W. Matthews Grant’s Dual Sources Account and Ultimate Responsibility

Abstract

A number of philosophers and theologians have recently challenged the common assumption that it would be impossible for God to cause humans actions which are free in the libertarian or incompatibilist sense. Perhaps the most sophisticated version of this challenge is due to W. Matthews Grant. By offering a detailed account of divine causation, Grant argues that divine universal causation does not preclude humans from being ultimately responsible for their actions, nor free according to typical libertarian accounts. Here, we argue that the kind of divine universal causation that Grant proposes is incompatible with a plausible interpretation of Robert Kane’s influential conception of ultimate responsibility. This conclusion is significant since Grant seeks to harmonize his divine causal account with Kane’s articulation of ultimate responsibility.

Proponents of libertarian freedom, and incompatibilism more generally, regularly claim that humans must bear ultimate responsibility for their free actions. While there are different ways of articulating the relevant notion of ultimate responsibility, the rough idea seems to be that an agent cannot be genuinely accountable for some act if the sufficient reason for this act rests outside of this agent’s control. Rather, the agent herself must in some way be personally responsible for the fundamental source(s) of her behavior; the buck, in some important sense, must stop with her. Given this characteristic emphasis on ultimate responsibility, it’s unsurprising that many libertarians, perhaps the vast majority, assume that if God causes each human action, then no (mere) human is ultimately responsible for what she does. ¹ W. Matthews Grant thinks this assumption is mistaken, however (see Grant, 2019, 2017, 2016, and 2010).

In a number of recent works, Grant draws from broadly Thomistic and scholastic resources to defend what he calls the ‘Dual Sources’ account of divine and human causation. In brief, this account refers to the idea that every intentional human action has two causal sources, one human and one divine, and that God’s causing of a human action poses no genuine threat to the exercise of human libertarian freedom, and, so, human ultimate responsibility. What’s more, on this Dual Sources account, God’s causing of each human action affords God a kind of meticulous providence where God gets all the details He wants within creation without thereby precluding humans from being ultimately responsible for a sizeable subset of their acts (e.g., Grant, 2010, 44; 2016, 214–215). Although this wedding of libertarian freedom and divine universal causation is perhaps attractive to some, we contend that Grant’s Dual Sources account fails in an important respect. Specifically, we contend that, despite Grant’s arguments to the contrary, the kind of divine universal causation that Grant proposes is incompatible with Robert Kane’s influential libertarian conception of ultimate responsibility, or at least a plausible interpretation thereof.

The failure of Grant’s attempt to reconcile divine universal causation with (a plausible interpretation of) ultimate responsibility among human agents is significant. This is because Grant’s Dual Sources account is, in our estimation, the most sophisticated contemporary attempt at such a project of reconciliation, of which there are many.² Thus, it stands to reason (although we will not contend for the point here) that the failure of Grant’s attempt to achieve the noted reconciliation generalizes to any account aimed at the compatibility of divine universal causation and rich, ultimate responsibility for humans.

The argument to come is structured as follows. In the first section, we present the central details of Grant’s Dual Sources account. Then, in section two, we highlight Grant’s attempt to show that his account is compatible with Robert Kane’s influential notion of ultimate responsibility. Finally, in section three, we contend that Grant’s account of divine agency precludes the relevant sense of ultimate responsibility, or, minimally, a plausible understanding of it.

1 Grant’s Dual Sources Account

Grant affirms two doctrines which seem to contradict one another. On the one hand, Grant affirms what he terms the “doctrine of divine universal causality” (hereafter, “DUC”). This is the teaching that “Necessarily, for any entity distinct from God, God directly causes that entity to exist at any time it exists.” DUC is said to have the implication that “God directly causes all creaturely actions, whether free or unfree, since whatever they consist in, creaturely actions are entities distinct from God” (Grant, 2019, 4). On the other hand, Grant affirms that DUC does not compromise human libertarian freedom, nor the ultimate responsibility often thought to be correlated with it (e.g., Grant, 2010, 22). Hence Grant assures us that we can have “a traditional view of God’s sovereignty,” specifically a view where God gets all the details He wants within creation, without thereby compromising “the sort of robust libertarian freedom that many contemporary theists have wanted to endorse” (Grant, 2016, 214–215). To make his case, Grant distinguishes two models of divine action and contends that his preferred model enables him to avoid placing DUC in competition with human actions that are free in the libertarian sense.

First, there is what Grant calls “the popular model” (Grant, 2019, 56). According to it, the following parts are involved when God brings about some creaturely effect E.

• (a) God.

• (b) E.

• (c) God’s choice, decree, or intention to bring about E, which is intrinsic to God, is that in virtue of which God causes E and which would not exist were God not causing E.

• (d) The casual-dependence relation between God and E (Grant, 2019, 56).³

If the popular model is true, Grant acknowledges that God’s causing of E precludes the human from freely causing E in the libertarian sense.

The reason for this incompatibility is straightforward. According to libertarians, an act is free in the libertarian sense only if its agent performs that act voluntarily and intentionally. Further, that free act must be either undetermined, in that there is no factor which is jointly prior to and logically sufficient for the act, or, if the act is determined, then the agent’s responsibility for the act must be derived from that agent’s performance of some prior voluntary and intentional free act that was not itself determined. However, if the popular model of divine agency is true, it looks like E is determined in the relevant responsibility-precluding sense. For, given (c), there is some real, intrinsic property, feature or state of God in virtue of which God causes E. Given that this feature or state of God is either His choice, decree, intention or etc., it seems inescapable that this state will be both prior to E, in the sense of explanatory priority, and logically sufficient for E, in that, given omnipotence, this divine decree, or choice, or etc. entails that E occurs. Consequently, when E concerns some intentional human action, God’s causing of E by way of (c) determines E, which thereby precludes the human from freely bringing about E in the libertarian sense—assuming, of course, that God’s causing of E precludes the human from being responsible for the determined E by way of her performance of some prior voluntary and intentional free act that was not itself determined (e.g., Grant, 2019, 56–58, 76–79).

Although Grant believes that the popular model is widely assumed by contemporary theologians and philosophers, he maintains that it’s not the only viable way of conceiving of divine action. Indeed, if God is perfectly simple (as characterized by the likes of St. Thomas Aquinas, for example), God is not the type of being that forms anything like contingent internal intentions to bring about various effects, as human agents do (e.g., Grant, 2019, 56–58, 76–79). Thus, a simple God does not undergo that which is referred to by (c) in the popular model. Because of this, Grant has us consider another model of divine action, that which he labels the “extrinsic model” (EM, hereafter). EM shares all the same parts with the popular model, only with two exceptions. The first exception is that (c) plays no role in EM. The second exception is that Grant adds a few more items in EM than appear in the popular model.

Taken altogether, EM characterizes God’s bringing about some creaturely effect E as follows:

• (a) God.

• (b) E.

• (d) God’s reason for causing E.

• (e) The cause-dependence relation between God and E.

• (f) God’s causal act, or causing of E, which consists in E plus the causal relation between God and E.

• (g) God’s willing or choosing E, which is nothing else than God’s causing E for a reason when God could have done otherwise (Grant, 2019, 60).

Grant contends that the removal of (c), within EM, allows for the compatibility of God causing a human act and yet that act being a free human action in the libertarian sense.

The reason for this is as follows:

[G]iven EM, none of the items on the scene when God causes [E] constitutes a factor both prior to and logically sufficient for [E]. But, then, on [EM], God’s causing [E] does not render [E] determined. What goes for [E] goes for any creaturely act. Given EM, God can cause all creaturely acts without rendering any of them determined. Thus, on EM, the only way a creaturely act caused by God would be prevented from being free in the libertarian sense is if God’s causing such an act precluded its creaturely agent’s performing the act voluntarily and intentionally (Grant, 2019, 61).

In other words, without (c) there is no factor that removes a human’s act being done voluntarily or intentionally. This is for two reasons. First, this is because neither God (i.e., (a)) nor His reasons for acting (i.e., (d)) are logically sufficient for the relevant effect’s occurrence (i.e., the creature’s act), since God is free to choose to bring about that effect or not. Second, this is because neither the effect produced by God (i.e., (b)) nor the resulting relation between God and the effect (i.e., (e)) are prior to the effect, even if they are logically sufficient for it. That is to say, although the occurrence of the effect and God’s resulting relation to the effect individually entail that the effect transpires, both of these entailments are simultaneous with, if not subsequent to, the occurrence of the effect. As Grant conceives of things on EM, God’s causing of a human action is not prior to this human action but simultaneous with it: “The co-operation between God and the creaturely agent is one which neither God’s act nor the creature’s act can be causally or temporally prior to the other. […] God’s act and the creature’s act are simultaneous (or concurrent) necessary conditions for each other” (Grant, 2019, 63; cf. Grant, 2010, 35–36).⁴ Hence, a voluntary and intentional human act has dual sources, one divine and one human, and is not exclusively determined by God. This is Grant’s Dual Sources account, which allegedly pairs free and incompatibilist libertarian human action, and the ultimate responsibility it affords, with DUC (e.g., Grant, 2019, 71, 99–144).

It’s worth pausing to consider the bar that Grant sets for his Dual Sources account. As noted, his goal is to show that DUC, by itself, does not introduce any factor that is jointly (i) prior to and (ii) logically sufficient for the occurrence any human action. Both conditions merit attention. Grant spells out the “logically sufficient for” relation as follows: “a is logically sufficient for b just in case it is not possible for a to exist (or occur, or obtain) without b’s existing (or occurring, or obtaining).” By “prior to” Grant has in mind specifically an asymmetrical dependency or explanation relation between two things (whether the dependency is temporal or not) (Grant, 2019, 6). Grant concedes that if DUC introduces any factor that satisfies both of these conditions, libertarian free action would be impossible for humans.

However, Grant doesn’t believe that DUC introduces any of the factors that would preclude creaturely libertarian freedom. On the contrary, he maintains that his Dual Sources account permits humans to be not only proximately but also ultimately responsible for much of what they do. And he turns to Robert Kane’s account of ultimate responsibility to make his case.

2 Grant on Ultimate Responsibility

In at least three separate publications, Grant contends that his Dual Sources account is compatible with Kane’s articulation of ultimate responsibility (Grant, 2019, 68–70, 2016, 226–229, 2010, 41–43). Kane’s influential statement of the doctrine runs as follows:

(UR) An agent is ultimately responsible for some (event or state) E’s occurring only if (R) the agent is personally responsible for E’s occurring in a sense which entails that something the agent voluntarily (or willingly) did or omitted, and for which the agent could have voluntarily done otherwise, either was, or causally contributed to, E’s occurrence and made a difference to whether or not E occurred; and (U) for every X and Y (where X and Y represent occurrences of events and/or states) if the agent is personally responsible for X, and if Y is an arche (or sufficient ground or cause or explanation) for X, then the agent must also be personally responsible for Y (Kane, 1998, 35).⁵

Grant understands the goal of Kane’s UR to provide necessary but not sufficient conditions for ultimate responsibility for some act, and Grant takes them to be such. Importantly, Grant maintains that his Dual Sources account, with its reliance on EM, satisfies these necessary conditions.

Start with condition R. The first portion of this is condition is satisfied, says Grant, “because the agent’s act is something he voluntarily does,” given the Dual Sources account (Grant, 2019, 68). It’s something the agent performs to secure some end in keeping with standard accounts of intentional action. Additionally, Grant thinks that the agent has the ability to do otherwise on his Dual Sources account even when all antecedent conditions remain the same, in step with standard characterizations of this ability by libertarians. He says, “S who performs action A has the ability to do otherwise only if there is a possible world the same as the actual world in all factors or conditions prior to A, but in which S does other than A” (Grant, 2019, 65). The Dual Sources account is said to be compatible with this agential ability since, given EM, God’s causing of A does not introduce any factor or condition prior to A and because the agent could have performed not-A while simultaneously being caused to do not-A by God. Said differently, once (c) of the previously described popular model is removed (i.e., once one gives up on the idea that there must be some choice, decree or intention to bring about A, which is intrinsic to God), we find that there is nothing peculiar to the Dual Sources account that would limit an agent’s ability to do otherwise in the noted sense. For, as explained, the Dual Sources account relies on EM wherein the popular model’s (c) is absent.

What about condition U of UR? Grant maintains that “[g]iven that R is satisfied, U will fail to be satisfied only if there is some arche for the agent’s action for which the agent is not personally responsible in the sense indicated by R” (Grant, 2019, 68). On this score, Grant locates two characteristics that appear to be essential for something to count as an arche on Kane’s understanding. First, there is the notion of an arche being a sufficient reason for some event or state. This comes in three kinds: (i) a sufficient condition, (ii) a sufficient cause, or (iii) a sufficient motive.⁶ The second characteristic of an arche is that is must be the origin, source, or cause of that for which it is a sufficient reason. Grant states that “[w]hat precludes ultimate responsibility according to condition U of UR is that there be sources sufficient for our actions, for which sources we are not personally responsible in the sense indicated by R” (Grant, 2019, 68). In Kane’s words, “[i]f the action did have such a sufficient reason for which the agent was not responsible, then the action, or the agent’s will to perform it, would have its source in something that the agent played no role in producing” (Kane, 1998, 73).⁷

Given this description of UR, Grant maintains that “God’s causing an agent’s action will preclude U’s being satisfied only if it introduces some arche for the action for which the agent is not personally responsible in the sense indicated by R” (Grant, 2019, 69). But the Dual Sources account doesn’t do this, thinks Grant, because it relies on EM. Grant explains,

We have already seen that on EM God’s causing some act does not introduce an origin, source, or cause of the act the obtaining of which entails or is logically sufficient for the act. Thus, God’s causing the act does not introduce any arche of sorts (i) or (ii) and hence does not introduce any arche of sorts (i) or (ii) for which the agent is not personally responsible. Nor if the agent’s act has an arche of sort (iii) does God’s causing the act preclude the agent from being personally responsible for that arche. For suppose that the act has a sufficient motive. God’s causing the act is perfectly consistent with its being the case that something the agent voluntarily did or omitted, and for which the agent could have done otherwise, causally contributed to the agent’s having that motive, and made a difference as to whether the agent had it [cf. condition R of UR]. For instance, decisions that the agent made in the past, and for which he could have done otherwise, may have causally contributed to the agent’s having the sufficient motive he now has (Grant, 2019, 69).

In short, since EM does not include the formation of some new and intrinsic divine choice, intention, or decree, there is nothing present within God’s causing of a human’s action that precludes that human from being ultimately responsible for this act as described by UR. And, as we have seen, a significant reason this is not precluded, according to Grant, has to do with the supposition that God’s causing a creaturely action or event via EM is compatible with the relevant creature satisfying UR’s condition R.

Grant’s defense of the compatibility of UR with Dual Sources is both more subtle and comprehensive than the present treatment captures. Nevertheless, we maintain that Grant’s argument for this compatibility rests upon a mistaken emphasis on what the removal of the popular model’s (c) implies for God’s causing of human action. More specifically, while Grant is correct that God’s causing a human action via that which is referenced in (c) precludes this human from being ultimately responsible for it in UR’s sense, Grant fails to recognize the way in which God’s causing of an action in accordance with EM, when placed within Grant’s wider Dual Sources account, likewise precludes this form of responsibility. In what follows, we’ll argue that this is true partially because Grant’s understanding of UR requires (what we will show to be) an implausible reading of UR.

3 On the Incompatibility of EM and UR

Our argument for the incompatibility of EM, when couched within Grant’s wider DUC, and UR runs as follows.

• 1) If DUC and EM are jointly true, then divine causation introduces an explanation for each human action that is both prior to and logically sufficient for each action’s occurrence for which no human is personally responsible.⁸

• 2) If divine causation introduces an explanation for each human action that is both prior to and logically sufficient for each action’s occurrence for which no human is personally responsible, then no human is ultimately responsible for any action she performs.

Therefore,

• 3) If DUC and EM are jointly true, then no human is ultimately responsible for any action she performs.

Premise (2) is part and parcel of Kane’s UR, and, as we have seen, something to which Grant would adhere. (Here and henceforth, let all references to “ultimate responsibility” and like terms refer to Kane’s UR, unless the context clearly dictates otherwise.) Premise (1) in effect is the claim that if God causes every human action in the manner referred to by EM, such causation constitutes an arche, in Kane’s basic sense, for each of these human actions for which no human can claim personal responsibility (again, in Kane’s basic sense, but about which more subsequently). Obviously, (1) is the controversial premise that Grant needs to resist. However, (1) strikes us as more plausible than its denial.

We begin our defense of (1) with an analysis of divine causation provided by Grant for the purpose of circumventing claims like (1). For this analysis, Grant draws a comparison between EM and human agent-causal libertarianism.

Reflecting again on the theory of agent-causation, it should be clear that […] an agent’s basic action [is not] explanatorily prior to the existence, say, of the executive state of intention the direct causing of which constitutes (or just is) that very action. For, whether we say that the agent’s basic action consists in just the causal relation between the agent and the intention, or whether we say that it consists in the intention together with the causal relation, the agent’s action will presuppose the intention. If the agent’s action consists in just the relation, then the action will presuppose the intention as a relation presupposes its relata. If the action consists in the relation and the intention together, then the action will presuppose the intention as that in which it partially consists. Either way, the agent’s basic action will not be explanatorily prior to the intention, since the action will presuppose the intention’s existence. In this way, God’s agency as characterized by EM is not in all respects sui generis. While both God and creaturely agent-causes cause, and therefore, explain or account for, the effects they bring about in their basic causal acts, these basic acts do not themselves cause, nor are they explanatorily prior to, the effects brought about in those acts (Grant, 2019, 62–63).

In other words, while both human and divine agent-causes do cause and explain various effects, the relevant agent’s causing of his or her basic acts are not explanatorily prior to these basic actions—say, an intention in the human case and creation in the divine case—since the agent’s causing of these basic acts are constituents of those actions or else identical to them. By extension, Grant’s thought seems to be that if EM, conceived along agent-causal lines, entails that God’s actions are not explanatorily prior to God’s effects within creation, (1) is false. And when Grant says that God’s actions are not explanatorily prior to God’s effects within creation, we take this to mean that God’s actions are not jointly prior and logically sufficient for these effects (cf. Grant, 2019, 60–61).

Unfortunately, though, turning to human agent-causal action fails clearly to escape the problem facing Grant’s EM. To discern this, consider Grant’s favored analysis of some human agent, Σ, causing her intention, φ, in a basic way. He analyses this in terms of the causal dependence relation between Σ and φ (i.e., φ being causally dependent upon Σ), together with φ (Grant, 2019, 59). This is what Σ’s causing of φ consists in, similar to the manner in which a mental state just is a brain state on reductive materialism (Grant, 2019, 64). As we have just seen, Grant maintains that Σ’s causing of φ is not explanatorily prior to φ (where, again, we take this to mean that God’s actions are not jointly prior to and logically sufficient for their effects), since this causal act presupposes φ plus the causal relation between Σ and φ. But, it seems, Grant moves too quickly here. Arguably, Σ’s immediate causing of φ amounts to an explanation of φ that is both logically sufficient for and prior to φ; that is, that Σ’s causing of φ is explanatorily prior to φ. This causation is sufficient for φ because Σ’s causing of φ is sufficient for the occurrence of φ; or we might say that the obtaining of the causal relation between Σ and φ is logically sufficient for the occurrence of φ. And, crucially (and in keeping with the language of premise (1) of our argument), Σ’s causing of φ is also prior to φ in the sense that φ depends upon or is asymmetrically explained by this causation—i.e., φ bears the property of being caused by Σ whereas Σ’s causing of φ does not bear the property of being caused by φ. If this is right, then Grant’s preferred analysis of basic agent-causal action is insufficient to shield the Dual Sources account, which is filtered through EM, from the relevant responsibility-undermining threat.

Here’s an illustration of the idea. Suppose through either divine design or by some strange cosmic accident, identical twins Tomax and Xamot became psychically and agentially connected, in certain respects, in utero. As a result, whenever (and only whenever) Tomax causes the executive intention ‘within himself’ to dance a jig, he also simultaneously and directly causes Xamot to cause the executive intention to dance a jig himself, no matter where Xamot is located. (Tell whatever metaphysical story you would like to preserve this dual, simultaneous causation. Maybe, e.g., substance dualism is true, and somehow the joint formation of Tomax and Xamot in utero enabled a psychic link between Tomax and Xamot such that Tomax’s immaterial soul acquires the ability to cause, through a single act, his own executive intentions as well as cause these intentions in Xamot on certain occasions.) When it comes to Tomax’s causing Xamot to dance a jig, we might, in keeping with Grant’s analysis of basic human actions, analyze Tomax’s basic action in terms of the causal dependence relation between Tomax and Xamot’s executive intention to dance the jig (i.e., Xamot’s intention being causally dependent upon Tomax), together with Xamot’s intention. (Of course, if Tomax causes Xamot to form the intention to dance, there will also be a story to tell about Xamot’s causing of his intention to dance. We suppose that Xamot causes his relevant intentions in keeping with Grant’s favored analysis of basic actions. So, Xamot’s intention to dance will have dual sources.) Given this analysis of Tomax’s causing of Xamot’s intention, the former’s causal act presupposes Xamot’s intention in that there is a sense in which relations presuppose their relata. But this hardly means that Tomax’s causation doesn’t introduce an arche of Xamot’s intention for which Xamot would need to be personally responsible (in keeping with UR’s U condition) if Xamot is to be ultimately responsible for his intention.

For suppose that Tomax causes Xamot to dance a jig at a funeral in a circumstance in which it would be deeply offensive to do so. Few with knowledge of the agential interconnection between Tomax and Xamot would be inclined to blame Xamot for his offensive dance, unless they thought Xamot had some ability to influence or cause whether Tomax caused him to dance when Xamot did. We take this as evidence that, although a basic action may presuppose its effect in one sense, effects can still be explanatorily dependent on this act—and in this respect, a basic action can be prior to its effect. More generally, we analyze this explanatory dependency relation in terms of the idea that the effects of basic actions bear the property of being caused by agents whereas agents’ causing of their effects do not bear the property of being caused by these effects. But even if this analysis is not quite right, it still does seem that one’s directly causing of another’s basic action can introduces an arche for the latter’s action, and so be relevantly prior to it. And, of course, this spells trouble for Grant’s pairing of EM and DUC.

Here, Grant might try to resist the above thought experiment by appealing to a principle of his, which we’ll call ‘CP’ (short for Counterfactual Principle).

(CP): “Agent S has counterfactual power over event E if S performs some act with respect to which S could have done otherwise all antecedent conditions remaining the same, and without which act event E would not have occurred” (Grant, 2019, 69).

The idea behind (what we’re calling) CP is this: for some actual event, E, if an agent performs p (and could have done otherwise), and ~ p implies ~ E, then the agent has counterfactual power over E.⁹ In this context, the claim would be that Xamot must satisfy CP with respect Tomax’s causing of his dance if he is to be personally responsible for the dance; or, minimally, the claim would be that Xamot satisfying the conditions in CP with respect to Tomax causing Xamot’s dance is a good indicator as whether Xamot might be considered personally responsible for his offensive dance. So, appealing to CP, Grant might suggest that Xamot isn’t personally responsible for his dancing a jig because he lacks counterfactual power over Tomax’s causing him to dance—and, therefore, even though Xamot isn’t personally responsible for his dance this isn’t because Tomax’s causing of Xamot’s dance is prior to and (at least materially) sufficient for the latter’s dance.¹⁰

But, we deny that Xamot lacks this power, as Grant spells it out. For, in the event of Tomax’s causing Xamot to dance a jig, given that Xamot dances the jig (when he could have done otherwise, holding fixed all antecedent conditions until the point of his deciding to dance a jig because Tomax decides to dance a jig), and Xamot’s not dancing the jig implies that Tomax doesn’t cause him to dance the jig, then it follows by CP that Xamot has counterfactual power over Tomax’s causing him to dance a jig. That is, Xamot could have chosen to remain reverent at the funeral, and (so it seems to us) that Xamot has the sort of power over Tomax’s causing of his decision to dance commensurate with CP. Even so, Xamot isn’t personally responsible for Tomax’s causing of his decision to dance.

Now, if Grant were to respond that Xamot does not have counterfactual power over his deciding to dance a jig, we’d need to hear why this is so. Perhaps one might argue that Xamot needs causal counterfactual control over Tomax causing his deciding to dance a jig in this scenario in order for Xamot to be personally responsible for his dancing a jig. But, he doesn’t have such causal control over what’s happening; so, he’s not personally responsible.

However, this objection isn’t one to which Grant can appeal. For Grant uses CP to explain the manner in which humans have counterfactual control over God’s causing their acts on his Dual Sources account. But, in the case of God, if God is the uncaused Cause (as Grant presumably maintains), nobody can be said to exercise causal counterfactual power over what God causes, even if humans have counterfactual power over what God causes in the sense commensurate with CP. So, it’s doubtful that Grant can consistently resist the thought experiment by affirming that CP requires causal, counterfactual control. And, if not, we’re left wondering why, given Grant’s CP, we shouldn’t say that Xamot is personally responsible for his dancing a jig (even though, we grant, Xamot doesn’t have causal counterfactual power over Tomax causing his dancing a jig just now). For, Xamot has counterfactual control (in the sense described by CP) over his dancing a jig. Nevertheless, we think it’s clear that Xamot isn’t personally responsible for his dancing a jig in the above situation. And we think most (all?) should agree. So, CP won’t be sufficient to address the worries expressed in the example of Tomax and Xamot.

But it’s important that we don’t lose the forest for the trees. The fundamental point we wish to make is that even if one directly causes an action, and we analyze this causation as Grant wishes, this doesn’t mean that this causation doesn’t introduce an arche for the action, or that which is prior to it. On the contrary, we think the Tomax and Xamot thought experiment illustrates how such priority might be introduced. This is significant; for, this thought experiment is relevantly like Grant’s EM.

More directly, God’s causing of some human act, E, in accordance with EM appears to introduce an explanation of E that is (following the language we use in (1)) both prior and logically sufficient for E. Why this seems to be so should now be clear enough. Recall the following manner in which Grant analyzes divine causation on EM:

(f) God’s causal act, or causing of E, which consists in E plus the causal relation between God and E.

However, along the same lines just detailed in the human agent-causal case, it’s plausible that God’s causing of E constitutes an explanation of E that is prior to and logically sufficient for E. As Grant concedes,¹¹ God’s causation is logically sufficient for E because God’s causing of E is sufficient for the occurrence of E. Indeed, given God’s omnipotence, divine causation cannot fail to achieve its effects. God’s causation of E is also prior to E in the sense that it asymmetrically explains why E transpires—i.e., E bears the property of being caused by God whereas God’s causing of E does not bear the property of being caused by E. Plus, the Tomax and Xamot case appears to be relevantly similar to the way in which divine causation works on EM, and yet Tomax’s causing of Xamot’s executive intention to dance appears to be prior and (at least materially) sufficient for Xamot’s intention. So, we maintain, God’s causation of E amounts to a prior and logically sufficient explanation of E.

Beyond that, according to Grant’s DUC God causes every event within creation, including every human action. But if God’s causing of every human action introduces an explanation for each of these actions that is jointly prior and logically sufficient for its occurrence, then it seems that no human agent can be ultimately responsible for any act she performs. For any act she performs has some arche (or sufficient ground or cause or explanation) that ultimately traces back to God, not, it seems, to herself. But if this is so, it looks very much as if (1) is true. Given Kane’s UR, the only apparent manner in which (1) can be avoided is by proposing that human agents can be personally responsible (again, in Kane’s sense) for God’s causing of her actions. For condition U of UR says that if something is an arche for S’s action, S must be personally responsible for this arche if S is to be personally responsible for this action. But, given the particulars of Grant’s Dual Sources account, such a claim seems implausible.

Grant, of course, will beg to differ. As partly explained in the previous section, a significant way in which Grant attempts to resist that idea that divine causation introduces an arche for which no human is personally responsible is by relying on condition R of UR. Given this reliance, we can imagine that Grant might reply something like the following:

If the only way to avoid (1) is by proposing that humans can be personally responsible for God’s causing of her action, then this isn’t at all that hard to do. Consider condition R of UR, again. It says that an agent, S, is personally responsible for the occurrence of some event or state, E, only if there is some act that S voluntarily (or willingly or intentionally) did or omitted, and for which S could have voluntarily done otherwise, that causally contributed to E’s occurrence and made a difference as to whether or not E occurred. Now, consider God’s causing S to A. Given my stipulations that when S wills A S could have done otherwise and that S’s Aing contributes to God’s causing S to A, it follows that S is personally responsible for her Aing. And, if so, then, according to your above line of reasoning, (1) is false (cf. Grant, 2019, 68–69).

In other words, Grant might seek to circumvent our argument by appealing to what we might (non-pejoratively) call the ‘loose interpretation’ of Kane’s idea of personal responsibility: S may be personally responsible for God’s causing S to A just in virtue of S causing A, so long as S’s causing of A meets the conditions just described in R. For, again, in meeting these conditions, S’s causing of A (for which S is personally responsible) contributes to God’s causing of A in that S’s causing of A partly constitutes God’s causing of A given EM’s (f), which is referenced above.¹²

In response, we think we need to slow down and home in on what Grant (and, maybe, Kane) means by ‘personal responsibility’. For, we think that a plausible understanding of this language is required for a plausible understanding of ultimate responsibility. But, once we see what a plausible reading of these notions will be, it will follow that Grant’s DUC (when paired with EM) is not compatible with UR, after all.

To begin to get clearer on what’s at issue, here, let’s zoom in on R by considering the following. Call the actual world (the possible world that is actual) ‘ALPHA’. And suppose that ALPHA includes our writing this paper voluntarily and freely, i.e., according to the conditions spelled out in what we are labelling the loose interpretation of Kane’s R. Call our writing this paper in accordance with R’s conditions, when interpreted loosely, ‘P’. Now, consider:

K: ALPHA implies P.

According to the loose interpretation of R, we are personally responsible for K just by virtue of the fact that we are (we assume) personally responsible for P. But, should we really conclude that UR (because R) has the (we think) strange implication that our being personally responsible for writing this paper confers on us personal responsibility for what some possible world (ALPHA) includes? We think that we should not draw this conclusion about UR; for, ALPHA, as a possible world, includes P even if we never write this paper—i.e., if ALPHA were not actual. It’s difficult for us to believe that we can be personally responsible for something that would be the case (i.e., K) even if we never write this paper. Moreover, K is a necessary truth. That is, in all possible worlds, it’s true that ALPA implies that we write this paper. So, on this way of understanding UR, UR (because of the loose interpretation of R) has the implausible (impossible?) implication that we are personally responsible for a necessary truth. We deny that we are personally responsible for any necessary truths, as one of us has argued elsewhere (see Turner 2015; Turner and Capes 2018). So, we think that this understanding of UR (because of the loose interpretation of R) is implausible.

There are two ways to take this conclusion, neither of which is favorable for Grant’s account. The first way to take this is that UR, as understood, is implausible and not such that anyone should affirm it because R, as defined by the loose interpretation, should be rejected. The second way to take this conclusion is to see that Kane never meant for UR to be amenable to the loose interpretation of R. Indeed, Kane’s more recent statements of UR don’t seem obviously to lend themselves to anything relevantly like a loose interpretation of R’s personal responsibility.

For example, in (Fisher, Kane, et al. 2007, 14), Kane says this:

Free will also seems to require that the sources or origins of our actions lie ‘in us’ rather than in something else (such as the decrees of fate, the foreordaining acts of God, or antecedent causes and laws of nature) outside us and beyond our control…I call this…the condition of Ultimate Responsibility (or UR, for short)…The basic idea of UR is this: To be ultimately responsible for an action, an agent must be responsible for anything that is a sufficient cause or motive for an action’s occurring. (Italics in the original)

The language Kane uses to define UR is similar in his (2011).¹³ The point we wish to make is this: there is no mention (not in the bits quoted or mentioned just above, nor the sentences surrounding those quoted) of personal responsibility. Condition R of UR is left undefined. More importantly, Kane’s UR implicitly in its original context (Kane, 1998, 5–8, 35, 74, 81–82, 88) and explicitly in later statements (Kane, 2007; 2011) seems to be intended to be incompatible with divine decrees and things ‘outside’ the human fixing which actions humans perform. We think that we should take Kane’s intent seriously; so, we think that the loose interpretation of R should be rejected. Plus, when the loose interpretation of R is embraced, this leads to the implausible conclusion that we are personally responsible for a necessary truth, as we’ve argued.

So perhaps condition R of UR should be left undefined as Kane does in some of his later works. An alternative option is to refine R so as to capture what might have been closer to Kane’s intent for this condition all along. Given Kane’s emphasis on the sources or origins of our actions lying ‘in us’, we think a better way to understand what’s at issue with the language of ‘personal responsibility’ is something like the idea of self-determination or the springs of action originating in that which is identical to the agent. That is, for an agent to be personally responsible for her action, she must meet the description found in Kane’s R (i.e., the relevant action depends upon something she voluntarily did, when she could have done otherwise, which makes a difference as to whether or not A occurs, etc.), but also the source of the action must be identical to her, qua agent. Here’s a way to put the idea from Richard Taylor which we find helpful:

If I believe that something not identical to myself was the cause of my behavior—some event not identical to myself was the cause of my behavior—some event wholly external to myself, for instance, or even one internal to myself, such as a nerve impulse, volition, or whatnot—then I cannot regard the behavior as being an act of mine, unless I further believed that I was the cause of that external or internal event. (Taylor 1974, 55)

What Taylor expresses here (especially before the final clause) gets at the idea of self-determination or actions originating from that which is identical to the agent. And, importantly, this admittedly vague but useful idea of self-determination fits with what is plausibly thought, for the reasons explained, to be a crucial ingredient of Kane’s intent for UR.

Furthermore, this idea of self-determination can be assimilated more explicitly into Kane’s UR. Keeping all else the same within UR, UR’s R may be modified as follows (with the addition in italics):

(R*) the agent is personally responsible for E’s occurring in a sense which entails that something the agent voluntarily (or willingly) did or omitted, and for which the agent could have voluntarily done otherwise, either was, or causally contributed to, E’s occurrence and made a difference to whether or not E occurred and the sufficient cause or explanation of that which the agent did or omitted with respect to E’s occurrence originates from that which is identical to the agent qua agent.

So, in the broader context of UR, condition R* essentially says that S’s personal responsibility for E requires that E is sufficiently caused or explained by that which is identical to S qua agent (whether that be an event or the agent herself). Condition U essentially says that if E has as its arche another event or state, then S must be personally responsible for that arche if S is to be personally responsible for E. We find it helpful to characterize the language of personal responsibility as found in R*. And, as stated, we think there is a case to be made that R* is anything but foreign to what Kane intends for UR.

The salient point, here, is that R’s compatibility with DUC was supposed to show how Grant’s DUC model could be co-possible with Kane’s UR. But this compatibility holds only if something like the loose interpretation of R holds (and, to be fair to Grant, Kane leaves open the loose interpretation of R in his older statements of UR). However, the loose interpretation of R should be rejected for reasons we’ve now seen, reasons that are both apparently internal to Kane’s purposes as well as more general philosophical reasons. And, importantly, once we abandon the loose interpretation of R, and opt instead for something like ‘personal responsibility’ as found in R*, we find that Grant’s DUC and UR (or a plausible interpretation thereof) are incompatible.

Recall that on Grant’s Dual Sources account, creaturely events have both a divine and creaturely source. Additionally, Grant maintains that for any action that a human performs, God could have caused her to perform a different action in precisely the same circumstances. This is indicated in (g) of EM. So, if Joanie decides to kiss Chachi, this kissing event is caused by both God and Joanie. Hence this event has the following two sources:

(I) Joanie kissing Chachi (or Joanie causing herself to kiss Chachi, or causing the kissing event, or however precisely one would like to analyze the action) in circumstance C,

and

(II) God causing Joanie to kiss Chachi in C, when God could have caused Joanie not to kiss Chachi in C.

Our contention is that while Joanie could be, under certain conditions, personally responsible for (I), in the sense of personal responsibility indicated by R*, she could not be similarly responsible for (II). (Unless otherwise clear from the context, let all subsequent references to ‘personal responsibility’ be understood in R*’s sense.)

To see why we think this, remember that EM includes the following ingredients of divine causation:

• (a) God.

• (b) (I).

• (d) God’s reason for causing (I).

• (e) The causal-dependence relation between God and (I).

• (f) God’s causal act, or causing of (I), which consists in (I) plus the causal relation between God and (I).

• (g) God’s willing or choosing (I), which is nothing else than God’s causing (I) for a reason when God could have done otherwise.

As originally presented, Grant used ‘E’ within (b)-(g) of this schema to denote the occurrence of some generic human action or effect. For our purposes, we have substituted ‘E’ with (I) to denote the event in which Joanie kisses Chachi. With that in mind, let GCA refer to the conjunction of (a), (d), (e), (f), and (g) from the immediately presented way of casting EM with respect to (I). We maintain that Joanie cannot be personally responsible for GCA (again, here and in what follows, in R*’s sense). That is to say, Joanie cannot be personally responsible for the fact that the conjunction of (a), (d), (e), (f), and (g) is true. And if this is so, we argue, Joanie cannot be personally responsible for (II), as would be required for ultimate responsibility for (I).

Let’s start by considering the parts of GCA. Obviously, Joanie cannot be personally responsible for (a), for the fact that there is a God. Whether Joanie is plausibly thought to be personally responsible for (d), (e), (f), or (g) is bound to be more controversial. So, these components require greater attention.

Plausibly, Joanie cannot be personally responsible (d), that is, for God’s reasons for causing her to kiss Chachi. For whether and to what extent God has reasons to act seems to be up to God and God alone—or, perhaps, God plus certain necessarily true moral principles. Think about it this way. Suppose that a father, Julio, has decided to give his son, Enrique, a new car for his birthday. Additionally, suppose that Julio has decided that if Enrique sneaks out of the house late at night to go serenading (something Enrique is prone to do), then this will be a sufficient reason not to give Enrique the new car. Sadly, Enrique sneaks out to go serenading, and, as a consequence, doesn’t get his new car. Should we conclude that Enrique is, in R*’s sense, personally responsible for Julio’s reasons for withholding the car? It’s difficult to see how this could be so. For, it’s not up to Enrique that his late-night serenades provide Julio with the reason he will not give Enrique the car for his birthday. That decision—what will count as the reason for Julio’s deciding not to give Enrique the car—is not something for which Enrique is personally responsible. Plausibly, moreover, the same logic applies to God and any mere human whomever: the human might provide God with reason to act, but what reasons count as decisive for divine action and on which occasions appears to belong to God and God alone.

The point is strengthened when we consider GCA’s (g). According to (g), God may act to bring about (I) for specific reasons. But if He does so, God could have chosen to do otherwise and decided not to act to bring about (I) on the basis of the relevant reasons. So, even if one were to concede that Joanie is personally responsible for providing God with some reason or another for causing (I), this would be a far cry from entailing that Joanie is personally responsible for giving God decisive reasons for causing (I). Quite the contrary, (g) says that the reasons found in (d) are not decisive. We suppose, then, that Joanie could not be personally responsible for (d).

What about (e)? Might Joanie be personally responsible for the causal-dependence relation between God and (I)? That’s unlikely. Suppose that we agree that Joanie is personally responsible for kissing Chachi as described in (I). All the same, we deny that Joanie would then be personally responsible for the existence of the relation found in (e). For, though it is true that there would not be this causal-dependence relation between God and Joanie if Joanie had not kissed Chachi in the relevant circumstance, whether or not there is a causal-dependence relation between God and Joanie having kissed seems to be something only for which God can be personally responsible. At most Joanie is personally responsible for her kissing of Chachi, and the causal-dependence relation between she and her having kissed Chachi on this occasion. Joanie is not, on the other hand, similarly responsible for there being a causal-dependence relation between God and Joanie’s having kissed Chachi—i.e., the relation of God having caused Joanie to kiss Chachi. It seems only God can be personally responsible for that, especially since, given (g), God could have caused Joanie not to perform (I).

Think of it like this. Suppose that our writing this paper causes Tim, Grant’s friend and fellow defender of the Dual Sources account, to write a reply piece. Should we think that Tim’s writing a reply confers on him some personal responsibility for the fact that we caused him to reply? Even if we grant that Tim would be personally responsible for his having replied, it seems implausible that Tim could be similarly responsible for our having caused him to reply; that is, for the causal-dependence relation that exists between our having written this paper and Tim having replied. For similar reasons, we conclude that Joanie cannot be personally responsible for (e).

Before turning to (f) we first examine (g). It’ll be remembered that (g) says that God’s willing or choosing (I) is nothing else than God causing (I) for a reason when God could have done otherwise. But, Joanie cannot be personally responsible for the fact that God causes (I) for a reason when he could have done otherwise. For, it’s up to God (and God alone) whether or not God acts for some set of reasons or another. And it’s (presumably) just a fact about God’s nature that God could have done otherwise than God, in fact, does (if God, in fact, does). So, Joanie cannot be personally responsible for (g).

One reason to think that Joanie cannot be personally responsible for whether God chooses (I) comes via consideration of Grant’s very meticulous doctrine of providence on account of his Dual Sources model. For Grant affirms that, according to his Dual Sources account, God is able to bring about “any state of affairs that is possible,” including each free human action (Grant, 2019, 158). So, in the exact same circumstances in which Joanie kisses Chachi, God could have caused Joanie not to kiss Chachi, and refuse to do so freely. But it’s doubtful that Joanie can be personally responsible for this contra-causal divine capability. For what Joanie does and when she does it, is the effect of God’s causal activity, but not vice versa. That’s to say, (I) bears the property of being the effect of God’s causal activity, (II), but (II) doesn’t bear the property of being the effect of (I). It’s therefore considerably plausible that Joanie isn’t personally responsible for (a), (d), (e), or (g). Now, what about (f)?

Recall GCA’s (f):

(f) God’s causal act, or causing of (I), which consists in (I) plus the causal relation between God and (I).

It might be tempting to suppose that Joanie can be personally responsible for (f) in virtue of being personally responsible for (I). And, importantly, it might be tempting to suppose that Joanie’s personal responsibility for (f) (if such were the case), in virtue of her responsibility for (I), renders her personally responsible for GCA. However, we shouldn’t conclude this.

We deny that Joanie can be personally responsible for (f). For, along the lines previously stated, if we allow that (I) bears the property of being caused by God, then it’s difficult to see how Joanie can personally responsible for (f) just in virtue of being personally responsible for (I). After all, UR, when augmented by R*, says that it’s a necessary condition of personal responsibility that the sufficient explanation or cause of the relevant event or state must originate from that which is identical, in this case, to Joanie, qua agent. Yet God’s causing of (I) certainly doesn’t do that. Hence, even if Joanie is personally responsible for (I), it seems that she cannot be personally responsible for (f), i.e., the way in which God’s causing of (I) is to be analyzed according to Grant’s preferred version of DUC.

In order to counter the above argument, Grant could rely on his aforementioned CP. As indicated, Grant maintains that on EM one can exercise counterfactual power over God’s causing one’s act via CP. The conclusion: “Since whether God’s causing my act occurs is not outside my power, there is no reason on EM to think that whether or not my act occurs is ultimately up to God and not, at least also, ultimately up to me” (Grant, 2019, 70).

By extension, we understand Grant to be effectively proposing that Joanie could be personally responsible for (f) (because ultimately responsible) by way of CP. The thought seems to be something like this. Suppose that Joanie is personally responsible for kissing Chachi (i.e., (I)). But, this fact implies, given DUC, that God causes Joanie to kiss Chachi (i.e., (II)); so, it follows that Joanie is personally responsible for God’s causing her to kiss Chachi. And, because this is so, Joanie can be said to have power (perhaps counterfactual power) over God’s causing her to kiss Chachi.

But, even with CP in hand, if we understand ‘personal responsibility’ in terms of R*, CP’s truth (if it is true) does not suffice to show that Joanie is personally responsible for God’s causing her to kiss Chachi. Here’s why. Even if Joanie has counterfactual power over God’s causing her to kiss Chachi, it doesn’t follow that God’s causing her to kiss Chachi originates from something identical to her as an agent, a condition necessary to meet R*. And, if not, then there’s no reason to think that CP’s truth (if it is true) undermines our above argument.

Nevertheless, suppose that we’re wrong about (f). Suppose, for the sake of argument, that Joanie could be responsible for (f) in virtue of her being personally responsible for (I). Still, we deny that this suffices for making Joanie personally responsible GCA.

To begin to see why we cannot conclude that Joanie is responsible for GCA simply by being responsible for (f), consider the following proposition.

L: That the jar breaks in the kitchen, Hitler commits mass genocide, and thousands die from COVID-19.

Now, let’s suppose that Smith is personally responsible (again, in R*’s sense) for the jar’s breaking in the kitchen. Nonetheless, it’s difficult to see how he could be personally responsible for the truth of L. There might be a number of reasons why this is. Perhaps there is a relevant epistemic condition that Smith fails to meet (we could imagine that Smith did not pay attention in history class, and that Smith is unaware of the severity of COVID-19, and so on) such that he cannot be reasonably expected to have thought about whether or not his actions play any causal role in making L true. And perhaps there are more conditions besides these that Smith fails to meet, maybe something having to do with the fact that two out of three of the conjuncts have nothing to do with Smith at all. Maybe.

But, then, consider the following:

M: That the jar breaks in the kitchen, and that Smith was born.

Again, suppose that Smith is personally responsible for the jar’s breaking. Here, Smith is (we might assume) obviously aware of the truth of M (or, at any rate, that both of M’s conjuncts are true), and that his actions play a causal role in making M true. Moreover, both conjuncts have something to do with Smith. Nevertheless—that is, even though Smith is both aware that M is true (or, at any rate, that both of M’s conjuncts are true), his actions play a causal role in making M true, and that both of M’s conjuncts have something to do with Smith—we find it implausible that Smith is personally responsible for M simply because he is similarly responsible for the truth of one of M’s conjuncts.

Certainly, there are conjunctions for which a person could be personally responsible even if only personally responsible for one of that conjunction’s conjuncts. Consider:

N: That Roger helped write the present paper, and Jordan helped write the present paper.

We think we are (each of us, individually) at least partly personally responsible for N. But we deny that N is like L or M, and we think this is pretty obviously the case. Moreover, we deny that GCA is like N, and, instead, we submit that GCA is like L and M. That is, we think that even if Joanie could be personally responsible for (f) in GCA, it doesn’t follow for a moment that she is also responsible for GCA.

So, here’s where we have arrived. Either Joanie is responsible for (a), (d), (e), (f), or (g) or she is not. We think we have shown that she is not. Thus, we think we have shown that she is not responsible for GCA, either. But, even if we were to grant, for the sake of argument, that Joanie is personally responsible for one of these things, namely, (f), we think we have shown that it doesn’t follow that she is also personally responsible for GCA. In fact, we think that most readers will conclude that GCA is closer to L or M than it is to N in that GCA does not make for a good candidate for that which Joanie may be personally responsible.

The upshot, then, is this. Given Grant’s DUC, we submit that Joanie could be ultimately responsible for (I) only by being personally responsible for some or all of GCA’s parts (i.e., (a), (d), (e), (f), or (g)). But we believe we have provided good reason to suppose that she is not personally responsible for any of the parts of GCA; and even if it were to turnout that she is personally responsible for one part, i.e. (f), she still wouldn’t be personally responsible for GCA. Moreover, we submit that if Joanie cannot be personally responsible for GCA, she cannot be ultimately responsible for (II) when spelled out in terms of Grant’s Dual Sources account. For GCA just is the application of the Dual Source’s EM to (II). And if (II) is the sufficient cause of (I), as would be the case on Grant’s DUC, then according to condition U of Kane’s UR, Joanie must be personally responsible for (II) in order to be ultimately responsible for (I). Yet, as we have seen, Joanie isn’t personally responsible for (II), so she can’t be ultimately responsible for (I).

And, of course, there is nothing unique about the case of Joanie kissing Chachi. What’s true of Joanie would presumably be true of all human acts for which they are supposed to be ultimately responsible. Hence, if both DUC and EM are true, then divine causation introduces an explanation for each human action that is prior and logically sufficient for each action’s occurrence for which no human is personally responsible. Thus (1) looks to be true.

We maintain, therefore, that (1) is more plausible than its denial. However, since we have already agreed to affirm (2), it follows that if DUC and EM are jointly true, no human is ultimately responsible for any action she performs. If this is right, then Grant’s Dual Sources account fails to meet the objective of offering a paradigm in which DUC is compatible with humans being ultimately responsible for many of the actions they perform.

4 Conclusion

We have endeavored to show that Grant’s Dual Sources account (together with its EM and DUC) precludes humans from being ultimately responsible for all of what they do. If we are right, then those who wish to affirm that ultimate responsibility (in what we take to be the most plausible understanding of this term) is necessary to preserve the deep agency human possess are best served by rejecting Grant’s Dual Sources account.