The Ontology of Action and Divine Agency (Do Not Cite Without Permission) Andrei A. Buckareff Marist College [email protected]

1. Introduction The concept of divine agency is central to the narrative traditions inherited by Jews, Christians, and Muslims.

The scriptures of the Abrahamic religions include

repeated references to the intentional actions and intentional outcomes of the actions of God. For instance, in the “Song of Moses” (Exodus 15:1-18), Moses celebrates the freedom of the Hebrews from bondage, declaring that Yahweh is “awesome in splendor, doing wonders” (5:11 NRSV). Alongside the picture of God as an agent who performs actions is a conception of God that developed over the centuries as a simple, immutable, impassable, timelessly eternal being. To many philosophers and theologians, this conception of God as a being who exists in a durationless eternity is inconsistent with the way God is characterized in the scriptures, viz., as an agent who acts in time, often in direct response to the actions of other agents. In this paper, I will assume that if the God of Judeo-Christian-Islamic theism exists, then God is an agent who has performed intentional actions. I will argue that if God is such an agent, then God cannot exist outside of time.1 This is because of how 1

For my purposes, the view of time that is accepted is unimportant. Whether we accept presentism,

eternalism, or growing universe theories makes no difference. On all of these theories an action is still a datable event. Some views of time are more or less amenable to the existence of complex events that are indexed to multiple moments of time; but none of these issues matters much for what I am doing here.

1

action-tokens relate to event-tokens. Specifically, every action-token is identical to an event-token. And every event-token is such that it can be indexed to some moment of time (and, in the case of composite events, successive moments of time). I will call this the ‘action-event identity thesis’, or ‘AE’, for short. If AE is correct, then God acts in time.2 Consider the following reductio in defense of my thesis.3 P1. Necessarily, for any A such that A is an intentional action, there is some event E, such that A is identical to E. (Assumption) P2. Necessarily, for any E such that E is an event, there is some time t such that E can be indexed to t. (Assumption) C1. So, necessarily, for any A such that A is an action, there is some time t such that A can be indexed to t. (from (P1) and (P2)) P3. God exists outside of time. (Assumption for reductio) P4. If (P3), then for any DA, such that DA is a divine action,4 there is no time t such that DA can be indexed to t. 2

I am assuming that the control relation an agent bears to her actions when exercising intentional agency is

such that it is concurrent with the action itself. If it is not, then the event that occurs is better regarded as the intentional outcome of an action and not the action itself. I offer reasons for why this assumption is justified in section 3.1. 3

It is worth noting that Hoffman and Rosenkrantz (2002, pp. 102-6) offer an argument similar to mine that

goes from the claim that God acts to God being a temporal agent. However, they do not explore the ontology of action in any detail. Thus, they leave many of their action-theoretic assumptions undefended. I should note that I became aware of Hoffman and Rosenkrantz’s argument after I had already written my paper. 4

I assume that any divine action is an intentional action. So adding ‘intentional’ before ‘divine action’ is

superfluous.

2

P5. If for any DA there is no time t such that DA can be indexed to t, then divine actions are not events. P6. If DAs are not events, then not all actions are events and so not all actions can be indexed to a time. P7. But, necessarily, all actions are events that can be indexed to a time. (from (C1)) C2. So it is not the case that God exists outside of time. (from (P3)-(P7)) C3. So God is a temporal agent. (from (P3) and (C2)) In what follows I will first focus on defending (P1) and (P2), given that the case against God as acting outside of time largely rests on the truth of these two premises. So I will defend AE. I will argue that the practical upshot is that the concept of God as an atemporal agent is untenable. I will then consider two objections to my case against atemporal agency.

2. Actions are Events: The Case from the Ontology of Action It would make my task easier if there were agreement among philosophers of action regarding the relationship of actions to events. While the current orthodoxy takes AE for granted, there are heretics in action theory. One argument against the current orthodoxy goes something like the following.5 While perhaps not the best criterion for the identity conditions of actions, if we assume the truth of Leibniz’s Law then some action A and some event E are identical if and only if for any property P, A has P if and only if E has P. For instance, the action of my moving my left arm and the concurrent event of my

5

This argument is based on the arguments against AE found in Alvarez and Hyman 1998, Bach 1980,

Ruben 1997 and 2003, and von Wright 1970.

3

left arm moving have different properties. We can distinguish the action from the event by some of the following properties. The action has properties we associate with agency, while the properties of the event are wholly neurobiological and could have obtained absent any exercise of agency on my part. While the event may be the intentional outcome of my action, it is not itself what I exercise executive control over in the way indicative of agency. This is evinced by the transitive form and intransitive form of the verb ‘move’ in the following sentences. Compare, ‘I move my arm’ to ‘My arm moves’. The transitive form of the verb is indicative of agency while the intransitive form is ambiguous at best. In any case, use of the transitive form of verbs is characteristic of action and distinguishes action sentences from sentences that describe events. Assuming these results generalize to all actions, while actions and what we may refer to as merely behavioral events share some properties, they do not share all the same properties. So actions and events belong to different kinds. Thus, AE is false. The foregoing argument against AE is not sound. Just because my arm moving and my moving my arm are not identical, it does not follow that my moving my arm is not an event. So it is not obvious that actions are not events. In the remainder of this section I will defend AE by arguing for (P1) and (P2) of the reductio argument against timeless agency. I will apply the consequences of my case for AE to the concept of divine agency at the end of this section. I will start by defending (P2): Necessarily, for any E such that E is an event, there is some time t such that E can be indexed to t. The status of events as a fundamental ontological category is not uncontroversial, although both folk practice and philosophical consensus favor an ontological commitment to events. Roughly, the major analyses of

4

events regard events as temporally located happenings, occurrences, or changes in the world. Qua changes, barring problems with vagueness and the individuation of events, events are taken to be datable, with each event-token being such that it is indexed to a moment or some moments of time. So events are temporally located.6 And if it is true of every tokening of an event-type that it is temporal, and actions are identical to events, then it is the case that every action is such that it is indexed to some moment(s) of time.7 Most defenders of events defend either property-exemplificationist theories (henceforth, PET) or non-exemplificationist theories (henceforth, NET).8 Defenders of NET have offered anemic accounts of the existence conditions for events. Well-known accounts simply assume that events are temporally located particulars and only focus on providing identity conditions for events that enable us to individuate events.9 In some of his earlier work on the identity conditions for events, Donald Davidson (1980) comes close to offering existence conditions for events. He grants that events are changes in objects, specifically, substances (1980, pp. 174-75). But he says little about the nature of 6

Many event theorists would add that events are spatially located (see e.g., Lombard 1979). For my

purposes, I need only assume they are temporally located while the effects of divine actions in the world are spatio-temporally locatable. 7

I will later take up an objection (in section 3) to the claim that events are necessarily temporal.

8

These do not exhaust all of the options with respect to event theory. Both theories take events to be

constituents of reality and they take them to be abstract particulars. They are abstract because multiple events can occur in the same location and they are particular because they do not recur. For a defense of the thesis that events are universals, see Chisholm 1970. For defenses of the thesis that events are concrete particulars, see Quine 1960 and Davidson 1985. For a defense of eliminativism about events see Horgan 1978. 9

Brand 1984 and Davidson 1980 defend versions of NET.

5

these changes and resists developing an account of the existence conditions of events in terms of changes in objects. So we get little from Davidson and other prominent defenders (e.g., Brand 1984) of NET by way of existence conditions of events. What is important for my purposes is that defenders of NET regard events as temporally located. The identity conditions on offer from versions of NET make this clear. For example, in his earlier work on events Davidson offered the following identity conditions for events (1980, p. 179). (NEIC) Necessarily, for any x and y, x and y are the same event iff x and y have exactly the same causes and the same effects. This is not very satisfying.10 Davidson recognized this later and abandoned his causal theory of event identity (see Davidson 1985). Other prominent NET theories that offer identity conditions focus on events being identical that necessarily occupy the same spatiotemporal location (e.g., Brand 1984). Such an approach fails to individuate events from objects such as substances. Moreover, it fails to account for the possibility that two or more non-identical events can necessarily occupy the exact same spatiotemporal location while being distinct because of their etiology or some other extrinsic property. I will ignore this approach. What is important for my purposes is that prominent NET theories of events regard events as temporally located. By offering little by way of the existence conditions of events, we have little reason to take it for granted that events are

10

This view implies that there cannot be two or more events that (i) are causeless but with the same effects,

(ii) are effectless but with the same causes, or (iii) are neither caused nor have any effects (Lombard 1998, p. 287).

6

necessarily temporally located. NET seems to simply report a widely held commitment to the temporal locality of events. Versions of PET and the arguments offered in their favor strike me as more promising than what has been offered to date by those who favor NET. If we suppose that any of the proposals for the identity conditions for events offered by defenders of NET just considered are satisfactory, the proponent of PET can avail herself of those same conditions if she wants, and she can provide more informative and satisfying identity conditions that issue from the existence conditions for events she offers.11 For instance, a defender of PET can agree that any events with exactly the same causes and effects are identical. But she can say more about why those events are identical. With respect to the myriad different versions of PET, I will avoid taking up some of the inhouse debates among defenders of PET. On some versions, events look an awful lot like facts,12 while other accounts leave it open that there are events that are changes that are not changes in anything at all.13 In the interest of perspicacity and avoiding some of the problems with more controversial versions of PET, I take events to fundamentally involve some change in the exemplifier of a property at a time.14 I assume that the

11

Kim 1976 makes this point.

12

This is the complaint leveled against Kim by Bennett 2002, Brand 1984, Hendrickson 2006, and

Lombard 1979. 13

Lawrence Lombard treats Bennett’s trope-theory of events as a species of the property exemplification

view and claims that it faces this difficulty (see Lombard 1998, p. 288). 14

The PET account of events offered here approximates the views of Lawrence Lombard (1979 and 1998).

Unlike Lombard I am avoiding talk of the movement of objects through “quality spaces” that occurs when they undergo some change in exemplifying a property at a time (see Lombard 1998, pp. 289-90).

7

exemplifying of a property is such that it involves a change in the exemplifier. An object moves, so to speak, from exemplifying some property to exemplifying another. So a very rough PET schema for the existence conditions of events that I believe should not be too controversial is the following: (PEEC) Necessarily, for any object x and any time t, some y is an event at t iff y is a change that occurs in x from exemplifying some property P to exemplifying some property Q at t.15 So we get the following identity conditions for events on this account: (PEIC) Necessarily, for any x and y, x and y are the same event at t iff x and y are each a change that occurs in exactly the same object S from exemplifying some property P to exemplifying some property Q at t. I do not wish to claim that either of these identity or existence conditions for events are wholly satisfactory or should not be critiqued and perhaps modified. I am sure they are open to lots of objections. But they are satisfactory for my purposes in this paper. What is important for my purposes is that the temporal locality of events should be evident on this type of theory of events. Moreover, events can be distinguished from other things that are temporally located. They are changes that occur when an object goes from exemplifying one property to another. It is worth noting that PET allows for relational events. If relations are dyadic properties, then the exemplifying of a relation is itself an event. For instance, a brick’s causing the breaking of a window involves changes with respect to the intrinsic and extrinsic properties of both the brick and the window. When such a causal relation

15

See Bennett 2002, Goldman 1970, Hendrickson 2006, Kim 1976, and Lombard 1979 and 1998.

8

obtains there is some change of the relevant sort that occurs that is distinct from both the causing event and the event that is the effect. The event would be the exemplifying of a causal relation between the relevant relata. This may seem trivial, but as we shall see, it will be of some importance later. I hope it is evident that the fact that events involve some change is indicative of their temporality. To conclude that events are temporally located on the basis of the foregoing may strike some as so much bootstrapping. However, (P2) seems safe for now if for no other reason than the standard, most widely accepted theories of events take events to be occurrences that are temporally located, involving some sort of change.16 If my interlocutor has both a commitment to events as a fundamental category in her ontology and wishes to reject the claim that they are temporally located, then if she wishes to falsify (P2) in my argument, she owes us some arguments for why the most widely accepted accounts of events are mistaken in taking events to be temporally located. For the time being, I will take the claim that events are temporal as prima facie justified and (P2) as true. But I will consider an objection to (P2) and (P5) in the next section. Turning now to a defense of (P1): Necessarily, for any A such that A is an action, there is some event E, such that A is identical to E. To many, AE is just obvious. It certainly is an assumption in action theory that has rarely been explicitly defended.17 But it seems to be a reasonable assumption. After all, if PET is correct, when a person acts, 16

We should expect similar results if events are universals. The instantiation of an event would be

temporally located. So long as P2 is read liberally, this would be enough to show that P2 is true even on a theory of events such as Chisholm’s. 17

The only explicit defenses of AE I know of are in Davidson 1980, pp. 113-14 and Zimmerman 1995.

9

she exemplifies some property (an action-type) that brings about a change in both the agent and in the world. So actions look like events. If actions are not events, the set of action-types constitute a fundamental sui generis ontological category alongside events. Time and space constraints will not allow me to offer the detailed defense of AE that it deserves. But the argument I shall offer should suffice to make a prima facie case for AE. For brevity’s sake, in the remainder of this section, I will offer a PET argument for the claim that actions are events. Then I shall apply the results to divine action. Why should we think that actions are identical to events? Consider the following example of a bodily action: Suguro’s kicking of a ball at 12:30 p.m. The relevant actiontype that is exemplified when Suguro acts is ‘kicking’. It should be fairly uncontroversial that Suguro’s kicking is temporally located since he did it at 12:30. But this does not guarantee that it is an event. After all, objects and other entities have temporal locations. But Suguro’s kicking is an event. Suguro qua subject of the action undergoes some change by exemplifying kicking. For example, he goes from exemplifying one property (e.g., running) to exemplifiying another property (viz., kicking). Suguro’s kicking is done intentionally and is under his control in the way appropriate for an action. Moreover, if we assume that there is an event in the same spatiotemporal location as Suguro’s kicking that is non-actional—e.g., the mere movement of his leg—it is not the case that the event is identical to Suguro’s kicking. The defender of AE does not have to defend the thesis that Suguro’s kicking and the mere movement of his leg are identical.

The mere

movement of Suguro’s leg occurs because he kicks. It is not the case that Suguro exemplifies the same properties when his leg moves as when he kicks. For one, the movement of his leg does not have the extrinsic property of being the intentional object

10

of Suguro’s intention to kick the ball and it does not lie in the motivational potential of his intention. His leg’s movement admits of a wholly neurophysiological explanation and is not what we aim at explaining by referring to his intentions. So the defender of AE does not have to be committed to the claim that there is an event that is the movement of Suguro’s leg that is identical to his action of kicking. They are not the same event under two different descriptions. So assuming that PET is correct, Suguro’s action of kicking the ball is an event. It certainly fits the description of an event. If all other actions resemble Suguro’s kicking, then I believe we have good prima facie reasons for accepting the thesis that all actions are events. All actions do resemble Suguro’s kicking with respect to the relevant features that make them events. If they do not resemble Suguro’s kicking, we need reasons from the opponent of AE for why the contrary impression about actions is mistaken. In any case, I believe we have good prima facie reasons for the truth of (P1). If the foregoing result is correct, we can generalize these results to divine actions: for instance, God’s freeing of the Hebrews from bondage in Egypt. If God freed the Hebrews, then the action of freeing occurred at some time in the past. In freeing the Hebrews God undergoes some change by exemplifying an action type, viz., ‘freeing’ or ‘liberating’. The relevant qualities of God’s action that would make a difference for whether or not it is an event are no different from Suguro’s. The action God performs by the freeing of the Hebrews is an event. Moreover, God undergoes some change by acting just like Suguro does. Both perform temporally located actions that are identical to events. Moreover, both God and Suguro are temporally located agents. God may be omnitemporal, but God is no less temporal than Suguro. If this is correct, then the

11

concept of God as an atemporal agent who performs intentional actions is untenable. This is the case even if God only performs one action. An action involves a change in the agent who acts. God cannot be timelessly eternal and act.

3. Objections I will consider two objections the defender of atemporal agency can offer. Of course, there are more. But addressing these two objections is necessary in order to make a strong prima facie case against atemporal divine agency. The first objection is based on arguments found in the action theoretic literature to the conclusion that actions are not events. The second objection assumes that actions are identical to events; but the claim that events are necessarily temporal is questioned. 3.1. Objection 1: An action is either a causal relation or a sui generis kind With respect to (P1), there are at least two options that appear in the action theoretic literature that defenders of atemporal agency can avail themselves of if they wish to reject AE. The first approach takes actions to be instances of a type of causal relation. Specifically, an action relates an agent or some proper part of an agent to an event. For instance, Kent Bach argues that an action is “the relation of bringing about (or making happen), whose terms are agents and events” (Bach 1980, p. 114). Georg Henrik von Wright claims, “An act is not a change in the world. But many acts may quite appropriately be described as the bringing about or effecting (‘at will’) of a change” (1970, p. 307). This feature of actions is evinced by the use of transitive verbs to describe actions in uncontroversial action sentences. Recall the earlier contrast between ‘I move my arm’ and ‘My arm moves’. The former sentence describes an action while

12

the latter describes some behavior that is not obviously actional. ‘Move’ in ‘I move my arm’ is a transitive verb that expresses a relation between myself and the event of my arm moving when I exercise executive control over my behavior. The defender of atemporal divine agency would do best not to avail herself of this type of response to AE. If we assume that this approach to denying AE is successful, it would still be the case that God must be temporal. This is the case because, to return to the example of God freeing the Hebrews, the action relation that would have obtained when God freed the Hebrews is still necessarily a datable occurrence. In any world, when an agent acts, the relation that obtains between the agent and the intended event that is the outcome can be indexed to some moment of time. An atemporal causal relation would be less troubling if the cause and effect are atemporal. But we are concerned with God bringing about temporally located changes in the universe. How an atemporal cause would relate to a temporal event that is its effect is mysterious. The relation would somehow cross an atemporal/temporal divide. Qua causal relation, the action would be both temporal and atemporal. This is paradoxical at best, logically impossible at worst. Defenders of atemporal agency do little to help their position by accepting a relational theory of divine action. Before moving on to another alternative to AE, it is worth noting that defenders of AE can accept the thesis that actions are relations. Recall that PET allows for relational events. If relations are dyadic properties, then the exemplifying of a relation is itself an event. The causing relation that obtains qualifies as an event on a PET analysis of events. What is perhaps more telling is that, as Michael Zimmerman notes, “causings can and sometimes do themselves enter into causal relations” (1995, p. 590). God’s causing the

13

liberation of the Hebrews caused the Hebrews to celebrate God’s actions in song. The causing in this instance seems to stand in an event-causal relation to the event of the Hebrews celebrating. So if actions are relations, they are still temporally located. And we have good reason to regard them as events. So AE is safe and atemporal divine agency appears no more promising if we accept the thesis that actions are relations. Maria Alvarez and John Hyman offer a strategy that is more promising than the relational approach. On their view, an action is “a causing of an event by an agent” (1998, p. 233). This sounds similar to the relational view. However, on Alvarez and Hyman’s theory of action, actions are sui generis. They are not reducible to events or relations. To act, on their view, “is to exercise a causal power—to cause, bring about or effect an event” (1998, p. 233). An action is of some kind x iff its result is of some corresponding kind y (Alvarez and Hyman 1998, p. 233). So, for instance, a killing is a causing of death.

And the killing is an action in virtue of its result being of a

corresponding kind, viz., a death. Notice that this view is tightly wed to a version of agent-causalism. The agent enters directly into a causal relation by acting, bringing about an intended outcome. Agents do not cause their actions on this view; but they cause events by acting. The action is itself the causing of an event by an agent (1998, p. 224).18 Alvarez and Hyman’s version of agent-causalism that treats actions as a sui generis kind might make room for atemporal agency. I will ignore the problems with this theory of action. I find the relation an agent bears to her actions and the events she causes to be no less abstruse than what we find on other versions of agent-causalism.

18

John Bishop has suggested to me that if Alvarez and Hyman are correct, then (P4) of the original

reductio is false.

14

Moreover, I am not convinced that their theory of action proves to be a threat to AE. For now, it is worth noting that Alvarez and Hyman’s view is not really all that helpful for the defender of atemporal divine agency, at least not without making significant modifications to the theory of action. Alvarez and Hyman claim that actions have spatiotemporal location (1998, p. 242). If an action has a temporal location, then, in virtue of the relation an agent bears to her actions on their theory of action, an agent must occupy the same temporal location as her action. To make matters worse for the defender of atemporal agency, on Alvarez and Hyman’s theory of action, while the location of actions may be imprecise, “if an action was performed at such and such a time and in such and such a place, then its result occurred at the same time and in the same place” (1998, p. 243). If God is an atemporal agent, then it seems that if Alvarez and Hyman’s theory of action is correct and actions are not events it turns out that all of the events God causes by acting would be atemporal.19 But as the Judeo-Christian-Islamic narrative traditions attest, the effects of God’s actions are temporal. The liberation of the Hebrews is an event that is datable, at least in principle. So if we assume that an agent-causal theory of action such as Alvarez and Hyman’s is correct, God’s actions are also datable and so God is a temporal agent. (P1) appears to be safe. 3.2. Objection 2: Events are not necessarily temporally located The defender of God’s atemporality may concede that actions are events. Moreover, he may even claim that events in the universe are necessarily temporal (call 19

Stewart Goetz suggested to me that if the defender of timeless agency relies on an agent-causal view of

action, then it is plausible to think that human agents remain outside of time when performing intentional actions since God can perform timeless actions. But, of course, we are temporal agents. So such a consequence of atemporal agent-causation is counterintuitive (at best).

15

these ‘worldly events’). But in spite of all of this, it may be possible that there are actions, and, hence, events, that are atemporal.20 Consider the following line of argument.21 I claimed that events are necessarily temporal and that the temporal location of events is due to the fact that an event involves some change in the world. But my interlocutor may argue that events do not necessarily involve change. Suppose that Amy is meditating, focusing on a feature of one of Andrei Rublev’s icons. She performs the mental action of dwelling upon this feature of the icon from t to t’. At t a change occurs, Amy goes from not exemplifying the act-property of dwelling to exemplifying it. At t’ Amy once again undergoes some change. She is no longer dwelling. At t and t’ events occur; and the events involve some change.

But what occurs between t and t’?

Assuming that t and t’ are not identical and that there is some point of time that is later than t and earlier than t’, an event occurs between t and t’. Amy is acting, my objector would claim, between these moments of time. So an event occurs. But there is no change during that time. At least, it is not obvious that there is some change that occurs when Amy dwells upon the one feature of the icon. If this is correct, then not all actions and, thus, events involve some change. If not all events involve change, then it is possible that there are atemporal events. And so it is possible for God to act outside of time. So (P2) and (P5) of the reductio against atemporal agency are false. Suppose I grant that what occurs from the onset at t of Amy’s dwelling on the icon to the cessation at t’ of her activity is an action. The objection seems to assume that it is

20

These would not be worldly events.

21

I owe this objection and the following example and argument to Hugh McCann.

16

a composite action that has among its proper temporal parts a simpler action that involves no change.22 The simpler action is sandwiched between the onset of acting and the cessation of acting. All three simpler actions are proper parts of one complex action of dwelling on the icon. 23 If we suppose the foregoing, a story can be told about the event of dwelling that involves some change. For instance, Amy changes from having an antecedent present-directed intention which triggers the action at t to having an intention in action that sustains her action between t and t’ (or at least the content of her intention to act changes to express a proposition about how she is acting). Moreover, I assume that if Amy is acting, she continues to make some sort of effort to dwell on the icon.24 The effort she exerts between t and t’ can be distinguished from the effort she exerted when she commenced dwelling. She undergoes some change from commencing an effort to sustaining it in dwelling. We may even locate additional changes that occur in Amy’s effort to continue dwelling between t and t’. There is at least one so-called Cambridge change that occurs. Specifically, Amy changes from dwelling on the icon at one moment of time to dwelling on it at another time. So Amy undergoes a change in one of her extrinsic properties that makes it no longer true that she is dwelling on the icon at the preceding moment of time; and she undergoes a change that makes it now true that she is dwelling on the icon at 22

The theory of action individuation being assumed is a version of the composite theory of action

individuation that allows for complex actions composed of simpler actions. See Ginet 1990, McCann 1998, and Thalberg 1977 for defenses of this view. 23

It is not obvious that all of the proper parts of a composite action must themselves be actions. I have

argued that they need not all be actions in Buckareff 2005. 24

I discuss the role of efforts or tryings in theorizing about mental action in Buckareff 2005 and 2007.

17

another moment of time. I recognize that a risk of appealing to such Cambridge changes is that it will turn out that a number of things we did not think were events are events. Still, it is not obvious that such Cambridge events should not be counted as real changes that are constitutive of real events. In any case, if Amy acts between t and t’, some change occurs. If this is right, then the temporality of events thesis is still safe. The upshot of the foregoing is that it is not obvious that temporal events can occur without some change occurring. If I am right, then the possibility of atemporal events and atemporal divine agency is still in doubt. So (P2) and (P5) are safe.

4. Conclusion If God is an agent who performs intentional actions, then God cannot be timelessly eternal. If God exists, God must be sempiternal or omnitemporal, but not atemporal. I have purposely remained neutral with respect to whether or not this means that classical theism of any stripe is untenable.25 It should be clear that God could not be 25

John Bishop has suggested that classical theism may be in trouble if I am right. For instance, it seems

difficult to make sense of creatio ex nihilo if God cannot perform atemporal actions (this same point was made by audience members when I read a version of this paper at the 2007 British Society for the Philosophy of Religion meeting). One way to avoid this problem is to take God as agent-causing God’s first creative act of creating. Before that, God did not act. But once God acts, there is at least one moment of time. A difficulty with this view is that it requires that the agency theory be true with respect to divine action. I believe there are good reasons to reject the agency theory as a theory of human action because it is needlessly obscure and mysterious. If events are not necessarily temporally located, then we cannot tell a story of divine action consistent with the standard story accepted by many action theorists today (viz., the causal theory of action). So either an agent-causal relation obtains between God and God’s first act (in creating) or the standard story of action applies to God and we reject creatio ex nihilo. Since I take human

18

an agent in the sense in which God is portrayed in the Judeo-Christian-Islamic narrative traditions if God is timelessly eternal.

I believe this implies that God cannot be

immutable or impassable. But that requires a separate argument. Taken together, these claims count against classical theism in its stronger forms and count in favor of various alternatives.26 In any case, an upshot of the central argument of this paper is that varieties of theism that conceive of God as an atemporal agent appear untenable.27

Works Cited: Alvarez, M. and Hyman, J. (1998) Agents and their actions. Philosophy, 73, 219-45. Bach, K. (1980) Actions are not events. Mind, 89, 114-20. Bennett, J. (2002) What events are. In R. Gale (ed.) The Blackwell guide to metaphysics (pp. 43-65). Oxford: Blackwell. Borghini, A and Varzi, A. (2006) Event location and vagueness. Philosophical Studies, 128, 313-36.

action as my starting point in theorizing about action and agency more generally, I am willing to bite the bullet and reject creatio ex nihilo (or at least the doctrine as it is traditionally articulated). 26

With further argument, it may be rational to accept some version of panentheism. This is the case

because if God is temporal it can be argued that God must also be spatial. See Helm 1980. For an attractive version of panentheism see Forrest 2007. 27

Thanks to John Bishop, Stewart Goetz, John Knight, Hugh McCann, and Henry Pratt for helpful

comments on an earlier draft of this paper. I am also grateful for useful feedback from the audience in the session at the 2007 British Society for the Philosophy of Religion meeting at Lady Margaret Hall, Oxford University where I read a shorter version of this paper.

19

Brand, M. (1984) Intending and acting: Toward a naturalized action theory. Cambridge, MA: MIT Press. Buckareff, A. (2005) How (not) to think about mental action.

Philosophical

Explorations, 8, 83-9. Buckareff, A. (2007) Mental overpopulation and mental action: Protecting intentions from mental birth control. Canadian Journal of Philosophy, 37, 49-66. Chisholm, R. (1970) Events and propositions. Nous, 4, 15-24. Davidson, D. (1980) Essays on actions and events. New York: Oxford University Press. Davidson, D. (1985) Reply to Quine on events. In E. Lepore and B. McLaughlin (eds.) Actions and events: Perspectives on the philosophy of Donald Davidson (pp. 17277). Oxford: Blackwell. Forrest, P. (2007) Developmental theism: From pure will to unbounded love. New York: Oxford University Press. Ginet, C. (1990) On action. Cambridge: Cambridge University Press. Goldman, A. (1970) A theory of human action. Upper Saddle River, NJ: Prentice-Hall. Helm, P. (1980) God and spacelessness. Philosophy, 55, 211-21. Hendrickson, N. (2006) Towards a more plausible exemplification theory of events. Philosophical Studies, 129, 349-75. Hoffman, J. and Rosenkrantz, G. (2002) The Divine Attributes. Oxford: Blackwell. Horgan, T. (1978) The case against events. The Philosophical Review, 87, 28-37. Kim, J. (1976) Events as property exemplifications. In M. Brand and D. Walton (eds.) Action theory (pp. 159-77). Boston: D. Reidel. Lombard, L. (1979) Events. Canadian Journal of Philosophy, 9, 425-60.

20

Lombard, L. (1998) Ontologies of events. In S. Laurence and C. Macdonald (eds.), Contemporary readings in the foundations of metaphysics (pp. 277-94). Oxford: Blackwell. McCann, H. (1998) The individuation of action and the unity of agency. In H. McCann, The works of agency: On human action, will, and freedom. (pp. 36-57). Ithaca: Cornell University Press. Quine. W. (1960) Word and object. Cambridge, MA: MIT Press. Ruben, D-H. (1997) Doing without happenings: Three theories of action. In G. Holmstrom-Hintikka and R. Tuomela (eds.), Contemporary action theory, Vol. I (pp. 267-86). Dordrecht: Kluwer. Ruben, D-H. (2003) Action and its explanation. New York: Oxford University Press. Thalberg, I. (1977) Perception, emotion, and action. New Haven, CT: Yale University Press. Von Wright, G.H. (1970) The logic of change and action. In M. Brand (ed.) The nature of human action (pp. 302-30). Glenview, IL: Scott, Foresman and Co. Zimmerman, M. (1995). Actions and events. Journal of Philosophical Research, 20, 585-94.

21