From Modal Collapse to Providential Collapse

Abstract

The modal collapse objection to classical theism has received significant attention among philosophers as of late. My aim in this paper is to advance this blossoming debate. First, I briefly survey the modal collapse literature and argue that classical theists avoid modal collapse if and only if they embrace an indeterministic link between God and his effects. Second, I argue that this indeterminism poses two challenges to classical theism. The first challenge is that it collapses God’s status as an intentional agent who knows and intends what he is bringing about in advance. The second challenge is that it collapses God’s providential control over which creation obtains.

1 Introduction

Traditional articulations of classical theism like those of Augustine, Aquinas, and Anselm affirm the Doctrine of Divine Simplicity (DDS). According to DDS, God is devoid of physical, metaphysical, and logical parts. He is identical to his essence, existence, attributes, action, power, and so on. On DDS, the distinctions that entail composition are (inter alia) those between (i) essence and existence, (ii) subject and accidents, (iii) individual and its essence, (iv) individual and its properties, (v) act and potency, and (vi) agent and the agent’s actions.¹ As Katherin Rogers summarizes the traditional doctrine, DDS “denies that God has any properties at all. God is an act... an eternal, immutable, absolutely simple act. … God simply is an act, and all the words we use to describe God refer to this act” (1996, p. 166).

In debates about DDS, x is part of S just in case x is a positive ontological item intrinsic to but distinct from S.² More simply (and applied to God), anything intrinsic to God is identical to God (Fakhri, 2021). As Augustine famously put it, God is what he has (Augustine, The City of God, XI, 10). Vallicella (2019) follows suit: “God is ontologically simple… there is nothing intrinsic to God that is distinct from God.” Other scholars are similarly explicit about this conception of parthood in relation to DDS.³ In motto form: whatever is in God is God.

A popular objection to DDS is the modal collapse objection. If God is identical to God’s act of creation (as is demanded by the identity claims in DDS), then it seems to follow that God necessarily creates. God, after all, is a necessary being, and thus whatever is identical to God is likewise necessary. And since God’s act of creation with which God is identical extends to any positive ontological item distinct from God, it would seem to follow that everything is necessary. All beings are necessary beings, and all truths are necessary truths. Modal categories collapse into metaphysical necessity.⁴

A flurry of recent articles have been published on modal collapse arguments, and my aim in this paper is to advance the debate. To accomplish this, I first briefly survey the modal collapse literature and argue that classical theists avoid modal collapse if and only if they embrace an indeterministic link between God and his possible effects (§2). Second, I argue that this indeterminism poses two challenges to classical theism. The first challenge is that it collapses God’s status as an intentional agent who intends what he is bringing about in advance (3). The second challenge is that it collapses God’s providential control over which precise creation obtains (§4). Let’s proceed to §2.

2 Modal Collapse: Survey and Solution

My present concern is action-based modal collapse arguments—that is, arguments that purport to deliver the absurdly fatalistic conclusion from the identity of God and God’s act(s).⁵ In §2.1, I briefly survey extant modal collapse arguments and their responses. In §2.2, I propose a biconditional solution to modal collapse arguments: classical theists avoid modal collapse if and only if they embrace an indeterministic link between God and his effects. This biconditional sets the stage for the two challenges accruing to classical theism developed in §§ 3 and 4.

2.1 Survey

Modal collapse arguments are found in a several authors, including Mullins (2013, 2016, p. 138, 2021), Leftow (2015, p. 48), Moreland and Craig (2003, p. 525), and Waldrop (2021). For simplicity, I will survey three.

The first is a simple, intersubstitution version:

• 1. Necessarily, God exists.

• 2. God is identical to God’s creative act.

• 3. Necessarily, God’s creative act exists.

As Tomaszewski (2019) points out, the problem with this argument is that it rests on an invalid substitution of a non-rigid singular term (‘God’s creative act’) for a rigid proper name (‘God’) in a referentially opaque or intensional context. Tomaszewski surveys and rejects various ways to repair the simple intersubstitution version.

A second version is found in Mullins (2021, pp. 94–95). For purposes of space, I won’t quote the argument here. In summary: Under DDS, God is identical to his intentional act to actualize this world. Since God exists necessarily, whatever is identical to God likewise exists necessarily. So, God’s intentional act to actualize this world exists necessarily. But if God’s intentional act to actualize this world exists necessarily, then this world exists necessarily, since an omnipotent being’s intention(s) cannot fail to issue in the obtaining of its object(s). So, this world exists necessarily.

The problem with this version of the argument is that if the link between God’s act and creation is indeterministic, then the necessity of creation does not follow from the (de re) necessity of God’s intentional act to actualize this world. But the classical theist is well within their epistemic rights—at least in the dialectical context at hand—in holding that such a link is indeterministic. Nothing in DDS in particular or classical theism in general demands a deterministic causal link between God and creation.⁶

The third version is found in Waldrop (2021). This version adds a metaphysically substantive essentiality thesis, (E), about the nature of divine action:

• 4. Necessarily, God exists.

• 5. God is identical to the divine creative act.

• 6. Necessarily, something is a divine creative act only if it is essentially the unique divine creative act. (E)

• 7. Necessarily, the unique divine creative act exits.

This argument is valid, and (7) rather swiftly entails modal collapse. The classical theist, of course, will reject thesis (E). For if God’s act merely indeterministically produces its effects, then it is false that God’s act—if it is the divine creative act—is essentially the unique divine creative act. Under such indeterminism, it is only accidental to God’s act that it is the unique divine creative act.

What about general responses to modal collapse arguments? Nemes (2020) argues that modal collapse arguments presuppose (what he terms) the difference principle, according to which any difference in effect (across worlds) presupposes a difference in cause (across worlds). But for Nemes, a classical theist is well within their epistemic rights in simply rejecting such a presupposition. Fakhri (2021) develops a similar line of thought but uses it to defend (a version of) the modal collapse argument.

Much more can be said, but that suffices for purposes of a brief summary. Let’s now consider my proposed solution to modal collapse arguments.⁷

2.2 Solution

Here’s my solution to modal collapse arguments:

Biconditional Solution: Classical theists avoid modal collapse if and only if they embrace an indeterministic link between God and his effects.

It’s clear that indeterministic causation is necessary to avoid modal collapse under classical theism. For the cause of everything apart from God, under classical theism, is God himself. (There is no act of creation distinct from God that mediates between God and creation.) And God is necessary. And by the distribution axiom, whatever is necessitated by something necessary is itself necessary. Since A’s deterministically causing B is a matter of A’s necessitating B—that is, since A’s deterministically causing B entails that necessarily, if A, then B—it follows that modal collapse ensues from a deterministic causal link between God and his effect(s). Thus, an indeterministic causal link here is necessary to avoid modal collapse under classical theism.

It is also sufficient. For if God indeterministically causes his effect(s), then (i) ‘God’s creative act’ only non-rigidly designates God, (ii) the necessity of creation does not follow from the (de re) necessity of God’s intentional act to actualize this world, and (iii) thesis (E) is false. And in that case, each of the three modal collapse arguments surveyed in the previous sub-section don’t work. More generally, so long as God only indeterministically causes his effect(s), God can be cross-world invariant, necessarily existent, and identical to his single act while creation is cross-world variant and contingent. This is part and parcel of indeterministic causation—the entity that is in fact the cause of E can exist without E. Hence, such an entity can be necessary while E is contingent, and moreover such an entity can be cross-world invariant while E is cross-world variant (e.g., existing in one world and not existing in another).

I take it, then, that Biconditional Solution is true.⁸ This is significant: not only can classical theists avert modal collapse by accepting <God indeterministically causes his effect(s)>, but also—if they are to avert modal collapse—they must accept this. Biconditional Solution is also significant because it undergirds two new problems for classical theism. It is to the first of these problems that I turn next.

3 Intentional Collapse

One thing we learn from the modal collapse debate is that contingent predications of God’s act(s) are extrinsic—they are not true of God in virtue of how God is in himself but instead in virtue things ad extra. Thus, Tomaszewski writes:

While God’s act is indeed intrinsic (and therefore identical) to Him, ‘God’s act of creation’ designates that act, not how it is in itself, but by way of its contingent effects. That is, whether ‘God’s act of creation’ designates God’s act depends on the existence of a creation which is contingent, and so the designation is not rigid. And since the designation is not rigid, the identity statement is not necessary, as it must be in order to validate the argument from modal collapse. (2019, p. 280)

Here, Tomaszewski points out that God’s act counts as creative act not in virtue of anything about the act itself but instead in virtue of creation’s obtaining. This point generalizes to other contingent predications of God, and there’s a rather straightforward reason for this. The reason is that everything intrinsic to God under DDS is identical to God, and hence everything in God is necessary. There is thus nothing about God as he is in himself that is contingent. God is so thoroughly uninfected by contingency that it is located entirely on the side of creation. Contingent predications of God, then, are true (at least in part) in virtue of facts about creation. Everything intrinsic to God is necessary, and so intrinsic predications of God are likewise necessary. Contingent predications of God, then, are extrinsic predications.

Another reason contingent divine predications are extrinsic is that God—under traditional articulations of DDS—is purely actual. He has no potential for change or for cross-world variance. Thus, everything about God as he is in himself is utterly invariant across worlds. (Otherwise, God would have potential to cross-world vary.) This debars contingent intrinsic predications of God, since a contingent intrinsic predication of God would entail that something about God as he is in himself is not invariant across worlds. (By definition, contingent is not cross-world invariant.)

Biconditional Solution bears this out, too. The link between God and any possible creation is indeterministic, and so God remains wholly self-same across all worlds despite different creations obtaining. The only variance is on the part of creation. Naturally, then, predications of God that vary across worlds track variations on the side of creation, not variations on God’s end. And this is just to say that contingent predications of God, under DDS, are uniformly extrinsic—they are true in virtue of things ad extra.⁹

But now a problem seems to arise. For God is an intentional, rational agent (even if only analogously so). Surely, then, God knows and intends what he is doing in advance. It is not as though God brings something about but doesn’t know or intend in advance what he is doing, i.e., what he is bringing about. The notion of ‘in advance’ here is a bit imprecise, but we can precisify it by speaking of the state causally prior to creation. As Brian Leftow explains the doctrine of creation within the classical tradition, “before all else existed, God existed, alone, or God and only God did not begin to exist” (Leftow 2012, p. 4). Leftow (2009) reiterates this commitment, in the context of Aquinas, that there is “the initial state of things which is God alone, causally prior to creating” (2009, p. 38). Leftow also writes that “we can… depict all possible worlds as trees of alternate possibilities branching out of a causally first state of things, which is God actually existing alone, causally though not temporally prior to creating” (2009, p. 25). Other authors are similarly explicit that the classical theist’s affirmation of creation ex nihilo involves a state causally prior to creation in which God alone exists.¹⁰ Call this state PRIOR.

Thus, causally prior to creation—in PRIOR—God knows and intends what he will create.¹¹ God is not ignorant of what he does in advance of doing it, and he does not create unintentionally. Otherwise, God would not be a perfectly rational, intentional agent. As Ward explains:

If such a God is personal, then his actions are personal actions, and this means they are intentional and free; and if such a God is perfect in character, then his actions are rational and good. These guideposts for reflection on the doctrine of creation force us to think of God as knowing what he is doing when he creates. When he says, for example, “Let there be light (Gen. 1:3),” he does not discover what light is when it comes into existence. He meant light. And if he meant it, then he knew about light before he spoke it into being.

The precise sense of ‘before’ is difficult to pin down; minimally, it is an explanatory or logical ‘before’. If the world is a product of God’s rational action then when God makes light he makes it, in part, because he knows about light, as when we say that the child aced the test because she knew her multiplication table. (2020, p. 5)

Plausibly, then, God knows and intends what he will create in advance of creating it, and his knowing and intending it is part of the explanation of why it comes about.

This is where the problem manifests. For God’s knowing and intending what he will create is a contingent matter. It is a contingent truth that God actualizes our world, and hence it is a contingent truth that God knows that he will actualize our world in advance of creating it. Similarly, it is only a contingent truth that God intends to actualize our world. Had another world come about, it would have been the case that God intended to actualize that world instead. (To say it is necessarily true that God intends to actualize our world is to grant modal collapse, since an infallibly omnipotent being intending to bring something about is sufficient for its coming about.) But we saw earlier that contingent divine predications are extrinsic, i.e., they characterize God not as he is in himself but rather as he connects (or fails to connect) to things ad extra. To put it differently, contingent divine predications depend on something apart from God.

But upon what could the contingent predication of ‘intending to actualize our world’ in PRIOR depend? What explains the truth of the predication in PRIOR? It can’t be God (alone), for this is an extrinsic predication. To propose that there is nothing upon which the predication depends strikes me as deeply implausible. Consider two worlds, each of which has its respective PRIOR state. Now, in one such world God intends one creation, whereas in another such world, God intends another creation. The proposal at hand renders the difference between these worlds utterly inexplicable. The difference seems to amount to magic. In each world, it just happens to be true that God intends the creation that comes about therein. Thus, surely something explains its truth. Could it be the absence of our world? Hardly—in every other possible world, there’s an absence of our world, and yet the relevant prediction doesn’t characterize God. (In itself, moreover, the proposal is facially implausible.) Could it be our world’s coming about? This is the only plausible candidate. But there is a serious problem with this proposal: it is viciously circular.¹²

For God’s intending and knowing our world in advance is surely part of the explanation of our world’s coming about. It is precisely because God intends to actualize our world and knows he will actualize it that our world comes about. But the above proposal flips the order of explanation. Under that proposal, it is because our world comes about that God intends to actualize our world (and knows he will actualize it). We therefore have a vicious explanatory circle on our hands: our world comes about because God intends (and knows) our world; but God’s act is one that intends (and knows) our world because our world comes about.

Upon reflection, this is entirely to be expected. Return to Tomaszewski’s point that that ‘God’s act of creation’ designates God not how he is in himself but instead by way of (i.e., by virtue of, by means of, because of) God’s contingent effects. By the same token, ‘God’s act of intending to actualize (and knowing that he will actualize) our world’ designates God not how he is in himself but instead by way of its contingent effect (viz. our world). Thus, God’s being an act to intentionally actualize our world is parasitic on and posterior to our world’s coming about. There is nothing about the act itself (and hence nothing prior to creation) that makes it an act to intentionally actualize our world. Its status as such is parasitic on the effect of that act (i.e., our world’s coming about). And this, we’ve seen, lands the classical theist in a problematic vicious circularity. We could term this vicious circularity the intentional collapse argument against classical theism.

Leftow (2009) considers a similar circularity worry as applied to God’s act of will within Aquinas’ model of God. Of particular relevance at this juncture is an objection Leftow considers:

One might reply here by distinguishing the senses of ‘because’: in ‘there are creatures because God wills there to be,’ ‘because’ expresses something efficient-causal, while in ‘God wills there to be creatures because there are creatures’ it expresses a non-causal relation in virtue of which a predicate applies. (Ibid, p. 33)

Simply replace ‘there are creatures’ with ‘our world comes about’ and ‘God wills there to be (creatures)’ with my ‘God intends (and knows) our world in advance of creating it’, and the objection equally targets my intentional collapse argument. Here’s how Leftow responds to the objection:

But it’s not clear that these two explanations really cohere. If God’s causation accounts for creatures’ existence, then logically before the creatures exist, God’s volition has a character sufficient to account for their existence. If it does, then at that point and for that reason it is a willing of creatures: there is nothing left for an extrinsic relation to the creatures themselves to explain. (Ibid, p. 33)

By my lights, this response is eminently plausible. For God’s act to constitute an intentional, knowledge-governed act to actualize this world, surely the intentional directedness toward its specific contingent effect is on the side of the act itself. It is surely not grounded in or explained by the relevant effect, precisely because the intentional directedness is prior to and accounts for the effect in question. And for this reason, it is surely the character of the act itself which grounds or makes true its intentional directedness toward the relevant effect. And yet—as we’ve seen—this is debarred by DDS, for the intentional directedness in question is only contingent (and hence extrinsic). We could formalize this line of reasoning like so:

• 8. God’s act in itself has a character adequate to provide an intentional explanation for creation’s obtaining.

• 9. God’s act in itself has a character adequate to provide such an intentional explanation only if God’s act in itself counts as an act of intending creation.

• 10. If God’s act in itself counts as an act of intending creation, then the contingent predication of intending creation to God is not extrinsic.1111

• 11. If the contingent predication of intending creation to God is not extrinsic, then DDS is false.12

• 12. So, DDS is false. (8–11)

I certainly don’t claim that this argument is insuperable. Per usual, there is substantial room for reasonable disagreement. By my lights, though—and as I hope to have shown throughout this section—each premise is deeply plausible. I offer the argument as a tool to advance modal collapse debates and serve participants thereof. Let’s now consider my article’s second challenge: the providential collapse argument.

4 Providential Collapse

As Biconditional Solution teaches us, to avoid the modal collapse objection classical theists must adopt a kind of radical indeterminism between God’s single act (on the one hand) and its various effects across worlds (on the other). Given that everything solely about God is fixed across all such worlds, the kind of indeterminism the classical theist must postulate is, indeed, quite radical: even fixing absolutely everything about God, any possible effect whatsoever can come about. The providential collapse argument reasons from this indeterminism to the denial of divine providence (i.e., God’s control over (i) whether creation obtains as well as (ii) creation’s precise contents if it does obtain). I develop the argument in §4.1 and assess objections in §4.2.

4.1 The Argument

The argument begins with the observation that if God is numerically identical to a single act across all worlds, then there is no distinctive willing of this particular creation to obtain, and nor is there any distinctive, intentional exercise of power to bring creation into being. As O’Connor points out, God’s willing that this world exist “will not be or involve a distinctive intrinsic state of God” (1999, p. 410).

Intuitively, though, creation’s precise contents then seem beyond God’s control, since fixing all the facts about God leaves open all possible worlds—from worlds in which an infinite multiverse co-obtains with God to worlds in which God exists by himself. Each of them could become actual, and God can do nothing distinctive in any such world to ensure any particular creation obtains.¹³

To draw out this intuition, consider the following analogy. Suppose that the temperature of a room can be any non-negative number—from 0 to π to 412.2 to 10^googol and so on. Suppose, moreover, that no matter what facts about you obtain—your actions, intentions, desires, bodily states and movements, mental states, and the like—none of these facts specify any particular value or even any subset of values among this infinite array of possible temperatures to be actualized. In any situation, everything about you—including your mental intentions, mental willings, and bodily actions—leaves perfectly open which of the infinitely many room temperatures becomes actual. I now ask: do you have control over the room’s precise temperature? I think the answer is obviously no. No matter what you do—no matter how you move your hands, exert your will, and whatnot—the temperature could still be any non-negative number.

But now compare this case with the classical theistic God: no matter what facts about God and what is within God obtain (e.g., any (intrinsic) desires, intentions, willings, actions, etc., all of which are numerically identical to God), none of these facts specify or determine any particular possible world to obtain or even any subset of possible worlds among the infinite array of such worlds. God just does something (which is the same as him just existing), and from this act some possible world or other is indeterministically actualized. But if one or another gets actualized, it won’t be due to anything different in God or in what God did (i.e., in God’s action). It’s hard to see how this is relevantly different from the temperature case, or from a case in which (i) you push the button on a lottery machine, and (ii) this causes the machine to randomly pick some numbers. In such a situation, you can’t control which set of numbers gets picked, even though you’re causally responsible for some number or other getting picked. It just seems self-evident that you’re not in control over which number gets picked. You perform one and the same action (which is the only one you can perform), and it can bring about any one of an arbitrarily large number of effects. It seems clear that you don’t have control over which effect comes about. And yet the same is true of God: God performs one and the same action (which is the only one he can perform, for he is numerically identical with it), and it can bring about any one of an arbitrarily large number of effects (indeed, infinitely many). Even though God is causally responsible for some creation or other coming about, it seems God isn’t in control over which creation gets picked.¹⁴

Thus far, I’ve only given an informal sketch of the argument. To make matters precise, I wish to formalize it. Before doing so, we need to get clear about the terms employed therein. By ‘facts about an agent and their act(s)’, I mean the facts about how the agent and their acts are in themselves. It includes things like internal mental willings, intentions, desires, the character of the act(s), the character and states of the agent, and so on. By ‘x is perfectly compatible with y’s obtaining’, I mean x leaves open whether y obtains; y’s obtaining is a live possibility on the supposition that x obtains—x’s obtaining does not ensure, settle, or determine ~y. By ‘divine effect’, I mean whatever effect of God’s one necessary act obtains. We can then formalize the providential collapse argument as follows:

• 13. If fixing all the facts about an agent and their act(s) is perfectly compatible with the obtaining of any possible effect of their act(s) among an arbitrarily large range of possible effects, then the agent is not in control over which effect of their act(s) obtains.¹⁵

• 14. If DDS is true, then fixing all the facts about God and his act is perfectly compatible with the obtaining of any possible divine effect among an arbitrarily large range of possible divine effects.

• 15. So, if DDS is true, God is not in control over which divine effect obtains. (13, 14)

• 16. But since God is provident, God is in control over which divine effect obtains.

• 17. So, DDS is false. (15, 16)

Premises (14) and (16) are relatively uncontroversial. The crux of the argument, then, is premise (13). To motivate it, I appeaed to (i) its eminent intuitive plausibility and (ii) the relevant similarity between (a) the classical theistic God and his effect(s) and (b) the link between an agent and their effect(s) in the temperature and button-lottery cases.

Like the intentional collapse argument, I do not claim that this providential collapse argument is insuperable, or that there is no room for reasonable disagreement about its soundness. Once more, I offer the argument as a tool to advance modal collapse debates and serve participants thereof. In the following section, I will assess objections to the argument.

4.2 Objections and Replies

Objection One

Don’t non-classical theistic models of God face the same (or an exactly and relevantly similar) problem? If so, then the argument is no mark against DDS in particular vis-à-vis non-classical theistic models of God. For non-classical theism grants that there are necessary features of God (e.g., God’s nature, goodness, perfection, reasons, and so on) and contingent features of God (e.g., God’s contingent intention(s) and act of will). Call the former N and the latter C. Now, non-classical theists grant that there is some dependence of C on N. It’s not as though C inexplicably floats free (as it were) from N. For instance, God’s contingent act of will is partly dependent on more fundamental, necessary features of God (e.g., God’s goodness). But in that case, we have a problem of control relevantly similar to premise (13). In particular, fixing all the facts about N (as it is in itself) is perfectly compatible with the obtaining of any possible C among an arbitrarily large range of possible C’s. How, then, is God in control over which C obtains given (13)?

Reply

I have three replies.

First, this tu quoque objection—while valuable—does not actually constitute an objection to my argument. Instead, it only constitutes an objection to the claim that classical theism faces some distinctive problem of control not faced by non-classical theism. But that claim is neither part of nor implied by my argument. Even if successful, then, the objection doesn’t threaten the argument.

But—and this is my second reply—I wouldn’t grant that the objection is successful in the first place. Recall (13): If fixing all the facts about an agent and their act(s) is perfectly compatible with the obtaining of any possible effect of their act(s) among an arbitrarily large range of possible effects, then the agent is not in control over which effect of their act(s) obtains. To have a truly symmetrical principle, the objector would have to replace each instance of ‘agent and their act(s)’ with ‘N’ and each instance of ‘effect’ with ‘C’:

13*. If fixing all the facts about N is perfectly compatible with the obtaining of any possible C (arising from N) among an arbitrarily large range of possible C’s, then N is not in control over which C obtains.

But the non-classical theist, I say, should embrace (13*) with open arms. Nothing in (13*) entails that God is not in control over which C obtains. It only entails that N is not in control over which C obtains. And God can be in control over which contingent act he performs even though the necessary features of God are not. Moreover, (13*) seems eminently plausible for precisely the reasons (13) did. I don’t think the objection at hand, then, should worry the non-classical theist.

Third, nothing in (13) by itself presents a problem either for human libertarian freedom (if we have it) or non-classical divine libertarian freedom (if God has it). For when we fix all the facts about a human agent (e.g., you or me) and a given act (or set of acts) of theirs, it is false that this is perfectly compatible with the obtaining of any effect whatsoever from their actions, willings, intentions, and so on. Instead, their actions, willings, intentions, and so on are almost always intrinsically directed toward specific states of affairs and can typically bring about only a small range of effects (oftentimes only one) among all possible effects of us as agents. (Remember, we’re taking about fixing all the facts about the agent and their act(s).) In fact, when indeterminism is introduced between our act(s) and the extrinsic effects thereof, surely our control is diminished. Suppose you’re driving along the road and no matter how you turn your hands and manipulate the steering wheel, the wheels indeterministically swivel left and right jaggedly and unpredictably. This surely diminishes your control over the car’s trajectory, precisely because the indeterminism is located entirely downstream of you and your acts.¹⁶

This point similarly holds for non-classical theism. Fixing all the facts about the non-classical theistic God is only compatible with one effect obtaining. For the non-classical God is infallibly omnipotent, and moreover the non-classical God’s intending and willing this creation is neither (partly) constituted by nor (partly) grounded this creation itself; instead, there is something about God’s action which is intrinsically directed toward the production of this creation. In other words, non-classical theists reject the extrinsicality of divine predications of willing and intending creation. Thus, if we fix all the facts about God, one of those facts will be the fact that God willed this universe. And this fact isn’t compatible with any effect other than this universe. (By contrast, the fact that God willed this universe, under classical theism, is not a fact about God as he is in himself; it’s a fact about the universe itself coming into being with a dependence relation on God.) Thus, (13) threatens neither human nor non-classical divine freedom.

For these three reasons, I conclude that Objection One doesn’t work. Note that I will say more about the asymmetry between non-classical and classical theism in response to Objection Two. What I’ve said here, though, suffices for present purposes.

Objection Two

The providential collapse argument seems to be a version of—or, at least, seems relevantly similar to—the luck objection to libertarianism. As Clarke and Capes (2017) summarize the objection, “[i]f a decision is nondeterministically caused… [t]here is… nothing about the agent prior to the decision—indeed, there is nothing about the world prior to that time—that accounts for the difference between her making one decision and her making the other. This difference, then, is just a matter of luck.” And if this is a matter of luck, responsibility for the decision seems undermined.

But the literature on the luck objection is substantive, and libertarians have proffered a variety of responses that are overlooked in the providential collapse argument.¹⁷ Without addressing such responses, the providential collapse argument is either underdeveloped or inadequately motivated.

Reply

This is a valuable objection. I have three replies.

First, I don’t think it’s plausible that my argument is a version of the luck objection. For the luck objection, if successful, rules out libertarianism. But the providential collapse argument, if successful, does not rule out libertarianism. As I explained in the third response to Objection One, nothing in (13)—and, by extension, nothing in the other premises of my argument—presents a problem either for human libertarian freedom or non-classical divine libertarian freedom.

But perhaps the idea behind Objection Two is that the motivations for (13), if applied consistently, would equally rule out libertarian accounts of freedom. Once more, though, this doesn’t seem plausible. Surely both proponents and opponents of libertarianism can unite behind the deeply intuitively plausible claim that in the temperature, button-lottery, and steering wheel cases from earlier, the agent is not in control over which precise effect comes about. In such cases, the indeterminism is problematically located downstream of everything about the agent and their action(s).

That, then, is my first response to Objection Two: it is not true that the providential collapse argument is a version of the luck objection to libertarianism, since the former (but not the latter) is entirely compatible with libertarianism.

My second response builds off the previous one, focusing in particular on the locus of indeterminism. Consider that by many libertarians’ own lights, indeterminism located downstream of our decisions, decision-making process, and actions does undermine control. As Levy (2005, p. 53) explains:

The demand for contrastive explanations of actions stems from the oft-expressed suspicion that indeterminism would not enhance freedom; it would merely introduce an element of randomness into human action. Certainly, and as libertarians themselves recognize, indeterminacy in the wrong place undermines control. Any indeterminacy ‘downstream’ of a decision would seem to reduce, rather than enhance, freedom. If it was indeterminate whether our decisions would result in our coming to have appropriate intentions, or our intentions in appropriate actions, then we would be less free than otherwise: we would sometimes be prevented from implementing our decisions, by indeterministic processes which intrude upon our actions from outside ourselves. Libertarians therefore locate the indeterminacy they claim to be required by free will upstream of decision, or in the decision-making process itself. Indeterminacy here, and here alone, enhances freedom, they claim.

Libertarians like Clarke agree. Clarke (2002, pp. 367–368) writes:

[W]e need to distinguish the following two significantly different varieties of case: those in which there is indeterminism between a basic action and an intended result that is not itself an action, and those… in which the indeterminism is in the production of a basic action itself. For the first sort of case, suppose that you throw a ball attempting to hit a target, which you succeed in doing. The ball’s striking the target is not itself an action, and you exercise control over this event only by way of your prior action of throwing the ball. Now suppose that, due to certain properties of the ball and the wind, the process between your releasing the ball and its striking the target is indeterministic. Indeterminism located here inhibits your success at bringing about a nonactive result that you were (freely, we may suppose) trying to bring about, and for this reason it clearly does diminish your control over the result—it constitutes control-diminishing luck.

As can be seen in the above quotations, libertarians make precisely the point I’ve made in this article about the control-diminishing nature of indeterminism downstream of one’s act(s). This is significant for two reasons. First, it bolsters my first reply to Objection Two (namely, that my providential collapse argument is not simply a version of the luck objection to libertarianism). Second, and more importantly, it constitutes a reply to Objection Two in its own right. For under classical theism (but not under non-classical theism and (typically) not in human free action), the indeterminism is located precisely in the control-diminishing locus: downstream of God’s act (and, indeed, downstream of everything about God as he is in himself). The classical theistic God seems relevantly analogous to the ball-target case from Clarke: The throwing of the ball is like God’s act (with which God is identical). But posterior to the throwing, the ball could indeterministically go anywhere on the target. And precisely because of this, you are not in control over which precise position on the target it lands. Similarly, posterior to God’s one, absolutely simple act, any possible effect (and, hence, any possible world) could indeterministically arise. And precisely because of this, God is plausibly not in control over which precise effect arises.

My third response is that there are reasonable accounts of control that help avert (or mitigate) the luck objection unavailable to classical theism but available to libertarianism about human and non-classical divine freedom.¹⁸ In particular, many defenders of libertarianism hold that the possession of a variety of reasons upon which one’s acts or decisions across worlds are dependent can help mitigate the luck objection, since one’s actions—though not necessitated by the reasons in question—are still appropriately dependent upon and guided by reasons that distinctively favor each action in the respective worlds in which such actions obtain. One example of this response to luck objections is in Franklin (2012), who holds that something only counts as an agential action if it is appropriately caused by the agent’s relevant mental states like “beliefs, desires, reasons, and intentions” (p. 396–397). Even if one disagrees about the causal link between such prior mental states and the agent’s act, few will deny the plausibility of there being some kind of dependence relation here.¹⁹

In keeping with this response to luck objections, non-classical theists can hold that there exists a multiplicity of reasons upon which God’s different intentional acts across worlds are dependent, such that different reasons factor differentially into the explanation of God’s choices across worlds. But the existence of such a multiplicity of reasons would plausibly entail that there are positive ontological items intrinsic to but numerically distinct from God, meaning that classical theists cannot avail themselves of this maneuver. Moreover, the proposal in question plausibly entails that God’s acts are dependent things—they are dependent on prior divine psychological states. But this is incompatible with classical theism, since God’s act(s)—being identical to God himself—is utterly independent.

Mele’s (1995, 2006, pp. 9–14) proposal for resolving the problem of luck is likewise uniquely unavailable to the classical theist given its explicit affirmation of agent-internal indeterminism (as opposed to indeterminism intervening between the agent’s act(s) and the effect(s) thereof). O’Connor’s (2000) proposal is similarly uniquely unavailable to the classical theist. O’Connor (2000, p. 74) describes a person named Tim deliberating about whether to keep working or to take a break. Tim decides to keep working. On O’Connor’s view, Tim “had the power to choose to continue working or to choose to stop, where this is a power to cause either of these mental occurrences. That capacity was exercised at t in a particular way (in choosing to continue working), allowing us to say truthfully that Tim at time t causally determined his own choice to continue working” (ibid). But the classical theistic God—unlike the non-classical theistic God and Tim—cannot causally determine a given outcome. This is debarred by Biconditional Solution. By contrast—as in Fakhri’s (2021) proposed non-classical model—the non-classical God’s act of will can causally determine a given outcome (even though that act was not itself causally determined by something prior).²⁰ Kane’s (1999, pp. 231–240) proposal of dual tryings is available to neither classical nor non-classical theism, though it is available for human freedom. For Kane, an agent is responsible for (and, hence, in control of) which decision is made because the agent is trying to make each decision, so long as the agent—once an outcome obtains—“endorse[s] the outcome as something she was trying and wanting to do all along” (ibid, p. 233). But, of course, it cannot be the case that an infallibly omnipotent being tries to actualize an outcome but fails to do so.

Other authors have developed theories of divine action. Rice (2016), for instance, argues that agent-causal theories of action—including those applied to God—face difficulties accounting for what action-for-a-reason consists in. Rice’s causal theory of divine action that avoids the aforementioned objection—an account according to which “[f]or every event e in the life of God, e is an action of God’s iff e is caused by the appropriate (rationalizing) [divine] mental items” (2016, p. 272)—is uniquely unavailable to classical theism. This is because her account (and others like it) explains divine action in terms of divine mental items of a particular sort event-causing God’s actions. And this clearly introduces a distinction between God’s act(s) and God’s psychological states, contra DDS.

That, then, is my third response to Objection Two: some of the most prominent libertarian accounts employed in response to the luck objection are uniquely unavailable to the classical theist. Before turning to Objection Three, however, I want to consider a rejoinder to my point about the unavailability to classical theism of a response to the luck objection appealing to the reasons-dependence of an agent’s action(s).

The rejoinder: classical theists can hold that—in some sense—God’s act is differentially explained by different divine reasons across worlds. For while God is identical to each such reason, God’s reasons can still differentially explain God’s act given the hyperintensionality of explanation.²¹ Given that explanation is hyperintensional, it can be true (in our world) that divine reason R₁ but not divine reason R₂ explains God’s act even though ‘R₁’ and ‘R₂’ necessarily co-refer. (In another world, R₂ but not R₁ explains God’s act.) In particular, they both necessarily refer to God himself, given DDS. (At least, that’s what the rejoinder at hand requires.)

I have five replies. First, it is controversial whether explanation is hyperintensional, after all. It is a significant and valuable result if my argument pushes classical theists to adopt a controversial thesis in another domain of philosophy. Second, and more importantly, the kind of explanation with which we are concerned is not merely an epistemic one but a metaphysical one, i.e., one that connects items or relata in extramental reality. We can helpfully think about it in terms of dependence. One’s actions are dependent on prior realities (viz. one’s reasons and other psychological states); and it is (in part or in whole) in virtue of this dependence—so the response to the luck objection goes—that one’s choice is not a matter of luck but is instead under agential control. But in that case, since dependence is asymmetric, and since it is relating items in extramental reality, it simply follows that we have distinct items here. And hence even if God’s various reasons necessarily co-refer to God, what’s problematic for DDS is that God’s act will nevertheless depend upon God’s reasons and hence be distinct therefrom. But this introduces a distinction between God’s act(s) and some prior feature(s) of God, contra DDS.

Third, it seems deeply implausible—at least by my lights—that (say) ‘God’s reason to create this universe’ refers to God himself. Reasons are considerations that count in favor of something (e.g., some action or some outcome thereof). But is God a consideration? Does God count in favor of something? It’s hard to see how. (Indeed, it’s hard to see how this is even intelligible.) Reasons also point towards (i.e., are of themselves directed towards and referred to) that which they favor. But surely God himself (under DDS) does not point towards and is not of himself directed towards or referred to something ad extra.²²

Fourth, even granting that classical theists can avail themselves of reason-based action as a response to luck objections, they still seem to face a unique problem not faced by non-classical theisms and libertarians about human action: vicious circularity. When an agent A acts for a reason R to bring x about, it is surely true that A acts to bring x about because A takes R to be the reason for which A acts. A’s taking R to be the reason for which A acts in this case is thus prior to A’s bringing x about. It is thus false that A takes R to be the reason for which A acts because A brings x about. That gets the order of explanation wrong.

But here’s the rub: this is precisely what the classical theist must reject, which seems to land them in a problematic vicious explanatory circle. For it is a contingent matter that God takes R₁ (which, we can suppose, uniquely favors creation C₁) to be the reason for which God acts. Hence—as we’ve seen—this must be an extrinsic divine predication; nothing intrinsic to God (under DDS) explains why this is the reason for which God acts. But in that case, it’s going to be something extrinsic to (external to, outside of) God that explains this. But in that case, it is in virtue of creation itself (or some portion thereof)—in this case, C₁ itself—that God takes R₁ to be the reason for which he acts. But as we’ve seen, this gets the order of explanation wrong. It is precisely because God takes R₁ to be the reason for which God acts that C₁ results from God’s act. Hence, it is false that it is because C₁ results from God’s act that God takes R₁ to be the reason for which God acts. And that contravenes classical theism, as we’ve seen.

As it turns out, a similar point is made in Leftow (2015). Applying and slightly adapting Leftow’s point to the present context, the marriage of DDS and reason-based action as a response to the luck objection would imply that

the character of the universe determines God’s purpose in creating. It determines the reason God made it. For the character of the universe determines which purposes it matches up with… This can’t be right. Surely God’s reasons for creating determine which universe He makes, rather than which universe He makes determining His reasons for creating. Surely God first has some rather than other purposes for the universe, and so creates it rather than another universe for those purposes. Surely God’s purposes explain His choice of universe. … But some rather than other divine purposes cannot explain God’s choice if which purposes explain it—which purposes God seeks to serve by creating—is settled by what He creates. On [the view under consideration], the universe has a purpose only retrospectively, once it exists. One can’t say that God created it with that purpose rather than others in mind beforehand. (ibid, p. 52)

For if God created it with that purpose rather than others in mind beforehand (i.e., explanatorily prior to the existence of anything extrinsic to God), then this would amount to a contingent predication that is not dependent on facts extrinsic to God—in other words, a contingent intrinsic predication. And that’s debarred by DDS. Leftow continues: “But then it is hard to see what point there is to purpose-talk here. Its point has to be explaining why God created what He did, but how can God’s purposes explain what He creates if what He creates determines what His purposes in creating were?” (ibid). That, then, is my fourth response: even granting the rejoinder under consideration, it leads to a vicious circularity. (Or, at least, it gets the order of explanation wrong.) And this problem is undergirded by DDS’s unique commitment to the extrinsicality of contingent divine predications.

Fifth, even if the hyperintensionality rejoinder successfully rebuts my third reply to Objection Two, my first two replies remain intact. And so even if the rejoinder at hand succeeds, Objection Two is not vindicated. (As I’ve argued, though, the hyperintensionality rejoinder does not succeed.)

Onward we march to Objection Three.

Objection Three

Perhaps the classical theist simply doesn’t share the intuition behind premise (13) or behind the idea that control requires an intrinsic state securing a certain outcome. If so, then surely the argument is powerless to move such a classical theist to abandon their position. What price is paid by such a classical theist?

Reply

This is valuable objection. In response, I make four points.

First, even if a classical theist doesn’t share the relevant intuition(s), the argument can still be deeply valuable. For the argument can still weigh on the evidential scales of those who do share the intuition, including both classical and non-classical theists (as well as non-theists). Argument evaluation is highly person-specific, depending on a whole concoction of factors like one’s prior experiences, beliefs, books and articles read, videos watched, body of testimonial evidence, and plausibility structures. Given this person-specific nature of argument evaluation, the argument can retain profound value for those who do share the relevant intuition(s). It can also broaden the perspectives of those who don’t share such intuitions.

Second, I can at least point the classical theist who doesn’t share the relevant intuition(s) to several other authors (libertarians among them) who find the kind of ‘downstream indeterminism’ required by classical theism severely control-diminishing. (I quoted earlier Levy and Clarke on this precise point, and there are others besides.) This indicates that the relevant intuition(s) is widespread and hence one worth addressing even if one doesn’t share it.

Third, I have not merely appealed to (13)'s intuitive plausibility. I have also appealed to well-nigh self-evident cases wherein control is significantly diminished and have argued that they are relevantly similar to the God of classical theism. Among the cases considered were the temperature case, the button-lottery case, the steering wheel case, and the ball-target case. What’s more, I have connected accounts of direct control to classical theism, arguing that the latter don’t satisfy the former. (Cf. footnote 13.) I would contend, moreover, that the providential collapse argument is strengthened by the various considerations raised in response to Objections One and Two. Thus, registering that one doesn’t share the relevant intuition(s) is not sufficient to address the providential collapse argument.

Fourth, the fact that many find (13) and its support deeply plausible urges classical theists to develop workable, classical-theist-friendly accounts of divine control on which the downstream indeterminism doesn’t collapse (and, indeed, doesn’t even diminish) God’s control over which effect obtains. To be sure, if the providential collapse argument succeeds, no such account can ultimately work. But the contrapositive is likewise true: if any such account can ultimately work, the providential collapse argument doesn’t succeed. The argument can therefore facilitate the development of such accounts and thereby advance discussions about God and divine control. To that end, one purpose of my providential collapse argument is akin to Mele’s purpose in defending the luck objection: “[M]y aim in developing this… problem for agent causationists and other conventional libertarians is to present it sufficiently forcefully to motivate them to work out solutions to it—proposed solutions that I and others can then assess. That is the way of progress” (2006, p. 70). My argument is likewise a path toward progress. I am not here to knock down classical theism, but to serve it.

5 Conclusion

After a brief survey of the modal collapse literature, I argued for a Biconditional Solution to the modal collapse argument: classical theists avoid modal collapse if and only if they embrace an indeterministic link between God and his effects. I then developed and defended two challenges to classical theism raised by this solution. The first challenge—the intentional collapse argument—is that it collapses God’s status as an intentional agent who knows and intends what he is bringing about in advance. The second challenge—the providential collapse argument—is that it collapses God’s providential control over which creation obtains.

Once again, I stress that the arguments I’ve developed here aren’t weapons used to attack a position or defend a tribe. They are meant, instead, to advance the modal collapse debate and to further our collective inquiry into the ultimate nature of reality.²³