Understanding theological incompatibilism

Theological incompatibilism has been formalized many different ways.Footnote ⁴ The following is faithful to most arguments that defend it:

• Assume that God exists and call whatever action you will perform at noon tomorrow ‘X’.

  1. (1) If you can freely perform X at noon tomorrow, it must be possible for you to not perform X at noon tomorrow. [from the principle of alternate possibilities]Footnote ⁵

  2. (2) God believed yesterday that you will perform X at noon tomorrow. [this follows from God's omniscience]

  3. (3) God's beliefs are events.

  4. (4) All past events are necessary (the past cannot be undone).

  5. (5) God's past belief that you will perform X at noon tomorrow is necessary. [from (2), (3) and (4)]

  6. (6) Necessarily, if God believes something will occur, it will occur. [this follows from God's necessary omniscience]

  7. (7) It is necessary that you will perform X at noon tomorrow. [from 5 and 6 by a ‘transfer of necessity’]

  8. (8) It is not possible for you to not perform X at noon tomorrow. [from (7)].

  9. (9) Therefore, you cannot perform X at noon tomorrow freely. [from (1) and (8)].

It is common for those proposing the argument to justify the move from [(5), (6)] to (7) by a transfer of necessity principle (see Zagzebski (Reference Zagzebski1991), ch. 1). In such principles, often expressed ‘[□P . □(P⊃Q)] ∴ □Q’, the necessity of the antecedent is logically ‘transferred’ via entailment to the consequent if the antecedent necessarily implies the consequent. In the argument above, the necessity of God's past belief is logically ‘transferred’ via entailment to your performance of action X, since God's past belief necessarily implies that you will perform action X.

Consequently, many have the impression that theological incompatibilism suggests God's past beliefs would make us un-free. Since, in the argument, the necessity of God's past belief is ‘transferred’ to your future action, it seems that your future action is necessary because God's past belief is necessary and necessarily implies it. But in fact, the argument suggests nothing of the kind. The word ‘transferred’ is misleading; ‘□P . □(P⊃Q)’ does not make ‘□Q’ true (Q need not have acquired its necessity from ‘□P . □(P⊃Q)’). Yes, ‘□P . □(P⊃Q)’ entails that ‘□Q’ is true, but entailment is a logical relation, not an explanatory or causal one. (For example, the truth of ‘{□(A⊃A). □[(A⊃A) ⊃ (1+1=2)]}’ entails but does not explain the truth of ‘□(1+1=2)’.) In the same way, the necessity of God's past belief plus its necessary implication of your future actions does not explain why your future action is necessary – it merely entails that it is. Thus, the question remains: If we would not be free if our future actions are necessary, why are our future actions necessary? What would make us un-free, if God existed, according to theological incompatibilism?

To answer this question, a deeper understanding of theological incompatibilism is needed, but an obstacle stands in the way: the argument is ambiguous. Notice that it repeatedly utilizes the words ‘possible’ and ‘necessary’ but never offers a definition of these words. As any good logic student knows, there are many different definitions of ‘possible’ and ‘necessary’ (i.e. there are many different modalities), and unless a definition is offered and the ambiguity clarified, a deeper understanding of the argument is unattainable.

Some, like Alston (Reference Alston1985), have identified how this ambiguity has negatively affected the debate; some have even tried to more precisely define the kind of necessity had by the past.Footnote ⁶ But to solve the problem, a specific modality needs to be chosen and consistently applied to the premises of the argument, and then the new unambiguous argument needs to be tested for validity and soundness. But which modality would work? Which modality would produce the most convincing argument? A thorough answer to this question would require lengthy elaboration but – regardless of how convincing other modalities might make the theological incompatibilist argument (see Johnson (Reference Johnson2006) ) – there is at least one modality that generates a sound theological incompatibilist argument: I call it ‘the ontological modality’.

The ontological modality is motivated by the modality utilized by Aristotle and Ockham in their writings on fatalism; they defined the relevant concept of possibility by the limits of ‘determinate reality’ (see Adams (Reference Adams and Norman1983) ). If something does not stand contrary to determinate reality, there is still ‘potency in things’ for its occurrence and thus it is possible. Of course, determinate reality may change over time (it depends on which temporal ontology is true); thus a formalized definition will need to be temporally sensitive. But, nuances aside, determinate reality is that which exists – that which would be included in an accurate ontology. Thus, we can formalize the Ockhamistic/Aristotelian definition: Something is now-ontologically possible if and only if it is not contrary to (in that it is not the opposite of) anything that exists. X is now-ontologically impossible if and only if there is some thing Y (i.e. Y exists) and positing the existence of X is inconsistent with positing the existence of Y.

By using the ontological modality to disambiguate the above argument, we produce the ontological version of the theological incompatibilism argument. It looks like this:

• Assume that God exists and call whatever action you will perform at noon tomorrow ‘X’.

  1. (1) If you can freely perform X at noon tomorrow, it must be now-ontologically possible for you to not perform X at noon tomorrow. [from the principle of alternate possibilities].

  2. (2) It is now-ontologically possible for you to not perform X at noon tomorrow only if your not performing X at noon tomorrow is not contrary to something that exists. [by definition of ‘ontological possibility’]

  3. (3) God's past belief that you will perform X at noon tomorrow exists.

  4. (4) God exists and cannot have false beliefs.

  5. (5) It is now-ontologically necessary that, if God believes you will perform X at noon tomorrow, you will perform X at noon tomorrow. [from (4)]

  6. (6) Your future performance of X at noon tomorrow exists. [from (3) and (5) by the ‘transfer of ontological necessity’]

  7. (7) Your not performing X at noon tomorrow is contrary to something which exists. [from (6)]

  8. (8) It is not now-ontologically possible for you to not perform X at noon tomorrow. [from (2) and (7)]

  9. (9) Therefore, you cannot freely perform X at noon tomorrow. [from (1) and (8)]

Only (3) and (4) are undefended. Premise (4) follows from the stated assumption that God exists, but one might question premise (3), wondering why God's past beliefs exist. The answer lies in debates (in temporal ontology) over questions like: To what extent can it be said that the past or the future exist? Those who suggest that the past exists defend their position by suggesting that, if there are truths about the past, one must believe that the past exists. If one does not, one has no truth-makers in one's ontology for truths about the past. For example, if you think that it is true that God had foreknowledge yesterday you must believe that God's past justified true beliefs about the future exist. If you do not, there is nothing in your ontology – you have no truth-maker – that explains the truth of ‘God has foreknowledge’.

The need for truth-makers is fuelled by the correspondence intuition: the intuition that propositions are true if and only if they correspond with the way the world is. (As Aristotle put it: ‘A proposition's truth or falsity is due, accordingly, to whether or not that proposition corresponds or fails to correspond with determinate reality’ (Adams (Reference Adams and Norman1983), 6). As others have put it more simply: ‘Truth supervenes on existence.’) Since most find this notion intuitive – Newhard (Reference Newhard2004) suggests that it is stronger than our intuition regarding non-contradiction – and most think there are past truths, most think the past exists – at least in some respect. We will return to the temporal ontology debate shortly – it is not yet clear if ‘truth about the past’ requires the past to exist – but we now have solid support for premise (3).

One might also challenge premise (2), wondering why your future non-performance of X is possible only if it does not stand contrary to something that exists (like your future performance of X). But we can find a defence for this premise by considering our relationship to the past. President Kennedy not being shot stands contrary to something that exists: the past event of Kennedy being shot. Since, once an event occurs it cannot un-occur, the existence of the event of Kennedy being shot is fixed. To suggest otherwise is to suggest that, even though Thursday's past contains the event of Kennedy being shot, Friday's past might not. But this is contradictory. 22 November 1963 is in both Thursday's and Friday's past and it cannot both contain and not contain the event of Kennedy's assassination. The existence of the event of Kennedy being shot makes its non-occurrence now-impossible. Kennedy was shot and there is nothing we can do about it now; the past already exists. The same would also be true of any present or future event that one supposed exists; once it has existence, it cannot later lack it. To suggest otherwise would be to suggest that some particular moment in time (whether present or future) can both contain and not contain some specific event. Although one might have expected Y on Thursday but not on Friday, the occurrence of Y next Saturday cannot be in Thursday's future without also being in Friday's.

Again, it still might be that truth about a time does not require that time to exist. But what the above argument makes clear is that, if any event already has existence, this status cannot be changed (regardless of whether it is a past, present or future). To suggest otherwise lands one in a contradiction. That being the case, the defence of the second premise is fairly straightforward. If you not performing X at noon tomorrow stands contrary to something that exists then you will perform X at noon tomorrow and there is nothing you can do about it now – it is ontologically necessary.

Lastly, one could question premise (1). It is justified by the principle of alternate possibilities, a classic statement of a necessary condition for free will. But one might wonder whether that principle is intuitive when expressed using the ontological modality. If it simply suggests that an action is free only if not performing the action is ontologically possible, it is not intuitive; that would make free action a logical impossibility. As you perform any action, the action exists and is thus ontologically necessary; therefore, as you perform any action, not performing the action is ontologically impossible. So, by the above criteria, all actions would be non-free – as they occur, their mere occurrence would entail they are not free. So the above criteria will not do.

Since the ontological necessity an action has while it is being performed does not prevent the action from being free, one might think ontological necessity cannot prevent any action from being free. But that is not the case. The following criterion seems quite intuitive.

And there are also reasons to think this true.

First, one reason our intuition that the ontological necessity an action has while it is being performed does not prevent the action from being free seems right is this: the action attains its ontological necessity, by attaining its ontological status, at the moment it is performed. But if it had that status and necessity before it was performed, it would not be attaining that status and necessity at that moment. So we could not use that as a reason to justify our intuition (which it seems we want to do). If there is no other reason for holding this intuition, we might be forced to reject it. This is at least some justification for the above criteria. Second, an action having ontological necessity before it occurs is contrary to the central assumption we all make when deliberating about whether or not to perform the action. When doing so, we assume that whether or not the action is going to be performed is yet to be determined – whether or not it will occur is what we are trying to decide – and not until the end of our deliberation will it be determined.

So both the performance and the non-performance of the action must be an ontologically possible result of our deliberation; otherwise our deliberation lacks the central quality we think it does. But if the performance of the action is ontologically necessary as we deliberate, even though our deliberation may still bring about the performance of the action, it could – contrary to assumption – do nothing else. Granted, once the deliberation is done and the action is being performed, the action being ontologically necessary at that moment does not contradict its being free. But if the action is ontologically necessary before it is performed, the central assumption we make in deliberation is grossly mistaken.

Objections to similar criteria have been raised: Rogers (Reference Rogers2007) imagines a deterministic world where actions usually are not free, but at one moment God steps in to divorce you from the deterministic system, allowing you to perform whatever action you wish, free of influence from the system, at that moment. Since the world was deterministic up until the moment you acted, it seems that you could not have done otherwise before you acted; yet – contrary to the criterion – whatever you do in that moment, you do freely. But despite problems regarding the impossibility of simultaneous cause-and-effect (which Rogers anticipates and attempts to answer), a number of objections can be raised.

For one, I highly doubt that free action is possible the very moment God frees you from the deterministic system. Some kind of non-determined deliberation seems to also be necessary for even the simplest of free actions – and that takes time. In addition, since God freeing you from the deterministic system is clearly possible in this scenario, it can hardly be said that your non-performance of the action was impossible before the action was performed. God freeing you from the system and you choosing not to do what you did is a genuine possibility in this story. In fact, the very thing that makes free action possible – the divine action of divorcing you from the deterministic system – is the very thing that makes it true before the action occurs that acting otherwise is possible. So perhaps Rogers's story actually supports my criterion.

The debate could go on, but it seems the above criterion will suffice as a necessary condition for free will that uses the ontological modality; the intuitive force of the criterion has been established. (Of course, once you see where my argument leads us, you may want to come back and deny it, but we will cross that bridge when we come to it.) And if an action is free only if not performing the action is ontologically possible before the action is performed, then if you are going to freely perform X at noon tomorrow it must be now (as well as all other times until the action is performed) ontologically possible for you to not perform X at noon tomorrow. So it seems we have a good defence of premise (1).

We can now clearly see what, according to theological incompatibilism, makes us un-free. The argument suggests that you not performing action X is possible only if your future performance of action X does not already (before the action is performed) exist. (And, since the former is required for free will, but the latter is false, you are not free.) But it is not God's foreknowledge that makes you un-free – it is not God's foreknowledge that causes your future performance of action X to already exist. What is responsible for your non-freedom is the mere fact that your future action already has – before you perform it – existence. If one could establish this without reference to God's foreknowledge, one would not even need God's foreknowledge to produce the fatalist conclusion.

The failure of the classic solutions

Why the classical solutions fail to solve the ontological version of the theological incompatibilist argument can now be made clear and doing so will also help us further understand the argument.

Many of the classic solutions try to solve the problem by re-describing God's foreknowledge in a way that is compatible with free will. But – since it is an existing future and not God's foreknowledge that hinders our free will – unless a solution's re-description of God's knowledge allows for the possibility of the future not existing, the solution will fail. For instance, suggesting that God is timeless – that God sits outside and views the entire timeline as a whole – does not solve the problem; the timeline must exist as a whole – past, present and future – if God is viewing its entirety and sitting outside of it. The timeless solution reinforces the problem (the fact that the future exists); it does not solve it. Suggesting that God's past beliefs are ‘soft facts’ – i.e. suggesting their content is determined by the future events to which they refer – does not deny (but again reinforces) the notion that the future exists. If their content is determined by the future, the future must exist. Additionally, suggesting that our future actions are what cause God's past beliefs does not deny – but again reinforces – the notion that the future exists. If you accept that God's past beliefs exist, but think their cause lies in the future, unless you accept that caused events can exist without their causes existing, you must also think that the future exists.

Another major type of solution that does not try to re-describe God's knowledge but fails all the same, is the ‘backtracking counterfactuals’ objection. This solution, raised first by Molina (see Molina (Reference Molina and Freddoso1998) ) and then by Saunders (Reference Saunders1966) and Plantinga (Reference Plantinga1977) in response to Pike's original paper, can be applied to the ontological version of theological incompatibilism in the following way:

But this solution completely misses the point of the argument. It is true that if you chose to refrain you would; it is also true that the mere fact that you will not refrain does not entail that you lack the power to refrain. But both are beside the point. If you cannot refrain, then refraining is beyond your power. And if God's beliefs and consequently the future cannot be different than they already are, but your choosing to refrain would require them to be, then – not only will you not refrain but – you cannot refrain. That God's beliefs and thus the future cannot be different than they are is exactly the point of the argument. The unchangeability of God's past beliefs is ‘transferred’ to your future action, and thus your future action is also unchangeable. Consequently, even though your refraining would make a different past and future exist, a different past and future no longer can exist; thus you cannot refrain. The backtracking counterfactual objection fails.

The fact that Molina subscribed to the above solution might be confusing to some readers; the assumption that Molina espoused the ‘middle-knowledge’ solution to theological incompatibilism is common. Middle knowledge is knowledge ‘of what any possible free creature would freely do in any possible circumstance’ (Zagzebski (Reference Zagzebski1991), 125). Although Molina thought that God had middle knowledge, he did not think that the doctrine solved theological incompatibilism. And he was right; in fact using middle knowledge as a solution to any version of theological incompatibilism is simply question-begging. Middle knowledge just is prior knowledge of the free actions of agents (both possible and actual); but whether or not such knowledge is even possible is exactly the issue; that it is impossible is what theological incompatibilism concludes. One cannot merely assume that God has such knowledge to successfully avoid the incompatibilist conclusion. Molina realized this and espoused the above solution instead (see Molina (Reference Molina and Freddoso1998) ).

None of the classic solutions show the ontological version of theological incompatibilism to be unsound.

The reduction of theological incompatibilism to logical incompatibilism

I have argued that if we suppose that God exists and has foreknowledge, it follows that we are not free. But under that supposition, God's foreknowledge would not be responsible for our non-freedom; but the existence of the future would be. But if the future exists, one wonders why it does.

The doctrine that the past, the present and the future exist is (what Armstrong (Reference Armstrong2004) calls) ‘omnitemporalism’ or (what Rea (Reference Rea, Loux and Zimmerman2003) calls) ‘eternalism’. Physicists call it the ‘block-world view’ and suggest that it is entailed by special relativity. The consistency of the speed of light in all reference frames entails that (among other things) simultaneity is relative. Even though, in my reference frame, present event B happened after past event A but before future event C, in the reference frame of another person (travelling quite fast), A and B occur simultaneously. As the other person experiences B (as I too experience it), he is also experiencing A. I must thus conclude that A ‘still exists’ even though it is, for me, a past event. If I coexist with someone, who also coexists with A, I must coexist with A, even though it is a past event. Another person in another reference frame will experience B and C simultaneously. In the same way, and for the same reason, I must thus conclude that C ‘already exists’ even though it is, for me, a future event. For any given two events, either there is some reference frame in which they coexist, or they coexist with a set of mutually coexisting events. Either way, since there is no preferred reference frame that tells us ‘the truth’ about simultaneity and existence, if special relativity is true all events throughout time and space coexist with each other. We live in a ‘block world’ where the past, the present and the future all equally exist.Footnote ⁷

But – just like foreknowledge – special relativity does not make the future exist, it merely entails that it does. If you ask a physicist why the universe is a ‘block world’ she will say: ‘It just is; the block-world view accurately describes the way the universe is, but there is no explaining why the universe is this way. We merely know that the universe is a block world because it is entailed by special relativity.’ In the same way, if you ask one who believes that we are not free based on the theological incompatibilism argument why omnitemporalism is true, she will say ‘It just is; omnitemporalism accurately describes the way the universe is, but there is no explaining why the universe is this way. We merely know it is this way because omnitemporalism is entailed by divine foreknowledge.’

One might wonder, however, why comprehensive foreknowledge commits one to omnitemporalism. The ontological version of theological incompatibilism claims that it does, but that claim relies on the transfer of ontological necessity principle. But could not that principle be challenged? I am inclined to think that the principle would withstand criticism; but whether or not it could is actually beside the point. Even if the principle is false, it can still be argued that God's comprehensive foreknowledge entails omnitemporalism. Recall the need for truth-makers for truths about the past; the past must exist if there are truths about it. But, in the same way, it can be argued that the future must exist if there are truths about it. But God's foreknowledge is – just that – fore-knowledge; and knowledge is (at the least) justified true belief and thus foreknowledge is (at the least) justified true belief about the future. Thus, the theological incompatibilist could defend his thesis by arguing that God's foreknowledge entails that there is comprehensive prior truth and thus entails that the future exists, regardless of whether the above transfer principle holds.

What lies at the root of theological incompatibilism is now clear. If theological incompatibilism is sound, the doctrine that there is comprehensive truth about the future, by itself, is enough to derive that the future exists; thus, if theological incompatibilism is sound, the fatalist conclusion can be derived solely from comprehensive prior truth; so, if theological incompatibilism is sound, logical incompatibilism is sound.Footnote ⁸ In fact, if theological incompatibilism is sound, God's foreknowledge entails the fatalist conclusion because God's foreknowledge requires comprehensive prior truth. Thus, we can conclude, theological incompatibilism reduces to logical incompatibilism; logical incompatibilism is the more basic and fundamental problem; the incompatibility expressed in logical incompatibilism is responsible for the incompatibility expressed in theological incompatibilism.Footnote ⁹ The majority view – that theological fatalism is the more severe problem – is mistaken.

Dealing with objections

Dummett, who acknowledges that comprehensive prior truth demands an existing future, does not think that this entails a fatalist conclusion. ‘The illusion derives from the idea that the present truth must compel the future action. The efficacy is in the reverse direction: a proposition about which I am going to do is true in virtue of my later action’ (Dummett (Reference Dummett2004), 81). But as an objection to the ontological version of logical incompatibilism, Dummett's argument falls short and contains a mistake similar to one contained in a proposed solution to theological incompatibilism. Pointing out that our future action is not brought about by God's past belief does not solve the theological incompatibilist argument for that is not what drives its fatalist conclusion. In the same way, pointing out that truths about the future do not compel future actions does not solve logical incompatibilism for that is not what drives its fatalist conclusion. Logical incompatibilism does not suggest that comprehensive prior truth entails we are not free because ‘truth compels action’. It suggests that we are not free because comprehensive prior truth entails an existing future that cannot be changed. Unless one can argue that comprehensive prior truth does not entail this – an argument to which we will shortly return – one will not even approach solving the problem. The fact that truth does not compel future action is irrelevant.

One might also argue:

This is a common view regarding propositions and if it is true, it is incorrect to say that a proposition is true at a time. I will not enter into the debate regarding what kind of entities propositions are; but the answer does not matter for our purposes. In the same way that suggesting God is timeless does not solve, but reinforces the problem, suggesting that propositions are timeless does not solve but reinforces the problem. If propositions are timeless they exist outside of the timeline, but the timeline must exist as a whole (past, present and future) if they are to do so. So suggesting that propositions are timeless does not avoid but reinforces the notion that the future and past exist and thus reinforces the fatalist conclusion. In addition, moving propositions outside of time does not deny the need for them to have truth-makers and thus will not avoid the ontological commitments that drive home my conclusion. What this objection does make clear, however, is that the suggestion that propositions are timeless stands contrary to free will; but this just means that a rejection of such a view will need to be added to the list of the many things that we need to do if we are to retain belief in free will.

But the most important objection to deal with is the one that that I have been alluding to – the one that is on the tip of everyone's tongue: the presentist objection.

It is true that the presentist temporal ontology is still a viable option; this is why I was careful not to endorse the temporal ontology suggested by the above incompatibilist arguments. No argument I have mentioned so far proves that a proposition about a time cannot be true unless that time exists. So the presentist could argue that both the theological and logical incompatibilist arguments presented above are unsound. But the presentist still cannot avoid theological and logical incompatibilism and the conclusions I have drawn about their relationship. Why?

Because: presentists do not deny the correspondence intuition; they suggest that their ontology can supply truth-makers for true propositions about the past and future. They do so by expanding ‘the class of objects, events, properties and relations that are taken to serve as truthmakers’ (Rea (Reference Rea, Loux and Zimmerman2003), 261), and suggesting that such things presently exist and serve as the truth-makers for propositions about the past and future. But this means that the only way that the presentist temporal ontology is fundamentally different from omnitemporalism is regarding where (or, more accurately, ‘when’) the truth-makers for such propositions are. For the omnitemporalist, the truth-maker is an event that exists in the future; for the presentist it is an abstract object or a future tensed property of a material object that exists in the present.Footnote ¹¹ But either way, the truth-maker exists. And since, if anything already has existence, this status cannot be changed (regardless of whether it is a past, present or future) adopting the presentist ontology cannot avoid the fatalist conclusion of the incompatibilist arguments or, consequently, the conclusions of my argument! Yes, strictly speaking, they can deny the existence of the future and the past, and could deny the truth of certain premises in the ontological versions of the above arguments – but they do not deny the existence of something just as damning to free will: presently existing and unchangeable truth-makers for propositions about the future.

We could thus re-term the entire argument we have seen so far in terms of ‘presently existing truth-makers’ and arrive at the same conclusions. The fatalist conclusion of theological incompatibilism would be driven by ‘presently existing truth-makers’ (instead of an ‘existing future’), it would be clear that the traditional solutions fail because they do not deny the existence of such truth-makers, and it would be obvious that logical incompatibilism is what truly lies at the root of theological incompatibilism because it is the truth of propositions about the future that requires those truth-makers to presently exist.

Whether one is a presentist or not, it is clear that theological incompatibilism reduces to logical incompatibilism; in addition one must acknowledge that some version of both theological and logical incompatibilism – perhaps in terms of an ‘existing future’; perhaps in terms of ‘presenting existing truth-makers’ – is sound.

The consequences

I have argued that theological incompatibilism has been misunderstood. Regardless of whether it is put in presentist or non-presentist terms, when the theological incompatibilist argument is disambiguated, it becomes clear that it reduces to logical incompatibilism. And, although presentists and non-presentists will disagree about ‘when’ truth-makers for propositions ‘are’ and thus disagree about which version of the argument is sound, they must acknowledge that one version is. The consequences of my argument are threefold.

First, theists are forced to recognize theological incompatibilism as a serious threat to free will – so serious that it requires revision of belief. Since foreknowledge requires truth-makers for propositions about the future to exist, but free will requires their non-existence, theists are forced to abandon one or the other.

Second, we must all take logical incompatibilism seriously. Comprehensive prior truth entails the fatalist conclusion. That there is comprehensive prior truth is entailed by the two most fundamental axioms of logic: bivalence and non-contradiction. If every proposition – whether timeless or not and including those about the future – has a truth-value (either ‘true’ or ‘false) and no proposition and its negation can have the same truth-value then, for any given proposition about the future, either it or its negation – which would also be about the future – is true. Thus, logical incompatibilism dictates a stern consequence: we must either give up bivalence, non-contradiction or free will.

Lastly, not even atheists can ignore theological incompatibilism as a threat to free will. The existence of God and His foreknowledge is not what drives the fatalist conclusion, so suggesting that God does not exist solves nothing. If the atheist merely admits that God could have foreknowledge if God existed, she admits to the existence of comprehensive prior truth, and that is enough to generate the fatalist conclusion.

Responses and conclusion

There are a few ways out of the problems I have raised. For example, one could reject the first premise by rejecting the principle of alternate possibilities. Those who do are often compatibilists; if they are right, and free will is compatible with determinism, then free will does not require your not performing X to be an ontological possibility and both the theological and logical incompatibility arguments are unsound. If, for example, Markosian agent causation is true (see Markosian (Reference Markosian1999) ), and one is free as long as one is the cause of one's actions – even if one could not have done otherwise – unchangeable truth-makers for propositions about the future could exist whenever you like and no threat to free will would be posed.

The debate about the principle of alternate possibilities is quite extensive. The most famous argument against the principle belongs to Frankfurt (Reference Frankfurt1969), who describes a situation where Black would ensure Jones does what Black wants but, since Jones does it anyway, Black does not intervene. It seems Jones acts freely, even though he does not have alternate possibilities. To reply to those who suggest that Jones does have alternatives in the form of neurological indications of his future actions, Fisher argues (1994, 2002) that such ‘flickers of freedom’ are not enough to ground moral responsibility (and, in turn, free will). Daniel Speak (Reference Speak2002), for one, disagrees. Others, such as Ginet (Reference Ginet1996) and Wyma (Reference Wyma1997), have argued that Frankfurt's example is impossible. If Jones acts freely, the world must be indeterministic; but if it is, Black cannot ensure Jones does what Black wants without intervening (thus making Jones's action unfree). In reply, compatibilists such as Mele and Rob (Reference Mele and David1998) have developed new versions of Frankfurt's original example. But since then, very good incompatibilist arguments have been presented by Hasker (Reference Hasker1998), Kane (Reference Kane2002c), and van Inwagen (Reference van Inwagen and Kane2002). The debate is not done, but accepting compatibilism is a way to avoid the fatalist conclusion. In general, however, it is not a solution that theists accept.

Another way out of the problems is to adapt the arguments of Katherin Rogers (Reference Rogers2007). She embraces a timeless-God solution to theological incompatibility arguments that is similar to C. S. Lewis's (Reference Lewis2001). She admits the solution entails omni-temporalism – what she calls a ‘four-dimensional or tenseless view of time’ (Rogers (Reference Rogers2007), 28) – but does not think it stands contrary to free will. Although she does not address the issue specifically, a faithful paraphrase of her argument could respond to mine:

If right, she can avoid the fatalist conclusion of the ontological versions of theological and logical incompatibilism altogether.

She wants her view to be consistent with libertarianism, but even she recognizes that not all libertarians will be satisfied with her position.

She defends against this objection, insisting that her position is libertarian, but in doing so she reveals that it is not. She admits that, in an omnitemporal universe, we lack the power to choose differently in a very significant sense, but she still maintains that we could be free.

She is able to do this because, on her theory, for a decision to be free it need merely be non-determined by outside forces and agent caused. But, contrary to her suggestion, this is not a libertarian definition of free will. On her theory, the ability to do otherwise in any sense really does not matter – only being the non-determined cause of the decision is required for the decision to be free. And, although most libertarians do suggest that this is required for free will, they do so because they think it is required for the power to do otherwise. Rogers's definition is actually closer to compatibilist definitions of free will; it is especially similar to Markosian's. If you are a libertarian, Roger's position is not for you.

For the libertarian, the only viable option for avoiding the fatalist conclusion is to make your non-performance of X ontologically possible before the performance of the action occurs; and the only way to do that is to eliminate the existence of the truth-maker of ‘You will do X at noon tomorrow’. And the only way to do that is to deny that ‘You will do X at noon tomorrow’ is true before it occurs; as long as you think it is true before it occurs – regardless of whether you think propositions are timed or timeless, or whether you are a presentist or not – you will need a truth-maker for it. But merely making it false will not work, since that will entail that another proposition about the future (‘You will not do X at noon tomorrow’) is true, has a truth-maker and thus is ontologically necessary. Thus, the only viable option for the libertarian who wishes to retain belief in free will, but not deny non-contradiction, is a rejection of bivalence via a rejection of comprehensive prior truth.Footnote ¹² If we hold that there are no truths about future free actions before they occur, the demand for the existence of the truth-makers for propositions about future free actions is lifted, and so is the fatalist conclusion.

Of course, one does not have to suggest that there are no truths about the future. There might still be future truths about non-free events and actions; perhaps present physical states of affairs physically determine some future events, and thus serve as truth-makers for truths about those future events. If this is the case, that is fine. What cannot be the case is that there are future truths about free actions; that would entail existing truth-makers for those truths and thus that those future actions are not free.

A rejection of bivalence will allow the theist to deny premise (3) of the ontological version of the theological incompatibilism argument, thus avoiding its conclusion. But it will also require the theist to give up belief in comprehensive divine foreknowledge. However, since knowledge requires truth and omniscience is simply believing everything true and nothing false, it will not require a rejection of divine omniscience. The view that God does not have comprehensive foreknowledge (‘open theism’) has been defended biblically (see Pinnock, et al. (Reference Pinnock, Richard, Sanders, Hasker and Basinger1994) ), philosophically (see Hasker (Reference Hasker1998) ) and theologically (Richard Rice (Reference Rice2007) has argued it is the only way to make sense of the doctrine of Trinity). Of course, this would also require everyone to abandon traditional bivalent logic. But so much good work has already been done on non-bivalent logic, this is not too tall an order; it seems that non-bivalent logics can work and in fact are quite intuitive (see Bourne (Reference Bourne2004) ).

One will also have to accept a specific kind of temporal ontology; one that does not include truth-makers for propositions about future free actions. This would include the ‘growing block’ temporal ontology, which suggests that only the past and present exists (but might contain truth-makers for future, non-free events). It would also include a presentist temporal ontology which only posits presently existing truth-makers for present, past and non-free future events. Either could deny the truth of propositions about future human free action and avoid the fatalist conclusion.

A reason to prefer the presentist option is Merricks's (Reference Merricks and Zimmerman2006) recent argument which suggests that the growing-block view entails certain philosophical difficulties. On the growing-block view, past and present moments are equally real, and the universe continually grows as new moments become real (as they become present) but then fall into the past. However, the only thing that changes about a moment in time is how it relates to the end of the block – when it first exists, it sits at the end of the block but as time progresses there are more and more moments between it and the end of the block; all of its other intrinsic properties stay the same. But this means, what it is like to exist in a present moment is indistinguishable from what it is like to exist in a past moment. Since there is very large (perhaps infinite) number of past moments, but always only one present moment, the growing-block theory has the unfortunate consequence of entailing that you should always conclude that what you are currently experiencing – when you exist – is the past, and not the present. Unless this bullet can be bitten or this argument answered, the growing-block view should be rejected. Consequently, the presentist view above seems the best way to avoid the fatalist conclusion.

This is not too surprising; Rogers predicted it.

But the fact that my conclusion is not happy does not mean that it is wrong, nor that my argument is flawed. What she helps me point out here is that if my argument cannot be answered, we are forced to conclude that – like classical theism and classic bivalent logic – the popular physical and metaphysical theories stand contrary to free will. If one really believes in free will, but cannot answer my arguments, one is forced to reject the popular theories. This is all the more reason to take the argument seriously; it is not a good reason to reject it, conclude that it is flawed, or ignore it.

But there are ways to revise the popular theories. As I mentioned before, special relativity entails the block-world view (i.e. omni-temporalism), so it will have to be dealt with if one wants to accept the presentist view I suggest. But if the reference frame in which the amount of cosmic microwave background radiation is constant throughout the universe is preferenced – as the ‘true’ one – then the simultaneity and coexistence of past, present, and future events in other reference frames is only apparent and one is not forced to conclude the block-world view is accurate. Tooley (Reference Tooley1997) also defends this conclusion (although in a slightly different way).

The need for brevity prevents me from developing these ideas further, but one thing is clear. The fatalist arguments have been misunderstood and brushed aside far too long. Both theological and logical incompatibilism present a serious threat to free will – both to the theist and atheist. We must work to revise our beliefs in response to these problems; we can ignore them no longer.Footnote ¹³