Linda Zagzebski, a distinguished philosopher who recently retired from the University of Oklahoma, has written a book titled Fatalism and the Logic of Time (Oxford, 2024). This book is a culmination of many years of reflection on the challenges posed by fatalistic arguments. Thirty-five years previously she wrote a well-received book titled The Dilemma of Freedom and Foreknowledge (Oxford, 1991) that focused on the challenges fatalistic arguments pose for divine foreknowledge and creaturely freedom. More recently (2021) she co-authored (with David Hunt) the Stanford Encyclopedia of Philosophy entry on “Foreknowledge and Free Will.” Her 2024 book covers some of the same ground, but focuses primarily on the challenges fatalistic arguments pose for thinking about the nature or “logic” of time, particularly the so-called “arrow of time” that distinguishes the past from the future.
In this blog post I comment on Zagzebski’s most recent book. Contrary to Zagzebski, I argue that fatalism has no essential connection with the arrow of time. In consequence, I believe her 2024 book is quite wrong-headed.
Now, Zagzebski is a very competent philosopher and a very clear writer, but with respect to fatalism she seems fixated on the idea that it essentially has to do with the “necessity of the past,” where she construes the latter as a modality that the past has simply qua past and the future lacks simply qua future. Consequently, she insists on construing the necessity of the past in such a way that it lines up with the arrow of time. The arrow of time is that which distinguishes the past (and present) from the future—why is time oriented only toward the future? Throughout her book she explores some different ways of cashing out the necessity of the past and, after examining fatalistic arguments based on those conceptions, argues that none of these proposed ways of distinguishing past from future actually work. She concludes that there is something fundamentally wrong with many fatalistic arguments and probably also with the notion of the arrow of time itself. Here’s one of her summary statements: “My aim is twofold: To significantly narrow the range of possibly successful fatalist arguments, and more importantly, to show how fatalist arguments reveal deep confusions in our views about time” (p. 106).
1. Defining fatalism
In her introduction Zagzebski defines a “fated event” as one that “is necessitated by something in its past, so when the event occurs, there is no other possibility” (p. 1, emphasis added). In Zagzebski’s terms, then, fatalism, or what she also calls “universal fatalism” (p. 6), is the thesis that all events are fated or necessitated in some way by something in their past.
Zagzebski also distinguishes between three “forms” of fatalism:
- logical fatalism = “the position that the truth of propositions about the future entails fatalism” (p. 6)
- theological fatalism = “the position that God’s infallible knowledge of the future entails fatalism” (p. 6)
- causal fatalism = “the position that the state of the world in the distant past in conjunction with the laws of nature entails fatalism” (pp. 6–7).
She notes that each type of fatalism “gets it force from the combination of two common principles” (p. 7):
- the necessity of the past, i.e., the idea that the past is “now-necessary” in some sense that distinguishes it from the future
- a transfer of necessity principle that allows us to transfer the necessity of the past to all future events.
Her general strategy throughout the book is to argue that if the necessity of the past is understood in such a way as to define the arrow of time, then the transfer principle, by virtue of transferring that necessity to all future events, thereby destroys the arrow of time. So, she reasons, something has to give. Either we should reject that understanding of the arrow of time—and perhaps doubt whether there even is an arrow of time—or we should reject the relevant transfer principle. While this doesn’t refute fatalism, it does (she thinks) undermine a wide swath of traditional arguments for fatalism.
Now, before going any further, I want to push back at the way Zagzebski sets things up.
First, despite what she says, fatalism has no essential connection at all to the necessity of the past. Zagzebski should have realized this because in the Appendix of her 1991 book she points out, correctly, that divine timelessness by itself does not sidestep the apparent incompatibility between creaturely freedom and infallible divine foreknowledge. Just replace “foreknowledge” with “knowledge of what to us is future” and the “necessity of the past” with the “necessity of timeless eternity” and run the argument exactly as before. This point alone is enough to undercut the central premise of her book. After all, if you can run a fatalistic argument without relying on the necessity of the past, then it immediately follows that fatalism is not essentially tied to the necessity of the past or, for that matter, to the arrow of time.
Second, it should be obvious that if fatalism is to pose a worry for creaturely freedom or future contingency more broadly, then it cannot be based on a purely temporal modality. It has to be based on a modality that, if not causal per se like causal necessity, nevertheless has causal implications (like logical and metaphysical necessity). In this context, in a 2021 paper that Zagzebski references in her SEP article, Andrew Law argues (correctly) that what matters for fatalism is not the fixity of the past but the “fixity of the independent.” In other words, all that matters is whether a unique and complete future is entailed by facts that are fixed independently of what happens in the future. If you have that, then you can construct a valid argument for fatalism regardless of whether those facts are in the past or not. The anti-fatalist must then say either (a) there is no collection of facts that picks out a unique and complete future (i.e., open futurism), or (b) there is such a collection, but some of those facts are ontologically dependent on what happens in the future, a view I call preventable futurism.
Third, Zagzebski’s three forms of fatalism are not that by her own definition of fatalism. It’s not that arguments for “logical fatalism,” “theological fatalism,” and “causal fatalism” issue in different kinds of fatalism. No, they all lead to fatalism simpliciter. And so, rather than speaking of different forms of fatalism, it would be much more accurate to speak of logical, theological, causal, etc. arguments for fatalism. (To be fair to Zagzebski, confusing various ways of arguing for fatalism with different types of fatalism is nearly ubiquitous in the scholarly literature, but that’s not a good excuse for perpetuating the confusion.)
2. Accidental necessity
The first way in which Zagzebski proposes to construe the necessity of the past is as accidental necessity. She devotes Part 1 (pp. 11–62) of her book to this conception of necessity, which she attributes to Ockham (p. 20). According to this conception, “the past qua past is accidentally necessary; the future qua future is accidentally contingent” (p. 35). This conception entails what Zagzebski calls a “modal arrow of time” because it grounds the distinction between necessary and contingent in the mere passage of time, wholly independent of any other type of modality, whether causal, metaphysical, logical, or what have you.
The problem, of course, is that if we apply the transfer of necessity principle to this sort of necessity, then it follows that the future is accidentally necessary just like the past. “The fatalist,” concludes Zagzebski, “assumes one half of the modal arrow of time in order to argue that there is no arrow of time” (p. 37). Later on she says, “Standard fatalist arguments implicitly assume a modal arrow of time in order to deny it” (p. 52). After arguing that accidental necessity has the absurd consequence that nothing in the past entails anything about the future (p. 42), Zagzebski concludes that this notion of a purely temporal modality is incoherent.
Now, I think she’s right that the notion of accidental necessity is incoherent as she construes it, but (1) she seriously misconstrues accidental necessity and (2) given how she does construe it, she’s wrong in thinking that anything like this modality motivates “standard” fatalist arguments.
Regarding (1), Zagzebski says that Ockham’s idea of accidental necessity derives “from the basic Aristotelian idea of potency and act … Necessity is what is settled by what is now in act. The contingent is what is now in potency” (p. 23). Now, she’s correct to see a connection between accidental necessity and the act/potency distinction, but she’s wrong to associate potency with contingency. The act/potency distinction most fundamentally has to do with change or becoming. In order for something to change it is first actually such-and-such but potentially otherwise and then that potency becomes actualized making the thing different in that respect from how it was before. This contrast between how things actually are but could be otherwise (act) and how things could be but aren’t yet (potency) is a plausible way to understand the arrow of time. But the contrast here is not between necessity and contingency (as Zagzebski supposes) but between actuality and non-actual possibility. While we can think of the actuality of the present and past as now-necessary because they are over and done with and cannot now be changed, it is a serious error on Zagzebski’s part to think of the potency of the future as contingency. There may well be future contingencies, but nothing in the act/potency dynamic requires that anything in the future be contingent. In a deterministic system, for example, future events are in potency relative to the past and present—they are possible but non-actual until they occur—but they are not at all contingent. In short, you can have accidental necessity without future contingency. Consequently, fatalistic arguments that transfer necessity from the past to the future do not (pace Zagzebski) “assume a modal arrow of time in order to deny it.”
Regarding (2), if we assume with Zagzebski that accidental necessity entails future contingency and that it is a purely temporal modality that has nothing whatsoever to do with causation, then the fatalistic arguments she focuses on become nothing but straw men. As I pointed out in section 1 above, for fatalism to pose any threat to creaturely freedom and future contingency it has to based on a modality that has causal implications, which accidental necessity—as Zagzebski articulates it—emphatically does not. Moreover, Zagzebski’s version of accidental necessity is so counterintuitive that no fatalistic argument based on it could ever hope to become “standard.” After all, hardly anyone believes that the future is meaningfully contingent simply qua future because nearly everyone believes that some future events (e.g., the next solar eclipse) are necessary on account of being causally determined by past and present events. That sort of necessitation (i.e., causal necessitation) is obviously relevant for thinking about creaturely freedom, but why anyone should suppose that the mere passage of time, in complete abstraction from causality, poses a threat to our freedom is puzzling, to say the least.
3. The causal closure of the past
This leads to Part 2 of Zagzebski’s book (pp. 63–111), where she (finally!) considers fatalism in relation to causality. This moves her a big step closer to the real target. However, she still wants a type of modality that grounds the arrow of time and thus that distinguishes the past qua past from the future qua future. This temporal fixation keeps her from actually finding the target.
Her proposal, which she calls the causal closure of the past, is that the past qua past lies outside the scope of causation and that, therefore, for all past events E, nothing can now cause E to have happened and nothing can now cause E not to have happened (p. 65). While that seems like a very plausible thesis, Zagzebski points out that if we transfer that sort of necessity to the future, then we get the result that the future is causally closed too. For all future events E, nothing can now cause E to happen or not to happen. And this consequence is very implausible indeed.
Of course, transfer of causal closure is a stronger assumption than any fatalist needs. Indeed, the fatalist doesn’t need to transfer anything from the past qua past to the future qua future. All the fatalist needs is for there to be a collection of facts that specifies a unique future (i.e., a unique and complete extension of the actual past and present) and for that collection to be “fixed” independently of any “say” we might suppose ourselves to have about it. If there’s a transfer principle at work here, it’s the transfer of “fixity” or, in Arthur Prior’s words, now-unpreventability. In terms of Zagzebski’s definition of the causal closure of the past, all the fatalist needs is the part which says that nothing can now cause E not to have happened. If E is now-unpreventable and E entails future F, then F is now-unpreventable too.
To her credit, Zagzebski almost comes around to recognizing that now-unpreventability is where the focus should be. She defines an unpreventability of the past principle, which says that “If E is an event in the past, nothing can now cause not E” (p. 71). And she uses this principle in conjunction with a transfer of unpreventability principle to set forth fatalistic arguments of the sort that “many philosophers take seriously” (p. 71).
Unfortunately, Zagzebski objects to reliance on the unpreventability of the past on the grounds that it most plausibly goes hand-in-hand with the causal closure of the past: “the two principles go together” (p. 77) because the “uncausability of the past and the unpreventability of the past are both part of the same intuition that the past is causally closed” (p. 75). But this is confused. The past, we may plausibly suppose, is causally closed and therefore unpreventable. But fatalism doesn’t require the transfer of causal closure, but only the transfer of unpreventability. Indeed, if some future events (e.g., eclipses) are causally determined then we have cases of future unpreventability without causal closure. The two principles, therefore, do not go hand-in-hand. The fatalist may affirm the causal closure of the past, but he doesn’t need to base his fatalistic arguments on that. Unpreventability suffices.
The reason Zagzebski has a hard time admitting this point is because she’s fixated on tying fatalism to a modality that neatly divides the past qua past from the future qua future. But that’s not what fatalism is about. Fatalism doesn’t essentially care about the arrow of time. It doesn’t start with a past/future split and then project a modality from the past half to the future half that effaces that split. No, it simply starts with the intuition—one we all share—that certain facts (such as but not necessarily limited to past facts) are fixed and therefore unpreventable and then observes that if those fixed facts entail all future facts, then the future is also unpreventable because it is, in effect, included within the collection of fixed facts. There’s no need for a supposition that the future is accidentally contingent, not capable of being caused, or not unpreventable.
4. Does time have an arrow?
In the final chapter of her book, Zagzebski asks whether time even has an arrow. “The existence of entailment relations between past and future,” she says, “shatters three arrows of time: the modal arrow, the causal arrow, and the counterfactual arrow” (p. 112). Since we think that the past does bear upon the future in some sense, we should affirm that these entailment relations obtain and give up, or at least consider how best to modify, our conception of time’s arrow.
Zagzebski goes on to explore some ideas along these lines, but I don’t wish to pursue those. As I see it the whole project is under-motivated. Contrary to Zagzebski’s project, the existence of entailment relations between past and future does not “shatter” the arrow of time. It only shatters her misconceptions about the arrow of time. More specifically, it shatters her assumption that the arrow requires the past to be necessary in some sense and the future to be not necessary (or contingent) in that same sense. But this is wrong. As my discussion of the act/potency distinction in section 2 above makes clear, the arrow of time marks a contrast between actual/necessary and not actual/possible. Non-actuality is opposed to actuality, but possibility is not opposed to either actuality or necessity. And so fatalistic arguments that trade on the necessity of the past—which not all fatalistic arguments do—do not undermine the arrow of time. If successful all they show is that we can have future potency (non-actual possibility) without future contingency.
5. What about open futurism?
Zagzebski says very little about open futurism, and what she does say is rather dismissive. On pages 50–51, in the context of discussing fatalistic arguments based on accidental necessity, she briefly references Patrick Todd’s 2021 work defending the idea that all will and will not propositions about future contingents are false. The reason she rejects open futurism is because she thinks (erroneously) that accidental necessity requires opposing the past as necessary to the future as contingent. And since what necessarily follows from what’s necessary is also necessary (transfer of necessity), this means (to her mind) that no proposition about the past can entail any proposition about the future if we grant her notion of accidental necessity. In short, she rejects open futurism because she sees it as not addressing the fundamental problem, namely, the failure of past-to-future entailment.
As I’ve already explained, however (see section 3 above), Zagzebski’s understanding of accidental necessity is incorrect. Accidental necessity does not oppose the past as necessary to the future as contingent. The arrow of time does not consist in the movement from necessity to non-necessity, but in the movement from actual but potentially otherwise to actually otherwise. The past is necessary because it is over and done with. It’s necessary, i.e., non-preventable, because it’s “too late” for us to do anything about it. The future is not over and done with, but it can also be non-preventable if it is “too late” to avoid it, like the Titanic ten seconds before hitting the iceberg.
Once we set Zagzebski’s confusions about the arrow of time aside, her main objection to open futurism vanishes. There is no problem with past-to-future entailment centered on the arrow of time. Her only other objection is that she finds it implausible that open futurism takes tense seriously to the point of denying that differently “tensed propositions about the same event are the same proposition” (p. 51). Here, though, it is Zagzebski and not open futurists like Todd or myself who defends the more implausible claim. The claim that tensed propositions about the same event are the same proposition expresses what today is known as the “old” B-theory of time. This theory says that language can be completely detensed without loss of meaning. Today this position has been almost completely rejected even by defenders of the B-theory because it has proven impossible to translate all tensed discourse into tenseless discourse without loss of meaning. The old B-theory has thus largely been replaced by the “new” B-theory which accepts that differently tensed propositions about the same event are different propositions while affirming that they have identical tenseless truthmakers. In this respect, then, it seems that Zagzebski has not kept up with developments in the philosophy of time.
6. Concluding remarks
While I have a lot of respect for much of Zagzebski’s work, I cannot recommend this particular book. Its central contention that fatalism is essentially connected to the arrow of time, which conception it then reveals to be problematic, is based on a faulty understanding of both fatalism and the arrow of time.