Introduction
The free-will defence is widely considered the best available theistic response to the logical argument from evil. According to this version of the argument, God's omnipotence together with His perfect goodness jointly entail that evil does not exist: for if God is omnipotent He is able to eliminate all evil, and if He is perfectly good He eliminates as much evil as is within His power to eliminate; yet evil exists, so God does not. The free-will defence consists roughly of the following claim: moral evil is directly attributable to human persons whose actions are not determined by prior conditions – including such potential determining influences as divine foreordination – for if all our actions were thus determined we wouldn't be significantly free in a way that confers moral responsibility for what we do. Since not even an omnipotent being can cause agents freely to do only what's right, God's omnipotence is not threatened by His failure to do so. And since among the worlds He can actualize, the worlds in which humans possess significant freedom are (ceteris paribus) more valuable than worlds in which they do not, God chose to actualize a world in which agents are free despite the fact that moral evil would result. And this decision does not impugn His moral goodness.
A number of philosophers have been persuaded that this familiar story (particularly as developed by Alvin Plantinga)Footnote 1 constitutes a successful defence of theism against atheological attack. But laying aside the problems specific to Plantinga's approach, a persistent band of critics has focused attention on an assumption considered more central to the viability of any free-will defence: the assumption that the libertarian analysis of free will is correct.Footnote 2 These critics contend that theological determinism – which we shall understand as the thesis that God's will is a sufficient cause of all worldly events – conflicts neither with free will nor with moral responsibility, because the correct analysis of these things reveals them to be fully compatible with determinism generally.
Now it seems plausible that if these critics are right, God could have created a world in which agents were determined to perform only morally virtuous actions without compromising the moral significance of their activities or the freedom they exercise in bringing them about.Footnote 3 And presumably He would have done this were He omnipotent and wholly good, assuming He wished to create any world at all. But He didn't. So free will, understood as per compatibilism, cannot serve as an explanation for the existence of moral evil that will be of any use to the theist, since the demands of free will present no obstacle to God's creating a world full of moral good and in which the absence of moral evil is guaranteed. Thus, David Lewis:
It seems that free-will theodicy must presuppose incompatibilism. God could determine our choices via our characters, thereby preventing evil-doing while leaving compatibilist freedom in tact. Thus He could create utopia, a world where free creatures never do evil … . The [theodicy's] opponents grant the value of compatibilist freedom. But they think that if God permits evil for the sake of incompatibilist freedom, what He gains is worthless.Footnote 4
Given the contentiousness of incompatibilism about freedom and determinism, then, the free-will defence appears vulnerable to attack. But instead of responding to this challenge by relying directly upon standard incompatibilist arguments (and potentially descending into the gridlock that typifies the free-will debate), it would be better if it could be argued that theological determinism is incompatible with freedom in a more dialectically useful way: namely, by beginning with varieties of theological determinism inconsistent with compatibilist, as well as libertarian, construals of freedom, and then arguing that the fundamental conflict between theological determinism and freedom as exemplified in these varieties survives in every manifestation of theological determinism.Footnote 5 That is what I aim to do here. My strategy is to present three versions of theological determinism, each of which is inconsistent with freedom on plausible, compatibilist-friendly principles, and then to argue that what makes them inconsistent with such freedom is entailed by any version of theological determinism.
Specifically, I will argue that in each case what generates the incompatibility is that every human action is the product of God's will (where ‘God's will’ and the related locution ‘God's willings’ are understood throughout to signify intentional actions on the part of God), together with a determination relation that holds between His will and what results as a necessary consequence of it, and that neither God's will nor the determination relation in question is sensitive to anything within the power of any human agent to do. This, of course, looks to be an incompatibilist analysis of the purest sort.Footnote 6 Yet, I shall argue, it is precisely the form of reasoning that underlies the compatibilist-friendly rejection of the versions of theological determinism we will begin by considering. If my analysis is correct, it will then be incumbent on the compatibilist to explain why this form of reasoning fails to apply to more garden-variety models of theological determinism, which may appear at first glance less problematic.
Theological determinism and continuous creation
The first version of theological determinism I wish to discuss owes to Jonathan Edwards (though its roots trace at least as far back as al-Ghazali),Footnote 7 and its presentation requires a brief digression into his ideas concerning divine governance. In keeping with traditional theism, Edwards sees God as playing two roles in relation to the world, that of the creator and that of the sustainer, or conserver, of all that exists. Not only did God bring into existence a bunch of contingent objects ‘in the beginning’, but He also, as it were, keeps the universe going in such a way that, if He were to cease conserving the universe at any time, the universe would at that time disappear. It is in this sense that all contingent things continually depend for their existence upon God.
It is easy to see that this doctrine is consistent with the idea that everything that comes to pass is determined by God. But Edwards incorporates the doctrine into his theory of theological determinism in a central way. On his picture, God determines the course of events by conserving the world in a law-like way, and He conserves the world and its contents by continuously creating them. As Edwards puts it, ‘God's upholding created substance, or causing its existence in each successive moment, is altogether equivalent to an immediate production out of nothing’, in a sense implying that the existence of a thing at a time depends in no ‘part, or degree’ on ‘its antecedent existence’.Footnote 8 Furthermore, in conserving the world, as in creating it, God is the total and exclusive cause not only of every object's existence, but also of every property instantiation or change any object undergoes. As we shall understand the terms here, to say that God is the total cause of an effect e is to say that God's will is sufficient for e and that no joint cause collaborates with God in bringing e about; and to say that God is the exclusive cause of e is to say that e is not causally overdetermined by God's will and any other potential cause.Footnote 9
Putting it all together, Edwards's theological determinism amounts to this: all created objects depend both for their existence and for the instantiation of whatever properties they exemplify on the immediate, total, and exclusive causal activity of God, and the lawful regularities in nature are expressions of God's faithfulness in ordering the world after a fashion that appears, sub specie temporalis, as though every event were causally necessitated by antecedent events. To complete the picture, we need to make explicit a final element Edwards leaves suppressed: it is not within the power of any human person to cause God to will what He does, and no human person has the power to frustrate the efficacy of the divine will.
The theory is elegant, but it immediately entails two untoward theses. The first is that there are no secondary causes in nature. The second is that human persons, like other created objects, fail to persist through time. I propose simply to shelve this latter difficulty and grant Philip Quinn's contention that ‘God can, so to speak, create an individual over and over again’, so that ‘contingent things once introduced into existence by God's creative power can also persist’.Footnote 10 Whether or not that's so is an intricate matter but it need not detain us, since the result that there are no secondary causes in nature follows solely from the totality and exclusivity assumptions.Footnote 11 And since the denial of secondary causation implies, as a special application, that we have no causal influence over what happens in the world – that our choices and intentions, e.g. are causally ineffective – it is this implication that is most immediately relevant to the question whether free will is consistent with this variety of theological determinism.
Without question, Edwards's vision of divine determination appears incompatible with any construal of human freedom.Footnote 12 To uncover precisely where the incompatibility lies we can outline its salient features as follows. Letting ‘G’ stand in for ‘God's willings’, and ‘D’ stand in for the determination relation, the present version of theological determinism encompasses the following states of affairs:
TD1
(1) There exists a set of conditions, G, such that a determination relation, D, holds between G and every state of affairs that obtains.
(2) For any agent S, it is not the case that S has the power to affect or causally influence either the obtaining of D or the obtaining of G.
(3) For any action A of S's, A is a total, exclusive and direct causal consequence of G and D.
Notice that condition (1) is entailed by any version of theological determinism, and according to a standard incompatibilist argument the conjunction of it with (2) is sufficient to rule out free will. For if (as condition (1) implies) all our actions are causally guaranteed by the conjunction of G and D, and if (as condition (2) says) we lack any sort of influence over whether G or D obtain, then we do not act freely, either because we have no available alternatives to any action or because our actions have their ultimate sources in something other than us. But presumably the compatibilist will want to reject this model on different grounds. Since the compatibilist holds that (1) and (2) may obtain consistently with human freedom he'll likely object to this model because of its third condition, for this reason:
(a) Condition (3) entails epiphenomenalism – that our minds are causally inert.
(b) Epiphenomenalism is inconsistent with free agency.
(c) Therefore, TD1 is inconsistent with free agency.
This argument seems significantly different from the incompatibilist's. But it is unsound as it stands, and the reconstruction required to render it sound reveals it to be quite similar in structure to the incompatibilist argument given above. The second premise is quite plausible.Footnote 13 What free agency requires, on anyone's understanding, is that we have an appropriate degree of influence over what happens within some limited sphere, that to some extent we can direct the course of events in virtue of what we think, believe and want. And insofar as mental causation is the best candidate for a mechanism through which such influence might be implemented, free agency is inconsistent with epiphenomenalism. So (b) should be accepted. But notice that (a) is false: what (3) entails is that no mental state of any human agent directly causes any of their actions, for inasmuch as any human action is immediately and exclusively caused by God, every other potential cause is blocked from contributing directly to its occurrence. But it does not follow from this that epiphenomenalism is true, nor does it follow that we lack any sort of influence over our behaviour: for it could be that our minds are causally relevant to whatever conditions do directly cause our behaviour, in which case the degree of influence required for freedom need not be jeopardized.
For example, suppose I have a headache and I want to get rid of it. I cannot simply will my headache out of existence – my desire to get rid of the headache won't directly cause it to disappear – but it doesn't follow that I have no influence over whether the state of affairs my having a headache obtains, because I can always alter the chemical state of my brain by taking some aspirin; and my desire to be rid of the headache can certainly be causally relevant to my doing that. Since I can influence the conditions which themselves determine whether or not my head hurts I have some measure of control over whether I have a headache; and I'm stuck with the headache, for as long as it lasts, only if I can do nothing about them. By analogy, I can influence my behaviour even if it is directly produced by God, as long as I can causally influence what God decides to do or whether God's willings will be effective; and I'm stuck with my behaviour, whatever it may be, only if I can do nothing about those things.
So to achieve the result that this version of determinism is inconsistent with freedom, the compatibilist, like the incompatibilist, must rely on condition (2) of TD1, and the argument for incompatibility will run like this:
(a) For any agent S, it is not the case that S has the power to affect or causally influence either the obtaining of D or G.[condition (2)]
(b) For any action A of S's, A is a total, exclusive and direct causal consequence of G & D.[condition (3)]
(c) (a) & (b) entail epiphenomenalism.
(d) Therefore, TD1 is inconsistent with free agency.
Our new argument makes essential reference to the assumption that we have no influence over the conditions that determine our actions, which brings its form one step closer to classical incompatibilist reasoning. Still, there is a clear difference between the two arguments. Whereas the incompatibilist thinks that determinism itself threatens our freedom, this version of theological determinism appears objectionable to the compatibilist only because it entails epiphenomenalism. But clearly not all varieties of theological determinism need have this consequence. Since the compatibilist thinks freedom is consistent with conditions (1) and (2), but not with the conjunction of (1), (2), and (3), he must believe there is a crucial freedom-relevant difference between deterministic worlds in which mental states are not causal and deterministic worlds in which they are; and this suggests that an appropriate alteration of the third condition would provide us with just what we need for freedom from a compatibilist point of view.
Now the problem with (3) was that it landed us in epiphenomenalism. And the most minimal alteration we can make to (3) to avoid this result is simply to suspend the specification that God's causal activity is exclusive, that if God directly causes an event, nothing else can cause it as well. So I propose we examine a model of theological determinism in which that offending assumption has been removed, with a view toward seeing whether it is consistent with free will. I will argue that it isn't for good, compatibilist-friendly reasons. Then I'll show how those compatibilist-friendly reasons once again mirror traditional incompatibilist arguments.
Suspending exclusivity
The second version of theological determinism is a close approximation to the first, and it bears interesting affinities to medieval Aristotelian and early modern concurrentism. According to it, God is still the sustaining cause of the universe and conservation is still a matter of continuous creation. Moreover, God remains the total and direct cause of everything that happens: everything He causes is caused by Him immediately and is fully brought about by Him. However, on this version, God's willings are not the exclusive cause of every event: some events are causally overdetermined by God's will and by the activities of human creatures.
Naturally, all overdetermining causes will have to display a precise harmony with what God wills: if God directly brings about an effect e, then for any event c that also causes e, it must not be possible for c to cause any distinct event e′ (where the occurrence of e′ is incompatible with the occurrence of e). But secondary causes may well be operative so long as they satisfy this constraint. And if our choices and intentions satisfy it, they may be operative too. As this model pertains to human action, then, the suggestion is that all of our actions are directly and totally caused by God in accordance with law-like patterns – that's the theological determinism – and, in addition, that certain of our own mental states may cause them as well – that's the overdetermination.Footnote 14
To make the elements explicit, our second model is as follows:
TD2
(1) There exists a set of conditions, G, such that a determination relation, D, holds between G and every state of affairs that obtains.
(2) For any agent S, it is not the case that S has the power to affect or causally influence either the obtaining of D or the obtaining of G.
(3′) For any action A of S's, A is a total and direct causal consequence of G and D.
Since (1) and (2) remain unaltered, the incompatibilist will object that TD2 conflicts with free will for familiar reasons. But since the exclusivity assumption has been jettisoned, (3′) is consistent with mental causation: our beliefs and desires may be among the elements that cause our actions, and the epiphenomenalist threat has been deposed.Footnote 15 Our question now is whether TD2 is plausibly consistent with freedom from a compatibilist perspective.
The suggestion that it is immediately appears suspect, but it's important to see why. One initial concern might be that we aren't really in control of our actions in this scenario. But I'm doubtful that charge can be sustained. Certainly, to be free with respect to an action is at least in part to have a sufficient degree of control over it. And a minimal compatibilist constraint has it that we possess the right degree of control if the processes culminating in our actions are sensitive to reasons in an appropriate way.Footnote 16 Roughly, if I perform an action for a given reason, and if it's at least possible for me to refrain from doing it in light of a sufficient reason to do something else – if I'm not acting under coercion or psychological compulsion, for instance – then it's sufficiently under my control. Only if conditions are in place that are sufficient to bring about that action regardless of the reasons I have for or against it does my action fail to be sensitive to reasons; but if I can regulate my behaviour in accordance with my reasons, it falls within the realm of my control.
Theological determinism as construed in this model is arguably consistent with such control. The suspicion that it is not is probably founded on the idea that even if an agent were apprised of a sufficient reason to refrain from performing an action A, God's willing that A should occur still would have brought that action about; and since it isn't possible for the agent to perform an action incompatible with A in those circumstances, the agent lacks sufficient control. But it isn't obvious that the counterfactual situation has to be evaluated that way. Plausibly, given God's reliable character, were He to produce a change in the circumstances leading up to an action, He would ensure suitable changes in the events that follow. Thus, were an agent to consider a different set of reasons for action, it is reasonable to think that the agent's subsequent behaviour would (or may) be modified accordingly. If that's right, then this version of determinism is no more inconsistent with compatibilist control than any other.
Still, the intuition remains that God's activity is ‘too close for comfort’. The real worry is not that we have no control over our behaviour but that it is in equal part being controlled by God, so that our actions are not clearly attributable to ourselves (rather than to Him). This isn't a specifically incompatibilist intuition. What the compatibilist claims is simply that our actions may be free and attributable to us even if determinism is true. But that (according to the vast majority of compatibilists) is a very different thing than our actions being under the direct causal influence of another agent, with purposes and intentions of their own, at every step. Indeed, much recent compatibilist work has been aimed at providing a principled distinction between ‘mere determinism’ on the one hand and direct intentional control by independent actors on the other, since there is considerable pressure in the latter case to conclude that the controlled agent's actions aren't really their own. As Dennett sums up the (near) consensus, ‘Only if foresighted, purposeful agents have been manipulating you for their own ends are you absolved of responsibility; those are not your deeds but your brainwashers’ deeds in such a case.'Footnote 17
Now I don't think this (or any other) model of theological determinism implies that we're being brainwashed by God. Importantly, there is no ‘short-circuiting’ or ‘bypassing’ of our considered values or faculties of judgement in this scenario. Nor does it involve our being forced to do things against our will, since God's control extends to our wills (reflections, evaluations) themselves. But whatever causal considerations might entitle us to claim ownership of our actions here are satisfied at least equally well by God, and whatever control we have over them is at least matched by His as well. According to a plausible compatibilist theory, in cases like these – which Alfred Mele dubs cases of ‘heteronomy’ – the agent being controlled lacks autonomy even if his actions are under his control, because he himself is unwittingly and non-consensually under the control of another agent; and people in such circumstances are neither free nor responsible for what they do.Footnote 18 So while this version of theological determinism does not imply that we lack (compatibilist) control or that we are subject to brainwashing and the like, its freedom-undermining feature is its implication that we act heteronomously.
These considerations suggest an argument to the effect that TD2 is inconsistent with freedom:
(a) An agent S is free with respect to an action A only if A is not a direct, intentional causal consequence of another agent's actions.Footnote 19
(b) For any action A of S's, A is a total and direct causal consequence of G and D.[condition (3′)]
(c) Therefore, for any action A of S's, S is unfree with respect to A.
Once again, our initial argument does not rely upon the assumption that we have no influence over the conditions that determine our behaviour, so it appears different from the incompatibilist's line of reasoning. But, as before, the argument is unsound; and the reformulation required to make it sound displays the same structural similarity to incompatibilist arguments that we noted above. The second premise, (b), is true by stipulation. But (a) is false, because it's absent the crucial specification that the subject of control (‘S’) himself has no influence over the agent whose activity is involved in the production of his behaviour.
To explain: what is concerning about the possibility that our behaviour is being controlled by another is the implicit assumption that her activities operate independently or irrespective of our own wishes and considered judgements, which introduces an important asymmetry.Footnote 20 Standard cases of brainwashing and hypnosis, for example, are cases in which ‘new pro-attitudes are not generated via an exercise or an activation of agents’ capacities for control over their mental lives; rather, they are generated despite the agents' capacities for this'.Footnote 21 But the external instilling of preferences, dispositions and so forth need not operate in this way. To take a simple case, suppose a person wants to quit smoking, finds it difficult to do so, and pays a third party to hypnotize him so as to eliminate his taste for tobacco. Maybe we think that the reformed smoker has ‘cheated’ in some way. But there is no reason to believe that he is not at least partially responsible for his newly acquired dispositions or that he does not act freely when he acts from them, since he himself arranged for the procedure and the outcome was proactively and retroactively endorsed.
So a qualification is in order. If one agent is controlled by another, or even manipulated in such a way as to bypass his ‘capacities for control over his mental life’, he fails to be responsible for the outcome only ‘provided that the bypassing was not itself arranged (or performed) by the manipulated person’.Footnote 22 The principle at work is simple and straightforward: if someone else directly and intentionally causes me to do something then I am neither free nor responsible for what I do – unless I have some influence over what they're doing. Now if that principle has at least a ring of plausibility even in such abnormal circumstances as hypnosis and brainwashing then it should be all the more plausible when we consider the version of theological determinism before us. God may be directly causing our behaviour at every step. But if God's causal activity is reliably responsive to our deliberations and decisions, and if He plays a normal role in the causal sequences that make up our actions (in a way that excludes such things as causal ‘deviance’ or ‘waywardness’), it is at least conceivable that we act freely.Footnote 23 What is hard to conceive is how we could act freely if we have no influence over what He does at all, even if our behaviour is intentional, responsive to reasons, and to that extent under our control.
Once again we find upon closer examination that a sound compatibilist argument for the inconsistency of freedom and TD2 must include the stipulation expressed in condition (2), that we cannot affect the conditions causally sufficient for our actions:
(a) An agent S is free with respect to an action A only if A is not a direct, intentional causal consequence of another agent's actions over which S has no influence.
(b) For any agent S, it is not the case that S has the power to affect or causally influence either the obtaining of D or the obtaining of G.[condition (2)]
(c) For any action A of S's, A is a total and direct causal consequence of G and D.[condition (3′)]
(d) Therefore, for any action A of S's, S is unfree with respect to A.
Our new argument relies essentially upon condition (2), just like the incompatibilist's argument does. Moreover, it highlights that we may lack compatibilist freedom even if our mental states are causally effective and even if we have a sufficient degree of control. Finally, the argument makes no mention of brainwashing, psychological compulsion, or other non-standard, independently freedom-cancelling features.
Now compatibilism does not object to our behaviour's being fully determined by conditions concerning which we have no choice, as per conditions (1) and (2). Since the typical compatibilist freedom-cancelling conditions are not entailed by this version of theological determinism, and since it is nevertheless inconsistent with freedom on plausible compatibilist principles, the conflict between freedom and TD2 must lie elsewhere. Specifically, it must lie in condition (3′).
Condition (3′) specifies that God's causal activity is both total and direct. To say that it is total is to say that God's willings are themselves sufficient, rather than either insufficient or partial in bringing about their effects. Any total cause will thus be a sufficient cause, but the operation of a sufficient cause need not exclude the possibility of independent causes for one of two reasons. First, as we have seen, an effect may be totally caused by God and also caused by something else if the effect is overdetermined. Second, an effect may be totally caused by God but only indirectly, by way of mediating causes. I have argued that the first option does not help to render theological determinism consistent with compatibilist freedom. But given the intuition that there is something objectionable about our actions being caused directly by another agent, it might be suggested that if the immediate effects of God's will were more distant (in space or time, say) we would secure a kind of protective buffer between our actions and His, and this would deliver enough ‘elbow room’ for free will. Accordingly, it might be thought that if condition (3′) were reformulated to exclude the assumption that God is the direct cause of every event, then, despite conditions (1) and (2), human agents might possess free will.
In the next section I explore this possibility. I will argue that it is a red herring: a slightly modified version of the last argument demonstrates that a buffer zone between God's intentional acts and their effects makes no principled difference to the compatibility of theological determinism and freedom. This slightly modified argument will once again mimic the incompatibilist's line of reasoning.
Suspending directness
Let us continue to work with models of theological determinism incorporating the doctrine of conservation as continuous creation. According to some present-day philosophers espousing continuous creation God brings it about only that every object exists when it does, but He is not the direct cause of every natural event that occurs. To bring out the contrast we may distinguish between what Jonathan Kvanvig and Hugh McCann call ‘productive’ and ‘alterational’ causation. God's productive causation amounts to His directly bringing it about that each object exists for as long as it does by means of continuous creation; but, as they put it, to ‘say this is not to say He is directly responsible for the states of things in the universe, or the changes they undergo’, since these ‘alterations’ may be brought about by natural causes without God's being immediately involved.Footnote 24
I do not think there is much content to this distinction. Specifically, I find it obscure how God could sustain the existence of an object without bringing about the properties or states essential to its persistence. But we may confer some intelligibility on this thesis if we combine medieval theology with a little contemporary physicalism. Suppose the physicalists are right that everything supervenes upon the physical, in the sense that the overall state of the world at a time is determined by and dependent upon the fundamental physical state of the world then.
Now if such supervenience obtains, it could be that God conserves entities by directly causing whatever physical properties they may instantiate (including those needed for life and persistence), but this does not entail that every property they instantiate is directly brought about by God. He need not directly produce any of the psychological states we experience, for example, because in directly bringing about the state of the physical world the mental features of the world are thereby guaranteed. And the same is to be said of biological, physiological, and whatever other supervenient states there may be. Details would of course need working out, but I find this combination surprisingly natural. (Perhaps not so surprising: the supervenience thesis is often glossed as saying ‘all God needed to do’ to create the world was to create the physical world since everything else, supervenient as it is upon the physical, is an ontological free lunch.)
So imagine a theory in which God determines the course of events by continuously creating entities in the way described. In so doing God does not directly cause the mental or physiological features an object exemplifies, if indeed it does, since the physical states He directly brings about are sufficient for that. So there is a buffer between the immediate activity of God and what gets determined as a result. And notice that consistently with this theory any event God causes indirectly, by directly bringing about another item that entails or otherwise necessitates it, may be independently caused by some other factor.Footnote 25 So it is conceivable that states which supervene upon physical processes are necessitated by the physical processes subvening them, and so only mediately by God, but also caused by antecedent events in a manner that involves no causal over-determination.Footnote 26
Once more, any such causes will have to work in exact agreement with whatever God's up to: if God brings about a state of affairs P such that P entails or otherwise necessitates e, then for any event c that causes e, it isn't possible for c to cause some other event e′ (where the occurrence of e′ is incompatible with e's occurring). As it relates to human action, an agent's choosing to A may supervene upon a particular physical process P, where P is directly brought about by God, but that choice may be independently caused by some prior state of the agent himself. And the same goes for any physiological or other higher-level bodily effects the agent's choice to A might cause.Footnote 27
Our new theory may then be outlined as follows:
TD3
(1) There exists a set of conditions, G, such that a determination relation, D, holds between G and every state of affairs that obtains.
(2) For any agent S, it is not the case that S has the power to affect or causally influence either the obtaining of D or the obtaining of G.
(3″) For any action A of S's, A is a total causal consequence of G and D.Footnote 28
The third condition no longer contains the stipulation that God's causal activity is invariably direct. Consequently TD3 does not entail that any of our choices or higher-level bodily effects are immediately produced by God, but merely that they are determined by God via mediating physical events.
Does our new theory allow that humans now act freely, in view of the modification introduced? A little reflection shows that whatever comfort this amendment was intended to provide is empty, if the argument of the last section has any force. What motivated the freedom-cancelling principle within that argument – that one isn't free if one's actions are ‘directly’ caused by another when one has no control over them – is certainly not that the external control at issue must be direct in any literal or precise metaphysical sense. After all, the standard stories in which agents are ‘directly’ controlled aren't cases of direct control in this sense at all: the hypnotist swings that funny watch by a string; the behavioural engineer steers you through some regimen, controlling you Walden-Two-style; the nefarious neurosurgeon implants that handy device in your head and activates it from a distance at will, and so on. In short, the controller in these stories controls you by doing something else, by causing a mediating sequence of events to occur.
So the worry does not concern whether the control is ‘direct’ or whether it's mediated by electronic devices. It is just that the control is there, that it's intentional, effective and, in particular, that it isn't within one's power to alter, affect or otherwise to change. Since these factors are not affected by the ‘directness’ assumption's suspension, whatever initial promise the buffer-zone intuition held out is erroneous, and the argument given above may be recast without loss of content to meet our present needs:
(a) An agent S is free with respect to an action A only if A is not an intentional causal consequence of another agent's actions over which S has no influence.
(b) For any agent S, it is not the case that S has the power to affect or causally influence either the obtaining of D or the obtaining of G.[condition (2)]
(c) For any action A of S's, A is a total causal consequence of G and D.[condition (3″)]
(d) Therefore, for any action A of S's, S is unfree with respect to A.
The conclusions of the last section thus apply mutatis mutandis to the model of theological determinism given here: incompatibility between such determinism and freedom does not hang crucially upon the spatial proximity of the immediate effects of God's will to the human actions which are determined as a result. What it does hang on is the truth of condition (2), that we have no power over the intentions or actions of God. In the next section I show how this result generalizes to theological determinism simpliciter.
Theological determinism and temporal distance
The last version of theological determinism makes no use of a clause indicating the distance between divine and human action. Moreover, it makes no special assumptions about the manner in which the world is conserved by God. Consistently with this theory God specified the natural laws, created the initial conditions and then, as Hume aptly put it, ‘bestowed motion on this immense machine’.Footnote 29 The remaining elements of our final theory, then, are these:
TD4
(1) There exists a set of conditions, G, such that a determination relation, D, holds between G and every state of affairs that obtains.
(2) For any agent S, it is not the case that S has the power to affect or causally influence either the obtaining of D or the obtaining of G.
Like the previous model this one allows for secondary causation without over-determination. Further, it accommodates any number of causes intervening between the immediate effects of God's will and the remote events necessitated by them. Last, it does not stipulate that the relation by means of which this necessitation is implemented is synchronic (as supervenience implies); we may take it to be wholly diachronic (as standard causal determinism implies) instead. It makes no difference: so far as the effectiveness of the relation goes it's neither here nor there whether ‘D’ operates in a ‘vertical’ or ‘horizontal’ fashion, since, as a formal matter, causal determination is strictly equivalent to supervenience turned on its ear.Footnote 30 Consequently, TD4 is compatible with a generous spatial and temporal buffer between God's immediate activity and any action any human undertakes.
Is TD4 therefore consistent with compatibilist freedom? It seems not, since the argument for incompatibility rehearsed in the previous section can be straightforwardly applied here. Recall that the initial premise of that argument, ‘An agent S is free with respect to an action A only if A is not an intentional causal consequence of another agent's actions over which S has no influence’, makes no specification at all concerning the time at which the controlling agent acts or the distance between his activities and those of the agent affected. Nor does it imply that the control at issue operates in a way that undermines the affected agent's psychological constitution. Still, the subjects of such control do not act freely on plausible compatibilist grounds.
But perhaps there are independent reasons to hold that compatibilist freedom can be insulated on this model, given the allowed temporal remoteness of God's activity. I can think of two such reasons, neither of which, unfortunately, is compelling. Consider first Harry Frankfurt's analysis of a similar pair of cases:
In the first state of affairs the [controller] manipulates his subject on a continuous basis … so that each of the subject's mental and physical states is the outcome of specific intervention on that part of the [controller]. In that case the subject is not a person at all. His history is utterly episodic and without inherent connectedness. Whatever identifiable themes it may reveal are not internally rooted; they cannot be understood as constituting or belonging to the subject's own nature. Rather, they are provided gratuitously by an agency external to the subject … . I believe, on the ground that since he lacks all autonomy, [his behaviour] cannot legitimately be ascribed to him as his actions.
The other possibility is that the [controller] provides his subject with a stable character or program, which he does not thereafter alter too frequently or at all, and that the subsequent mental and physical responses of the subject to his external and internal environments are determined by this program rather than by further intervention on the part of the [controller]. In that case [there are no] compelling reasons either against allowing that the subject may act freely or against regarding him as capable of being morally responsible for what he does.Footnote 31
Frankfurt's description introduces a false dichotomy. Nothing in the previous versions of theological determinism so much as hints that the continuous control portrayed therein is episodic, unintelligible, or inherently disconnected. To be sure, they do entail that any agent's actions are ‘provided gratuitously by an agency external to the subject’, and this does suggest that the agent's behaviour ‘cannot legitimately be ascribed to him’, but certainly not in view of the internalist reasons for which Frankfurt's analysis of freedom is famous. So while his attempt to distinguish these cases is eminently well-motivated, Frankfurt lacks the resources needed for the task.
If the intuition persists that temporal distance makes a difference, it is likely because we are accustomed to distinguishing between (i) intentional actions, as opposed to consequences for actions, and (ii) unintentional or unforeseen consequences, as opposed to consequences over which we exercise responsibility-conferring foresight and control. These considerations make quite a bit of sense when human action is at issue. After all, our actions may have any number of consequences that are not plausibly attributable to us, since the ripple effects of what we do may spread out in ways that we couldn't reasonably be expected to anticipate or that are strictly out of our hands. But such considerations are unduly anthropomorphic and consequently inapplicable here, as Hume rightly saw: ‘Ignorance or impotence may be pleaded for so limited a creature as man; but those imperfections have no place in our Creator. He foresaw, he ordained, he intended all those actions of men’ that are determined as a result of His activity. So even if, as our latest model allows, His intentions are executed through ‘a continued chain of necessary causes’, providentially arranged and stretching backward to ages long past, this implies neither lack of divine foresight nor lack of intentional control; and temporal distance notwithstanding, it remains as natural as ever to say, ‘While we act, we are, at the same time, acted upon.’Footnote 32
I conclude that the argument of the last section retains its force here: all our actions are caused by God's will and the determination relation that holds between it and its effects; it isn't within our power to influence God's will or the determination relation between it and our behaviour; since intentional control by another agent rules out freedom when we have no influence over what they do, theological determinism is inconsistent with freedom.
Remaining options
Does the foregoing amount to an argument for the ‘incompatibilist premise’ of the free-will defence? Yes and no. If such an argument requires an independent defence of incompatibilism about freedom and determinism generally, it does not. For my order of reasoning does not take the form: ‘Freedom requires alternative possibilities and determinism is inconsistent with alternative possibilities; therefore, since theological determinism is too, theological determinism is incompatible with free will.’ Rather, it is a direct argument for the incompatibility of freedom and theological determinism, modelled upon standard incompatibilist arguments, but constructed in a way that circumvents the traditional compatibility dispute.
Thus, when David Lewis (among others) insists that the free-will defence ‘must presuppose incompabilism’, and that ‘God could determine our choices via our characters, thereby preventing evil-doing while leaving’ a valuable, ‘compatibilist freedom in tact’, he's only half-right. Nothing in my argument requires me to denigrate the value of compatibilist freedom or to voice the conviction that libertarian freedom is more valuable still. Nor, so far as I can see, does the free-will defence require me to do so. If that is correct, the premise that freedom and theological determinism are incompatible does not beg the question against the compatibilist; nor is it a troubling point of weakness in the free-will defender's strategy to safeguard theism against the logical argument from evil.
Surprisingly, the conclusion is that the preservation of valuable freedom does indeed demand that our actions are not intentionally determined by God – not even via the device of standard causal determination – yet this conclusion does not presuppose libertarianism as the correct account of free will. Thus, we cannot assume, as Paul Helm contends, that the compatibility of freedom and theological determinism straightforwardly follows from the alleged compatibility of freedom with ‘secular determinism’, since the theistic context involves unique considerations that are strictly inapposite in the secular case.Footnote 33
But which options remain for the compatibilist who wishes to resist this result? It seems to me there are three, none of which is particularly attractive. The first option is simply to bite the bullet and endorse what Gary Watson and Robert Kane have termed ‘hard compatibilism’.Footnote 34 Hard compatibilists recognize that we may be unable to draw any principled distinction between ‘mere determinism’ and intentional manipulation or control, but they stick to their guns and insist that we may be just as free and responsible in the latter case as we are in the former. I suppose this option will always remain open. Like many, however, I think taking this route is not as much to uphold compatibilism as it is to reduce it to absurdity. And given the widespread dissatisfaction with internalist conceptions of freedom (such as Frankfurt's), it is unlikely that the majority of compatibilists will be content to maintain the compatibility of freedom and theological determinism on these grounds.
The second available option is to ‘go subjectivist’. That is, some compatibilists may wish to follow John Fischer and Mark Ravizza in incorporating a subjective component within their theory so that, provided certain conditions of reasons-responsiveness are met, an intentionally controlled agent may be morally responsible so long as she views herself as such. In this case, the compatibilist might contend that whereas an agent is not responsible if she believes God has determined her behaviour and that this implies a lack of responsibility on her part, other agents, though equally determined by God, may be responsible provided that they do not believe they are thus determined, or do not think such determinism conflicts with their responsibility. In my view this option is ultimately indistinguishable from the first. In particular, as I have argued elsewhere, this subjectivist manoeuvre is inconsistent with the intuitive requirements for moral responsibility which generally motivate philosophers to adopt it.Footnote 35
According to Fischer and Ravizza, for example, since ‘hard compatibilist’ theories do not deem historical considerations relevant to assessments of responsibility, they are subject to refutation on the grounds that an agent may fulfil the internal psychological conditions for freedom and responsibility even when they are subject to external control. In isolated cases of this kind, Fischer and Ravizza insist that such agents have not ‘taken responsibility’ for the ‘mechanisms’ productive of their behaviour – i.e. they do not genuinely view themselves as responsible when acting under manipulative influence – because they do not know about or have not explicitly considered the possibility that they are being controlled, so they fail to be responsible for what they do.Footnote 36 However, as they realize, an agent may be externally caused to accept responsibility or manipulatively induced to view herself as an appropriate target of praise and blame. And in this case they contend that, if the manipulative influence is sufficiently ‘global’ or comprehensive, a person can be completely and constantly controlled by another without failing to be responsible for what she does!Footnote 37 Clearly, this result is no more palatable in a subjectivist setting than it was above: the same bullet is bitten, and the second option collapses into the hard compatibilism it was designed to avoid.
The final option relies upon the possibility that considerations leading us to deny the freedom or responsibility of those who are controlled by human agents fail to apply to divine control. The merits of this response are, of course, hard to evaluate in the absence of a reasonably articulated description of that in which the relevant difference between divine and human control is supposed to consist. No doubt the manner and guaranteed success of these kinds of control differ, and the intentions of God – benevolent as we may assume them to be – may differ from those of human controllers as well. But these factors do nothing to show how there is a crucial difference in the effects of these kinds of control, nor do they explain why in the one case our freedom is violated while in the other it is preserved. I do not claim that no freedom-relevant difference between these kinds of control could ever be produced; but I think it fair to say that no such difference has so far been provided. So while I do not reject this possibility out of hand I think it must be viewed as (at best) a promissory note that has not yet been discharged, but which needs to be if the compatibilist who contests the free-will defence for this reason is to avoid the charge of special pleading.
