2.a. Libertarianism

One way to define libertarianism is to first define libertarian free will and then to define libertarianism as the conjunction of the following two theses:

(L1) is a conceptual claim, so it is pretty obviously compatible with God’s foreknowledge—i.e., with the thesis that God has foreknowledgeFootnote ⁵—and we will not concern ourselves with it here. What we want to argue is that (L2) is compatible with God’s foreknowledge. So when we use the term ‘libertarianism’ in this paper, we will be talking about (L2)—i.e., the thesis that human beings possess libertarian free will. But we need to say what libertarian free will is.

Broadly speaking, we can say that a person S has libertarian free will—or for short, L-freedom—if and only if S makes at least some decisions that are L-free in the following sense:

This is not very precise; in order to make it more precise, we need to say more about the four clauses of the definition. We will have a lot to say about clause (i), but let us start by saying a few words about (ii)-(iv).

In connection with clause (ii), different libertarians can define appropriate non-randomness in different ways, but on any reasonable view, appropriate non-randomness will involve a sort of agent-involvedness—e.g., it will involve the agent controlling which option is chosen, or authoring the choice, or some such thing. But the exact definition of appropriate non-randomness will not matter in this paper, so we will not say any more about it.

Moving on to clause (iii), the reason we include this clause in our definition is that if we did not, then the claim that we are L-free would be compatible with the claim that indeterminism is necessarily freedom-undermining—for it would be compatible with the view that there are insignificant indeterminacies in our decision-making processes and that, despite these indeterminacies, our decisions are sufficiently appropriately non-random to count as free.

To make clause (iv) more precise, let us suppose that a person S makes a decision D in which a specific option O is selected. Given this, we can say that D plurally satisfied conditions (ii) and (iii) if and only if there was at least one unchosen option O* such that (a) while S did not end up choosing O*, S could have chosen O*, even holding fixed all of the laws and all of the facts about the past, and (b) if S had chosen O* (instead of O), then D still would have satisfied conditions (ii) and (iii).

But for our purposes in this paper, the most important clause is clause (i)—i.e., the clause that says that L-free decisions are undetermined. In the remainder of this subsection, we will define three different ways in which a decision can be undetermined, and this will give us three different kinds of L-freedom that, we think, differ with respect to whether they are compatible with God’s foreknowledge.

First, let us say that an event E was entailment-determined if and only if the claim that E would occur when it did occur was already entailed by the conjunction of P and L, where P is a complete description of the universe at some time prior to E’s occurrence, and L is a complete statement of the laws of nature. And let us say that E was entailment-undetermined if and only if it was not entailment-determined; and in the case where E is a decision in which a specific option O was chosen, when we say that E was entailment-undetermined, what we’ll mean is this: the event of O being chosen in E was not entailment-determined.

This last definition gives us our first kind of L-freedom—a kind in which the decisions in question are entailment-undetermined. We can call this entailment-L-freedom. (To be clear, entailment-L-freedom is defined exactly as L-freedom was defined above, except that the two occurrences of ‘undetermined’ are replaced by ‘entailment-undetermined’.) Given this, we can say that entailment-libertarianism is the view that human beings are entailment-L-free—i.e., that they make at least some decisions that are entailment-L-free.

Second, let us say that (a) an event E was causally determined if and only if E was causally necessitated by prior events together with the laws of nature (or if and only if prior events caused it to be the case that, at some time prior to E’s occurrence, the objective probability that E would occur was 1); and (b) an event was causally undetermined if and only if it was not causally determined (and, again, if E is a decision in which option O was chosen, then in our lingo, ‘E was causally undetermined’ means that the event of O being chosen in E was causally undetermined). Given this, we can define a second kind of L-freedom—a kind in which the decisions in question are causally undetermined. We will call this weak-causal-L-freedom. And given this, we can say that weak-causal-libertarianism is the view that human beings are weak-causal-L-free.

Notice that weak-causal-libertarianism and entailment-libertarianism are both compatible with the claim that for every human decision, there is a unique option O such that the prior objective probability that O would be chosen was .999 or greater. Given this, it seems that weak-causal-L-freedom and entailment-L-freedom are both less robust than we might like.

To appreciate this point, consider torn decisions—i.e., decisions in which (a) the person in question (viz., S) feels completely torn between multiple tied-for-best options (i.e., it seems to S, in their conscious thought, that these options are all tied-for-best), and (b) S chooses while still feeling torn. To say that a torn decision was entailment-L-free and weak-causal-L-free is compatible with the claim that the prior objective probability that one of the tied-for-best options would be chosen was .999. This is surely not the result that libertarians want, and so it seems that we should try to define a more robust kind of L-freedom.

To get a more robust kind of L-freedom, we need a more robust kind of indeterminacy. To this end, consider the following definition:

This is a bit of a mouthful, but the basic idea is very simple; we can bring this out by thinking of a case in which there are only two options available to S; in a case like this, to say that D was strongly causally undetermined is just to say that the objective moment-of-choice probabilities of the two options being chosen (given all of the causally relevant facts and all of the laws) were both roughly 0.5. And if there were only three options available to S (and if all three options were tied-for-most-likely), then the probabilities were all roughly 0.333. And so on.Footnote ⁸

This definition gives us a third kind of L-freedom—a kind of L-freedom in which the decisions in question are strongly causally undetermined. We can call this strong-causal-L-freedom. And given this, we can say that strong-causal-libertarianism is the view that human beings are strong-causal-L-free.Footnote ⁹

We’ll argue in sections 3-5 that strong-causal-libertarianism is compatible with God’s foreknowledge (and notice that it follows from this that weak-causal-libertarianism is compatible with God’s foreknowledge as well).Footnote ¹⁰ Now, we admit that entailment-libertarianism is not compatible with God’s foreknowledge; but we’ll argue in section 6 that (a) strong-causal-libertarianism is the kind of libertarianism that we should be focusing on in connection with the problem of God’s foreknowledge, and (b) we should not care at all about entailment-libertarianism.

2.b. Eternalism

Eternalism is best understood in contrast with presentism. Presentism (see, e.g., Zimmerman Reference Zimmerman, van Inwagen and Zimmerman1998, Markosian Reference Markosian and Zimmerman2004, Crisp Reference Crisp, Loux and Zimmerman2005, and Merricks Reference Merricks2007) is the view that only present objects exist; and eternalism (see, e.g., Heller Reference Heller1984, Lewis Reference Lewis1986, Sider Reference Sider2001, and Hawley Reference Hawley2001) is the view that past, present, and future objects all exist. So, e.g., eternalists believe that there are dinosaurs and 22nd-century cockroaches, and presentists do not.

We can also think of eternalism as the view that physical reality is a 4-dimensional spatiotemporal block and temporally distant objects like dinosaurs are analogous to spatially distant objects like Saturn; so according to eternalism, just as Saturn exists in the outer solar system, dinosaurs exist in the Jurassic period, where the outer solar system and the Jurassic period are both just different regions of the 4-dimensional spatiotemporal manifold.Footnote ¹¹

2.c. Libertarianism and Eternalism Together

We are going to argue in section 3 that strong-causal-libertarianism is compatible with eternalism. It is worth noting, however, that it is easy to define a kind of libertarianism that is not compatible with eternalism. We can do this by defining a fourth kind of indeterminism. In particular, let us say that (a) an event E was settlement-determined if and only if it was already settled in advance that E would occur, in an eternalistic sense of ‘settled’ (so if eternalism is true, then all of our decisions are settlement-determined); and (b) an event E was settlement-undetermined if and only if E was not settlement-determined. Given this, we can define a fourth kind of L-freedom—a kind in which the decisions in question are settlement-undetermined. We can call this kind of L-freedom unsettled-L-freedom, and given this, we can say that unsettled-libertarianism is the view that human beings are unsettled-L-free.

Unsettled-libertarianism is obviously not compatible with eternalism. But we’ll argue in section 6 that (a) we should not care at all about unsettled-libertarianism, and (b) the kind of libertarianism that we should care about—and that we should be focused on in connection with the problem of God’s foreknowledge—is strong-causal-libertarianism. And, again, we are going to argue in section 3 that strong-causal-libertarianism is compatible with eternalism.

3.a. The Argument from the Silence of Eternalism About Causation And Agent-Involvedness

Let causal determinism be the view that all events are causally determined in the sense defined in section 2.a, and let causal indeterminism be the view that causal determinism is false. Given these definitions, we begin our first argument for the compatibility of eternalism and strong-causal-libertarianism by arguing for the following thesis:

While we believe that there is a straightforward and obvious argument for (A), this argument is curiously absent from the literature. Shanks (Reference Shanks1994) comes the closest to arguing for something like (A). He argues that eternalism is compatible with indeterminism, but he seems to think that eternalism entails an anti-realist or deflationist view of causation, and he seems to think of indeterminism as just a claim about the laws being probabilistic (and not about the causal structure of reality). So, when Shanks says that eternalism is compatible with indeterminism, he is saying something different from what (A) says. Also, while Hunt (Reference Hunt1999) argued that eternalism is compatible with freedom, he was not talking about an indeterministic kind of freedom, so he did not argue for (A) either.Footnote ¹²

So, our straightforward and obvious argument for (A) is also novel. Here is our argument in a nutshell. If eternalism is true, then all future events already exist. To speak metaphorically, these events are all sitting there, lined up in a row. But eternalism does not tell us anything about the causal relations that obtain between the events that are, so to speak, standing next to each other in line. Indeed, eternalism is perfectly compatible with the view that there are no substantive causal relations at all between these events. Eternalism is also compatible with causal determinism—i.e., with the claim that each event that stands in line is causally necessitated by prior events. And, finally, eternalism is compatible with causal-indeterminism and, in particular, with the view that while there are causal relations between some of the events that are “standing in line,” there are not always deterministic causal relationships between those events.

In short, eternalism—the thesis that all future events are, so to speak, already there, lined up in a 4^th dimension—does not tell us anything about the causal relations that obtain between these events. And so it does not tell us that causal determinism is true. And so eternalism is perfectly compatible with causal indeterminism.Footnote ¹³

More specifically, since eternalism does not say anything about the causal relations that obtain between any of the events that are, so to speak, “standing in line,” it does not tell us anything about the causal histories of our decisions. So eternalism is perfectly compatible with the claim that at least some of our decisions are causally undetermined—and, indeed, strongly causally undetermined. And so claim (A) is true.

But claim (A) is weaker than what we need; we need to argue not just for (A) but for the following claim:

We have already argued that eternalism is compatible with the existence of decisions that satisfy condition (i), so we just need to beef up the argument to show that eternalism is compatible with the existence of decisions that satisfy conditions (ii)-(iv) as well.

Showing that eternalism is compatible with (i)-(iii) is easy—we can argue for this in the same way that we argued for the compatibility of eternalism and condition (i). First, we need to show that the claim that a decision D satisfied conditions (i)-(iii) is a claim about the relations obtained between D and things that existed prior to D. (Different libertarians will argue for this in different ways because they will have different views of what appropriate non-randomness consists in. But on any reasonable view, the claim that a decision D was appropriately non-random will boil down to some version of the idea that D was made by the agent—or that it was authored and controlled by the agent, or some such thing—and so it will be a claim about the relations that obtain between D and other things that exist prior to D (or if eternalism is true, other things that are “standing before D in line”), most notably, the agent, or the agent’s reasons, or events involving the agent’s reasons, or some such thing.Footnote ¹⁴) Second, we just need to point out that eternalism does not say or imply anything at all about the relations that obtain between the things that exist at a time t and the things that exist at later times. It says that these things are all there—standing in line, so to speak—and that is all it says.

Finally, condition (iv) is a counterfactual; in essence, it says that if the agent had chosen differently, then conditions (ii) and (iii) would still have been satisfied. But that is just to say that if the agent had chosen differently, the relevant relations would still have held between D and the relevant things that existed prior to D. But, once again, eternalism does not say anything about these relations. And so it seems to us that eternalism is perfectly compatible with the existence of decisions that satisfy conditions (i)-(iv). And so it seems that claim (B) is true—i.e., that eternalism is compatible with strong-causal-libertarianism.

3.b. The Argument From the Parallel Between the Past and the Future

Our second argument for the compatibility of eternalism and strong-causal-libertarianism proceeds as follows:

1. (I) The claim that a past decision (e.g., Hitler’s decision to invade Poland in 1939) is already settled is compatible with the claim that that decision was strong-causal-L-free. But

2. (II) If eternalism is true, then if (I) is true, the claim that a future decision (e.g., Trump’s decision to order Bubble Gum Surprise at an ice cream parlor in 2028) is already settled is compatible with the claim that that decision will be strong-causal-L-free. Therefore,

3. (III) If eternalism is true, then the claim that a future decision (e.g., Trump’s decision) is already settled is compatible with the claim that that decision will be strong-causal-L-free.

This argument is valid, so we just need to argue for premises (I) and (II). We take it that (I) is obvious, but we can argue for it as follows. The claim that Hitler’s decision was strong-causal-L-free is a claim about the relationship between Hitler’s decision and things that existed before it. And the claim that it is already settled that Hitler decided to invade Poland is simply irrelevant to this. In saying that Hitler’s decision is settled, we just mean that the decision did occur. (Note that ‘settled’ does not mean real here; for then the claim that the past is settled would entail that presentism is false, and we do not mean to imply that by the claim that the past is settled.) In any event, the point here is that the claim that it is already settled that Hitler decided to invade Poland does not entail that Hitler’s decision was not strong-causal-L-free. Indeed, the claim that it is already settled that Hitler decided to invade Poland is not relevant at all to the claim that his decision was strong-causal-L-free. And so these two claims are compatible with one another, and so premise (I) is true.

The argument for premise (II), on the other hand, is based on the fact that if eternalism is true, then there is no relevant difference here between past decisions like Hitler’s and future decisions like Trump’s. All the things that motivate the idea that the claim that Hitler’s decision could be both settled and strong-causal-L-free are, if eternalism is true, present in connection with Trump’s decision as well as Hitler’s. In particular, if eternalism is true, then the claim that it is already settled that Trump is going to order Bubble Gum Surprise is not relevant to the claim that Trump’s decision will be strong-causal-L-free because (a) the claim that Trump’s decision will be strong-causal-L-free is a claim about the relations that obtain between Trump’s decision and things that exist before it, and (b) the claim that it is already settled that Trump will order Bubble Gum Surprise is, if eternalism is true, not relevant at all to the claim that these relations obtain. So if eternalism is true, then there is an exact parallel here between Hitler’s decision and Trump’s decision. In short, if the claim that Hitler’s decision is already settled does not undermine the claim that his decision was strong-causal-L-free—and it clearly does not—then (if eternalism is true) the claim that Trump’s decision is already settled does not undermine the claim that his decision will be strong-causal-L-free. And so premise (II) is true.

The idea that the past and future are deeply similar can seem counterintuitive to us—because, prima facie, it seems that while the past is already settled, the future is not. But if eternalism is true, then this is just a fact about our epistemic access to the past and future. If eternalism is true, then even if it is not causally determined what will happen in the future, it is already settled what will happen—in a way that is exactly analogous to the way that it is already settled what happened in the past.Footnote ¹⁵

5.a. The Argument from Omniscience About Existent Things

If strong-causal-libertarianism is true, then events of a certain kind—namely, strong-causal-L-free decisions—exist. And if eternalism is also true, then future strong-causal-L-free decisions could exist. But we just saw in section 4 that if future events exist, then an omniscient God would have infallible knowledge of them. So if eternalism and strong-causal-libertarianism are both true, then an omniscient God would have infallible knowledge of all future strong-causal-L-free decisions. So (a) strong-causal-libertarianism is compatible with God’s foreknowledge, and (b) eternalism straightforwardly entails that an omniscient God would have infallible foreknowledge of all of our future strong-causal-L-free decisions.

Moreover, for all the reasons brought out in section 4, eternalism comes close to explaining how an omniscient God would have knowledge of our future strong-causal-L-free decisions. More precisely, eternalism eliminates the troublingness of the question ‘How could God have knowledge of our future strong-causal-L-free decisions?’; for if eternalism is true, then this question becomes a special case of (Q3)—i.e., of ‘How could God have knowledge of facts that exist in reality?’—which, again, is clearly not troubling.

It is worth noting that in this scenario, an omniscient God would not just know which options we are going to choose in our future strong-causal-L-free decisions. Such a God would also know that those decisions will be strong-causal-L-free. For God would know that those decisions will satisfy the four conditions for strong-causal-L-freedom—e.g., that they will be strongly causally undetermined, appropriately non-random, and so on.

5.b. The Argument from the Parallel Between Past-Knowledge and Foreknowledge

Our second argument for the compatibility of strong-causal-libertarianism and God’s foreknowledge proceeds as follows:

• (I*) The claim that God has infallible knowledge of the outcome of a past decision (e.g., Hitler’s decision to invade Poland in 1939) is compatible with the claim that that decision was strong-causal-L-free. But.

• (II*) If eternalism is true, then if (I*) is true, then the claim that God has infallible knowledge of the outcome of a future decision (e.g., Trump’s decision to order Bubble Gum Surprise in 2028) is compatible with the claim that that decision will be strong-causal-L-free. Therefore,

• (III*) If eternalism is true, then the claim that God has infallible knowledge of the outcome of a future decision (e.g., Trump’s decision) is compatible with the claim that that decision will be strong-causal-L-free.

This argument is valid, so we just need to argue for premises (I*) and (II*). The argument for (I*) is exactly parallel to the argument for premise (I) from section 3.b. The claim that Hitler’s decision was strong-causal-L-free is a claim about the relations that obtain between Hitler’s decision and things (i.e., objects and events) that existed before it. And the claim that God has infallible knowledge of what Hitler did is not relevant at all to the claim that these relations actually obtained. Therefore, the two claims are compatible with one another, and so premise (I*) is true.

Moving on to (II*), the argument for this is based on the fact that if eternalism is true, then there is no relevant difference here between past decisions like Hitler’s and future decisions like Trump’s. All the things that motivate the idea that God’s foreknowledge is compatible with the claim that Hitler’s decision was strong-causal-L-free are, if eternalism is true, present in connection with Trump’s decision as well as Hitler’s. In particular, if eternalism is true, then the claim that God has infallible knowledge of what Trump is going to do is not relevant to the claim that Trump’s decision will be strong-causal-L-free because (a) the claim that Trump’s decision will be strong-causal-L-free is a claim about the relations that obtain between Trump’s decision and things that existed before it, and (b) if eternalism is true, then the claim that God has infallible knowledge of what Trump is going to do is not relevant in any way at all to the claim that these relations actually obtain. So if eternalism is true, then there is an exact parallel here between Hitler’s decision and Trump’s decision. So if the claim that God has infallible knowledge of the outcome of Hitler’s decision does not undermine the claim that that decision was strong-causal-L-free—and it surely does not—then (assuming eternalism) the claim that God has infallible knowledge of the outcome of Trump’s decision does not undermine the claim that that decision will be strong-causal-L-free. And so premise (II*) is true.

6.a. Why We Should not Care about Entailment-L-Freedom

Recall from section 2.a that entailment-L-freedom is just like strong-causal-L-freedom except that it requires the decisions in question to be entailment-undetermined rather than strong-causal-undetermined.

It seems clear that if God has foreknowledge of our decisions, then we are not entailment-L-free—i.e., none of our decisions is entailment-L-free—because if God knows at t₁ that S will choose O at t₂, then (assuming that God is part of the universe at t₁) a complete description of the universe at t₁ entails that S will choose O at t₂.Footnote ¹⁹

But we will argue in this subsection that if eternalism is true and an omniscient God exists, then we should not care about this. To make our argument here more precise, consider the following scenario (which, we have already argued, is at least possible):

We will argue in this subsection that in this scenario, entailment-L-freedom is not worth wanting, or worth caring about.

Let’s start by thinking about what we want out of a libertarian kind of free will. It seems to us that what we want is for our decisions to be up to us; and we want it to be the case that we are not made to choose in the ways that we do. If all of our decisions are causally determined by the Big Bang and the laws of physics, then there is a clear sense in which the Big Bang and the laws of physics make us choose in the ways that we do. And from a libertarian point of view, that is bad. We want it to be the case that (a) we make our decisions and (b) nothing makes us decide in the ways that we do.

The way that we defined strong-causal-L-freedom in section 2.a reflects these ideas about what we want out of a libertarian kind of free will. The requirement for non-randomness (or agent-involvedness) corresponds to the idea that we want it to be the case that we make our decisions; and the requirement for strong-causal-indeterminacy corresponds to the idea that we want it to be the case that nothing makes us decide in the ways that we do.

Let us focus on the nothing-makes-us-do-it requirement—i.e., on the idea that in order for a choice to be free in a libertarian way, it needs to be the case that nothing makes the agent choose in the way that they do. Focusing on this requirement brings out what is wrong with the idea that we should care about entailment-L-freedom. If the eternalistic-libertarian-foreknowledge scenario is true, then, as we have seen, we are not entailment-L-free—despite the fact that we are strong-causal-L-free. For instance, Trump’s decision to order Bubble Gum Surprise will be strong-causal-L-free in this scenario, but it will not be entailment-L-free because at an earlier time, God already knows that Trump will decide to order Bubble Gum Surprise, and this entails that Trump will make such a decision. But—and this is the really important point—the reason that Trump’s decision will not be entailment-L-free is not that Trump will be made to choose Bubble Gum Surprise. Rather, in the eternalistic-libertarian-foreknowledge scenario, Trump’s decision will not be entailment-L-free for the exact opposite reason; his decision will not be entailment-L-free because Trump’s choosing Bubble Gum Surprise in 2028 makes it the case that God already knew in 2010 that Trump will choose Bubble Gum Surprise in 2028. Footnote ²¹

We can bring this point out if we forget about free will and God’s foreknowledge for a moment and just think about eternalism and truthmaking. If eternalism is true, then in 2010, the sentence ‘Trump will win the 2016 U.S. presidential election’ was already true. And it was made true by Trump’s winning of the election in 2016. So if eternalism is true, then there’s backwards truthmaking. And since truthmaking is not a causal relation, there is nothing metaphysically weird about this.Footnote ²² It is just a straightforward consequence of eternalism. (If eternalism is true, then backwards truthmaking is no weirder than east-to-west truthmaking. If you say in Boston that Madrid exists, then the existence of Madrid makes your claim true in an east-to-west way. Likewise, if eternalism is true, then if you say in 2010 that Trump will win in 2016, then Trump’s winning makes your claim true in a 2016-to-2010 way. If eternalism is true, then these two kinds of truthmaking are exactly analogous to one another.)

If we bring free will into the picture, nothing changes. If eternalism is true (and if Trump’s decision to order Bubble Gum Surprise will be strong-causal-L-free), then the sentence ‘Trump will make a strong-causal-L-free decision to order Bubble Gum Surprise in 2028’ was already true in 2010; moreover, it was made true by Trump’s making a strong-causal-L-free decision to order Bubble Gum Surprise in 2028. So, again, the truthmaking in this scenario is backward truthmaking.

Finally, if we bring God’s foreknowledge back in, we get the same result. If the eternalistic-libertarian-foreknowledge scenario is true—i.e., if eternalism is true, and an omniscient God exists, and Trump makes a strong-causal-L-free decision to order Bubble Gum Surprise in 2028—then the following claim is true:

In this scenario, the following claim is false:

This seems entirely obvious to us. In the eternalistic-libertarian-foreknowledge scenario, God knows in advance that Trump is going to make a strong-causal-L-free decision to order Bubble Gum Surprise. But God does not make Trump choose in that way. Indeed, the claim that God does not make Trump choose in this way follows from the supposition that Trump’s decision will be strong-causal-L-free—because it follows from this that Trump’s decision will be strongly causally undetermined.

So, again, in the eternalistic-libertarian-foreknowledge scenario, Backwards-Because is true and Forwards-Because is false. This seems crucial to us. If the eternalistic-libertarian-foreknowledge scenario is true, then Trump’s decision will not be entailment-L-free, despite the fact that it will be strong-causal-L-free. But libertarians should not care about this, or see this as a lack of freedom of a kind that is worth wanting, or worth caring about. The reason that libertarians should not care about the lack of entailment-L-freedom here is precisely that, in this scenario, Backwards-Because is true and Forwards-Because is false.

We can put the point like this. What libertarians should care about is it not being the case that we are made to choose in the ways that we do. But they should not be bothered by the fact that our decisions can make other facts obtain. Therefore, if the eternalistic-libertarian-foreknowledge scenario is true, then entailment-L-freedom is not something that libertarians (or anyone else) should care about. For in this scenario, we lack entailment-L-freedom not because we are made to choose in certain ways, but because our decisions make other facts obtain—in particular, because they make God know about our decisions at earlier times. In short, we should not be bothered by our lack of entailment-L-freedom in this scenario because the direction of dependence here is backwards.

Finally, it is important to remember that we do not need entailment-L-freedom in order to have a robust libertarian kind of free will. For we could still have strong-causal-L-freedom.

6.b. Why We Should not Care about Unsettled-L-Freedom

Recall from section 2.c that unsettled-L-freedom is just like strong-causal-L-freedom except that it requires the decisions in question to be settlement-undetermined rather than strong-causal-undetermined.

It is obvious that if eternalism is true, then we are not unsettled-L-free—because if eternalism is true, then all of our decisions are settled in advance. But we think that libertarians should not care at all about this. The argument for this claim is analogous to the argument that we just ran in section 6.a. In short, we think it can be argued that if the eternalistic-libertarian-foreknowledge scenario is true, then unsettled-L-freedom is not worth wanting, or worth caring about. We will not argue for this point in full—because the argument is so similar to the argument of section 6.a—but it is worth saying a few words about this here.

If the eternalistic-libertarian-foreknowledge scenario is true, then Trump’s decision to order Bubble Gum Surprise will be strong-causal-L-free, but it will not be unsettled-L-free. But the reason that Trump’s decision will not be unsettled-L-free in this scenario is not that Trump will be made to choose Bubble Gum Surprise. Rather, the reason that Trump’s decision will not be unsettled-L-free in the eternalistic-libertarian-foreknowledge scenario is that, in that scenario, it was already the case in 2010 (and at every other moment in time) that Trump’s making of the decision—i.e., the event in which (a) Trump himself chose Bubble Gum Surprise (in a strong-causal-L-free way), and (b) nothing made him choose in that way—was part of reality.

But it seems to us that libertarians should not care about this at all. If Trump himself made the decision, and if nothing made him choose in the way that he did, then that gives us what we want—or what we should want—out of libertarian freedom. Why should we care whether our strong-causal-L-free decisions—the decisions that we make, without anything making us choose in the ways that we do—exist eternalistically? The eternalistic existence does not do anything to undermine what we care about—or what we should care about—vis-à-vis libertarian freedom. For the eternalistic existence does not do anything to undermine the extent to which these decisions are up to us.

In short, the point is as follows: It might seem that if all of our future decisions are already settled right now, then we do not have a genuinely libertarian sort of free will. But if eternalism is true, then that thought is simply mistaken. For if eternalism is true, then the already-settled-ness of our future decisions is perfectly compatible with the claim that all of our future decisions are such that (a) they are made by us, and (b) nothing makes us choose in the ways that we do in those decisions. Moreover, as we have seen, the claim that our future decisions are already settled in an eternalistic way is perfectly compatible with the claim that these decisions are free in a robust libertarian way because it is compatible with the claim that they are strong-causal-L-free.

6.c. Fatalism

Given the remarks of section 6.a, we can now say what is wrong with standard fatalist arguments for the incompatibility of free will and God’s foreknowledge. There are various ways to articulate the fatalist argument, but they all rest on premises like the following:

For example, Pike’s classic (1965) version of the argument contains the following premise:

And similar premises appear in versions of the argument presented by others, e.g., Hasker (Reference Hasker1989), Zagzebski (Reference Zagzebski1991), Hunt (Reference Hunt1999), Finch and Rea (Reference Finch, Rea and Kvanvig2008), Merricks (Reference Merricks2009), and Todd (Reference Todd2013).

We admit that if ‘power’ is defined in terms of entailment-L-freedom, then (F) is true. But if ‘power’ is defined in terms of strong-causal-L-freedom, then if the eternalistic-libertarian-foreknowledge scenario is true—i.e., if eternalism is true, and if human beings are strong-causal-L-free, and if God has foreknowledge of our strong-causal-L-free decisions (and, remember, we argued in section 5 that these claims are all compatible)—then (F) is false. The key to appreciating this point is to recall from section 6.a that if the eternalistic-libertarian-foreknowledge scenario is true, then whenever people make decisions that are strong-causal-L-free, they make it the case that God had certain beliefs in the past. Now, if a decision is strong-causal-L-free, then (by definition) the agent in question could have chosen differently—in a strong-causal sense of ‘could have chosen differently’—and it follows from this that if we define ‘power’ in terms of strong-causal-L-freedom, then if a decision is strong-causal-L-free, then the agent in question had the power to choose differently. But if we combine this with the argument of section 6.a, we get the following result:

But (F) just says that people do not have the power mentioned in (*). Thus, since (*) is true, it follows that if the eternalistic-libertarian-foreknowledge scenario is true—and, again, we have already argued that that scenario could be true—then if ‘power’ is defined in terms of strong-causal-L-freedom, (F) is false.

It is important to note that we are not saying that people have the power to change the past. In other words, we are not endorsing claims of the following kind:

That is not our view. Our view is that if eternalism is true (and if an omniscient God exists, and if Jane’s decision was strong-causal-L-free, and if ‘power’ is defined in terms of strong-causal-L-freedom), then God’s t₁ belief was always already influenced by Jane’s t₂ decision. Remember, if eternalism is true, then it is not the case that t₁ came into being and then, later on, t₂ came into being. No, if eternalism is true, then Jane’s t₂ decision already existed when God believed at t₁ that Jane would decide to do A at t₂; and the reason that God believed at t₁ that Jane would decide to do A at t₂ is that (a) her t₂ decision already existed, and (b) God was already aware of her decision. So the view here is that if eternalism is true (and if an omniscient God exists, and if human beings are strong-causal-L-free, and if ‘power’ is defined in terms of strong-causal-L-freedom), then (F) is false not because people have the power to change a pre-existing past—a past that existed before our decisions came into being—but because, when people make decisions that are strong-causal-L-free, they have the power to influence what God’s beliefs were all along. So, e.g., if at t₂, Jane made a strong-causal-L-free decision to do A rather than B, then our claim is that, at t₂, Jane had the power to do A and she also had the power to do B—and so she had the power to make God believe at t₁ that she would decide to do A at t₂, and she also had the power to make God believe at t₁ that she would decide to do B at t₂.

So, again, if (a) eternalism is true, (b) human beings are strong-causal-L-free, (c) God has foreknowledge of our decisions, and (d) ‘power’ is defined in terms of strong-causal-L-freedom, then (F) is false. And exactly analogous remarks can be made about other (F)-style premises in the literature, e.g., Pike’s premise (4)—they too come out false in the (a)-(d) scenario.

Now, some philosophers (e.g., Todd (Reference Todd2013)) have argued against the rejection of (F)-style premises. But their arguments have more or less assumed presentism—or at any rate, they have not considered how the truth of eternalism would affect the situation. As we have argued, if eternalism is true, then the rejection of (F)-style premises becomes non-problematic.

Now, you might object here that there is a powerful argument for (F) that is based on the idea that we humans cannot change the so-called “hard” facts about the past. For instance, you might argue for (F) by appealing to the following two premises:

Whether facts about God’s past beliefs (about what was then the future) count as hard facts or soft facts depends on how we define ‘hard fact’ and ‘soft fact’. But we do not think this matters. For if we define these terms in a way that makes premise (ii) come out true, then if eternalism is true (and if humans are strong-causal-L-free, and if ‘power’ is defined in terms of strong-causal-L-freedom), then premise (i) will be false. We have already argued for this claim. For we have argued that if eternalism is true (and if an omniscient God exists, and people are strong-causal-L-free, and if ‘power’ is defined in terms of strong-causal-L-freedom), then when people make strong-causal-L-free choices, they have the power to make it the case that God has different beliefs in the past. And it follows from this that if ‘hard fact’ is defined in a way that makes the facts in question here (i.e., facts about God’s past beliefs about what was then the future) count as hard facts, then human beings have the power to make it the case that at least some of the hard facts about the past are different than they actually are.

Our conclusion, then, is this: if we read fatalist arguments as arguments for the incompatibility of God’s foreknowledge and strong-causal-libertarianism, then they are not good arguments because they rely on (F)-style premises, and we can reject those premises by endorsing the eternalistic-libertarian-foreknowledge scenario—i.e., by maintaining that eternalism is true, that human beings are strong-causal-L-free, and that God has foreknowledge of our strong-causal-L-free decisions. Or to put the point differently, (F)-style premises are question-begging because they assert that people do not have powers that we in fact do have if the eternalistic-libertarian-foreknowledge scenario is true.

Now, our arguments here do not do anything to undermine fatalist arguments that target the compatibility of God’s foreknowledge and entailment-L-freedom. But we argued in section 6.a that if eternalism is true (and if an omniscient God exists), then entailment-L-freedom is not worth wanting, or worth caring about.

6.d. Give Us a Metaphysical Model (of Liberty) or Give Us Death

We end by comparing our views to those of a few other philosophers. The first point we want to make here is that while others have responded to fatalist arguments by rejecting premises like (F)—see, e.g., Talbott (Reference Talbott1986), Merricks (Reference Merricks2009), and Westphal (2011)—none of them appeals to eternalism, and the claim that (F)-style premises are false is much less plausible if eternalism is not true.Footnote ²³ Or to put the point differently, the difference between us and these other writers is that, unlike them, we have provided a metaphysical model that explains why (F)-style premises are false.

We think this is crucial. Zagzebski (Reference Zagzebski1991, p. 179) says that in order to solve the problem of freedom and foreknowledge,

Zagzebski—who tries to solve the problem of foreknowledge without appealing to eternalism—admits that her own view falls short of providing such a metaphysical story. But we take ourselves to have provided precisely the sort of metaphysical story, or metaphysical model, that is needed here. It is a story or model on which strong-causal-libertarianism, eternalism, and God’s foreknowledge are all true. The point we are making here is that others who have rejected (F)-style premises—e.g., Talbot, Merricks, and Westphal—have not provided such models.

Similar remarks can be made about philosophers who have responded to fatalist arguments in different ways. We will just mention three cases here. First, Wasserman (Reference Wasserman2020) argues that as long as God’s foreknowledge of our future actions is not causally relevant to those actions, it is compatible with the claim that those actions are free. But Wasserman does not have a story to tell about how God could know what we are going to do in the future in a way that is compatible with our being free.

Second, Tooley (Reference Tooley2010) responds to a certain fatalist argument by providing a linguistic analysis that enables him to say that one of the premises in the fatalist argument is false. But he does not give us a metaphysical model, and what is more, his analysis is extremely implausible—or so it seems to us. He argues that sentences of the form ‘It is true at t₁ that person S will do A at t₂’ are not true—even if eternalism is true, and even if S does do A at t₂. But this is hard to believe; if eternalism is true, and if S does A at t₂, then it just is true at earlier times that S will do A at t₂, and no convoluted linguistic analysis can change this. (It is also worth noting here that, unlike us, Tooley does not argue that libertarianism and eternalism are compatible; moreover, according to Tooley, these two views are not jointly compatible with God’s foreknowledge.)

Finally, similar remarks can be made about Oaklander (Reference Oaklander and Le Poidevin1998). He argues that what he calls “the new theory of time”—which seems to entail eternalism, although Oaklander is not very clear about this—is compatible with the existence of human free will. But like Tooley, Oaklander’s argument is based on a linguistic analysis of certain kinds of sentences; indeed, Oaklander is concerned with sentences of the same general kind that Tooley is concerned with, and he has a very similar view of those sentences. (It is also worth noting that, unlike us, Oaklander does not argue that eternalism is compatible with the existence of libertarian freedom.)

We end by pointing out that while we should all agree with Zagzebski that what we really need here is a metaphysical model—and that we should not be satisfied with a mere linguistic analysis—libertarians in particular should agree with this. For they are already on record (or at any rate, they should be on record) as disliking linguistic solutions to problems about free will. They should claim with Kant that compatibilist views of free will involve a kind of “petty word jugglery.” So, if they think that compatibilist views are unacceptable for this reason, then they should not engage in “petty word jugglery” themselves; they should not try to solve the problem of freedom and foreknowledge by merely coming up with a clever analysis of certain kinds of sentences. They should instead try to provide a metaphysical model that explains how libertarianism and God’s foreknowledge could both be true.