travel. Indeed, these examples help to bring out an important difference between OD and FCP. Consider, for example, the original case of Smith and Grandfather. As noted earlier, Grandfather’s not being killed as a baby is a part of Smith’s causal past, since her existence depends (causally) on his meeting Grandmother. Thus, FCP tells us to keep this fact fixed when determining what Smith can do at that time. However, Grandfather’s not being killed at that time also depends (causally) on what Smith does in the past. After all, if Smith had fired a fatal bullet just a moment earlier, then Grandfather would have been killed at t–. So, Grandfather’s not being killed depends (causally) on Smith not firing the bullet. Hence, OD tells us to keep the fact that Grandfather was not killed open when determining what Smith can do. But that again seems like the wrong result, since it precludes the natural explanation of Smith’s inability to kill Grandfather. The same point can be made with a less homicidal example. Suppose that, instead of trying to kill her grandfather, Smith simply leaves him the plans for her time machine (at t–). Suppose further that Grandfather eventually gives the plans to Father, who passes them onto Smith (at t), who uses them to build her time machine (so that she can go back in time and leave the plans with Grandfather). In 28 See also the discussion in Law (Forthcoming b). Although we approached this topic independently, Law and I come to very similar conclusions. 29 The distinction between an agent’s causal history and the causal history of her actions turns out to be important when it comes to certain unusual cases of time travel. See Wasserman (Unpublished, section 3). 13 this case, Smith’s receiving the plans at t is both causally dependent on, and is itself a cause of, her leaving the plans at t–. Hence, OD tells us to keep the former fact open (when determining what Smith can do at t–), whereas FCP will tell us to keep it fixed (since it is a part of Smith’s causal history at that point). As before, FCP seems to get things right. After all, Smith would never have been there in the first place (just before t–) if she had not had the time machine, and she wouldn’t have had the time machine if Grandfather hadn’t received the plans at t–. So, she cannot (then) avoid giving him the plans. Of course, Smith could try to avoid this (by, for example, trying to light the plans on fire). But we know in advance that she will fail—Smith will either slip on a banana peel, or drop her match, or have a change of heart, etc. One way or another, Grandfather will assuredly receive the plans, which means that Smith will invariably fail. But, if an agent would invariably fail to perform an action no matter what she tried to do (and no matter how many times she tried), then she cannot perform that act.30 It follows that Smith cannot avoid giving the plans to Grandfather. Once again, the natural explanation for this is that Grandfather’s receiving the plans is already a part of Smith’s causal history, and that facts about one’s causal history must be held fixed when determining what one can do. Since this explanation contradicts OD, we have yet another reason to prefer the independence approach. I conclude that there is much to say in favor of the proposed solution. Unlike traditional responses to fatalism, it does not require us to take a stand on controversial claims about the nature of eternality or omniscience or truth.31 Moreover, unlike the currently-popular dependence response, it does not require us to identify a sense in which God’s past beliefs depend on our future actions (though it is of course consistent with there being such dependence). Perhaps most importantly, the independence solution follows from an independently motivated distinction—the distinction between the causal and temporal past—and it provides us with a uniform solution to both the problem of theological fatalism and the problem of logical fatalism. All of this speaks in favor of the proposed account.32 30 See Vihvelin (1996, p.318) and the discussion in Wasserman (2018, pp.114-130). Here I have in mind the kind of responses suggested by Stump and Kretzmann (1991), Swinburne (2016), and Geach (1977). For an overview of the traditional responses, see Zagzebski (2017). 32 For comments on earlier drafts of this paper, I thank Dan Howard-Snyder, Frances HowardSnyder, Hud Hudson, Andrew Law, Christian Lee, Neal Tognazzini, Dennis Whitcomb, and an anonymous reviewer for this journal. 31 14 References Byerly, T. R. (2015). Do God’s beliefs depend on the future? On the importance of the mechanics of divine foreknowledge. Journal of Analytic Theology, 3, 124129. doi: 10.12978/jat.2015-3.19-5113012424a Carrol, J. (2016). Ways to commit autoinfanticide. Journal of the American Philosophical Association, 2, 80-91. doi: 10.1017/apa.2015.37 Craig, W. L. (1987). The Only Wise God. Grand Rapids, MI: Baker Book House. Cyr, T. W. and Law, A. (2020). Freedom, foreknowledge, and dependence: a dialectical intervention. American Philosophical Quarterly, 57, 145-154. doi:10.2307/48570844 Fischer, J. M. (1994). The Metaphysics of Free Will. Oxford: Blackwell. Fischer, J.M. and Todd, P. (2013). The truth about foreknowledge. Faith and Philosophy, 30, 286-301. doi: 10.5840/faithphil201330329 Fischer, J. M. and Tognazzini, N. (2014). Omniscience, freedom, and dependence. Philosophy and Phenomenological Research, 88, 346-367. Geach, P. (1977). Providence and Evil. Cambridge: Cambridge University Press. Hasker, W. (1989). God, Time, and Knowledge. Ithaca: Cornell University Press. Hoffman, J. (1979). Pike on possible worlds, divine foreknowledge, and human freedom. Philosophical Review, 88, 433-442. doi: 10.2307/2184958 Law, A. (Forthcoming a). The dependence response and explanatory loops. Faith and Philosophy. Law, A. (Forthcoming b). From the fixity of the past to the fixity of the independent. Philosophical Studies. Lewis, D. (1976). The paradoxes of time travel. American Philosophical Quarterly, 13, 145-152. McCall, S. (2011). The supervenience of truth: freewill and omniscience. Analysis, 71, 501-506. doi: 10.1093/analys/anr051 Merricks, T. (2011). Foreknowledge and freedom. Philosophical Review, 120, 567-586. doi: 10.1215/00318108-1334487 Murray, M. J. and Rea, M. (2008). An Introduction to the Philosophy of Religion. Cambridge: Cambridge University Press. 15 Origen. (c.246/2002). Commentary on the Epistle to the Romans: Books 6-10. T.P., Scheck (Trans.), Washington D.C.: The Catholic University of America Press. Pike, N. (1965). Divine omniscience and voluntary action. Philosophical Review, 74, 27-46. doi: 10.2307/2183529 Rea, M. (2015). Time travelers are not free. Journal of Philosophy, 112, 266-279. doi: 10.5840/jphil2015112516 Rota, M. (2010). The eternity solution to the problem of human freedom and divine foreknowledge. European Journal for Philosophy of Religion, 2, 165–186. doi: 10.24204/ejpr.v2i1.359 Stump, E. and Kretzmann, N. (1991). Prophecy, past truth, and eternity. Philosophical Perspectives, 5, 395–424. doi: 10.2307/2214103 Swenson, P. (2016). Ability, foreknowledge, and explanatory dependence. Australasian Journal of Philosophy, 94, 658-671. doi: 10.1080/00048402.2015.1130731 Swenson, P. (2017). Fischer on foreknowledge and explanatory dependence. European Journal for Philosophy of Religion, 9, 51-62. doi: 10.24204/ejpr.v9i4.2034 Swinburne, R. (2016). The Coherence of Theism, 2nd edition. Oxford: Clarendon Press. Todd, P. (2013). Prepunishment and explanatory dependence: a new argument for incompatibilism about foreknowledge and freedom. Philosophical Review, 122, 619-639. doi: 10.1215/00318108-2315315 Todd, P. and Fischer, J. M. (2015). Introduction. In P. Todd and J.M. Fischer (Eds.), Freedom, Fatalism, and Foreknowledge. Oxford: Oxford University Press. van Riel, R. (2017). Free will, foreknowledge, and future‐dependent beliefs. The Southern Journal of Philosophy, 55, 500-520. doi: 10.1111/sjp.12259 Vihvelin, K. (1996). What time travelers cannot do. Philosophical Studies, 81, 315330. doi: 10.1007/BF00372789 Wasserman, R. (2018). Paradoxes of Time Travel. Oxford: Oxford University Press. Wasserman, R. (Forthcoming). Freedom, foreknowledge, and dependence. Noûs. doi: 10.1111/nous.12317 Wasserman, R. (Unpublished). Time travel, incompatibilism, and the fixity of the past. 16 Westphal, J. (2011). The compatibility of divine foreknowledge and freewill. Analysis, 71, 246-252. doi: 10.1093/analys/anr023 Zagzebski, L. (1991). The Dilemma of Freedom and Foreknowledge. Oxford: Oxford University Press. Zagzebski, L. (2017). Foreknowledge and free will. In E. Zalta (Ed.), Stanford Encyclopedia of Philosophy. 17