Causal pluralism: agent causation without the panicky metaphysics

Abstract

An important divide in the free will literature—one that is arguably almost as common as the distinction between compatibilism and incompatibilism—concerns the distinction between event and substance causation. As the story typically goes, event-causalists maintain that an action is free only if it is caused by appropriate mental events, and agent-causalists maintain that an action is free only if it is caused directly by a substance (the agent). This paper argues that this dichotomy is a false one. It does this by introducing a new view called Causal Pluralism, which maintains that free will is compatible with both event and substance causation. Furthermore, it is argued that agent-causalists have good reason to adopt Causal Pluralism, and also that that the view has interesting implications for the free will dialectic.

1 Introduction

Within the free will literature, there exists a deep divide between event- and agent-causalists. Simplifying things somewhat, event-causalists maintain that a free action is caused by appropriate (mental) events. In contrast, agent-causalists maintain that a free action is caused, not by any events, but by a substance—namely, the agent. So construed, agent-causalists maintain that substance causation is necessary for free will, while event-causalists maintain that event causation is necessary for free will. A central point of contention within the free will dialectic, then, surrounds whether substance or event causation is necessary to secure free will.

It is often assumed that the choice to include agents as fundamental causal relata marks a radical departure from the standard event-causal framework—so much so that, for the last few decades, the postulation of fundamental substance causes has often been taken as sufficient grounds to dismiss agent-causalism as a credible theory of free will. Because of this, event-causalist views have solidified themselves as the orthodox position within the free will literature.

In this paper, I introduce and motivate a new view on free will, which I call Causal Pluralism. To foreshadow, Causal Pluralism is the view that free will is compatible with both (fundamental) event and substance causation. Indeed, as I develop the view here, free will is compatible with both kinds of causal relata because event causation can accommodate much of the metaphysics of substance causation. I show this by considering the nature of events more closely. In particular, I argue that, if the standard agent-causalist wants to reject the Causal Pluralist view I develop here, this requires showing that there is a freedom-relevant difference between event causation (properly construed) and substance causation. In addition to this, I explain how Causal Pluralism has the advantage of capturing most of what extant agent-causalists have wanted to say about free will, all while making fewer ontological demands. By doing all of this, I hope to show that agent-causalists cannot be so quick to dismiss event causation.

Still, even if one is not ultimately persuaded by the view, considering the Causal Pluralist proposal helps to draw out at least two important points. First, there is a lack of clarity regarding how the two types of causal relata—i.e., events and substances—relate. Second, getting clearer about how the two causal relata relate suggests that, in some important respects, the divide between event- and agent-causalists may be more superficial than it initially seems.

Here is the plan for the rest of the paper. Section 2 starts by providing a bit more detail regarding the event- and agent-causalist views, with a focus on how they differ. Following that, Sect. 3 explains why there is a false dichotomy when choosing between event and substance causation in a theory of free will. The Causal Pluralist view is introduced as a way of drawing out the false dichotomy. Section 4 turns, then, to providing further motivation for why extant agent-causalists should take Causal Pluralism seriously. Because the initial motivation for Causal Pluralism in Sects.  3 and 4 relies heavily on a property exemplification model of events, Sect. 5 returns to examining the nature of events. More specifically, it is argued that Causal Pluralism does not stand or fall with a property exemplification model of events. Section 6 finishes with some reflection on Causal Pluralism’s relation to agency reductionism and event-causalism. An important upshot of these remarks is that event causation and agency reductionism can be prized apart.

2 The dichotomy

Understanding the dispute between event- and agent-causalists will require a brief excursion through both campgrounds. While there is much to be said for and against each of these positions, my aim in this section is not take to a stand on which position should be endorsed. Rather, it is merely to lay out the positions in order to show that an event-substance divide exists. In the next section, I explain why the event-substance divide may be less important than initially thought.

2.1 Event-causalism

Before beginning, in order to preempt any confusion, it is important to note up front that the accounts which will be the focus of our discussion will be those that accept the control condition for free will, especially in its causal form. So, on this assumption, if an agent’s action is to be free, the agent must stand in some suitable causal relation to that action. With that said, a view is a species of event-causalism when it seeks to explicate the control condition on free will wholly in terms of events causing one another. This would include both compatibilist and incompatibilist theories of free will.¹ While a number of different accounts fall under this heading (e.g., Franklin, 2018; Fischer, 1999), the various accounts appear to build upon a common core:

Event-causalist condition: An agent freely performed some action $\phi$ only if $\phi$ was caused by mental events involving the agent (e.g., the agent’s reasons).

Though not essential to these views, it is commonly assumed that directly free actions are mental actions, such as the formation of an intention or the making of a decision. On this position, bodily movements—such as raising an arm or pulling a trigger—are derivatively free; that is, they are free in virtue of being caused by (directly) free mental actions. Though not much should hinge on this, for ease of exposition, I will presume for simplicity that bodily movements (like the mental actions that cause them) are directly free.

With that said, consider a simple example. Suppose that Angelica makes an utterance. According to the event-causalist condition, Angelica’s utterance was freely performed only if it was caused by mental events involving Angelica. As indicated in the description of the event-causalist condition, it is typically assumed that the appropriate mental events are the agent’s reasons.

There are at least two ways to understand what the reasons mental state consists in. On the standard story, following Davidson, reasons are reduced to two more basic mental events: a belief-desire pair. So, for instance, Angelica’s having a reason to make the utterance consists in her having both a desire to make the utterance and some kind of belief that her utterance will accomplish a particular goal. This, we might note, invokes an internalist account of reasons insofar as it posits that the appropriate reasons are internal to the agent.

An alternative account—one that is more sympathetic to externalist accounts—grants that reasons are, perhaps, something like considerations that count in favor of a particular action, but goes on to add that the appropriate mental event for the event-causalist condition is the agent’s taking something to be a reason. If some reason R is a consideration in favor of making an utterance, then Angelica’s taking R to be a reason can play the requisite role for the event-causalist (Franklin, 2018).

According to event-causalism, (the control condition on) free will consists in events causing one another. On the one hand, we have mental events playing the role of causes, and, on the other hand, we have actions (whether mental or bodily [i.e., physical]) playing the role of effects. Whatever role the agent plays in the performance of free action, it will be mediated by their relation to these events. Furthermore, note that, on event-causalism, it is a necessary condition on free will that a free action be caused by a (mental) event.

The nature of events is generally left unanalyzed, even for event-causalists; and we will do so as well for the time being. Of course, later on, we will have more to say about the nature of events. It should be sufficient for our purposes here to describe events at a high level of generality. Events, we may say, then, are simply “happenings” or “things that occur.” They may also be further picked out with common examples such as the tree’s falling, Old Yeller running a mile, and Juan’s wanting a drink.

Applying what we’ve said to a case of free action, the relevant events and their causal relation would look something like this: when Angelica has freely made an utterance, then Angelica’s having a reason to make the utterance has caused Angelica’s making the utterance.

2.2 Agent-causalism

Agent-causalism denies that the control condition on free will is explicable wholly in terms of events causing one another. Instead, they posit an irreducible agent as a cause. Just as with event-causalism, I am understanding agent-causalism to include both compatibilist and incompatibilist theories (e.g., Nelkin, 2011; Markosian, 1999; Jacobs & O’Connor 2013). Here is the core of such accounts:

Agent-causalist condition: An agent freely performed some action $\phi$ only if $\phi$ was directly caused by the agent (qua substance).

Agent-causalism requires that a free action not be fully causally explained by any event(s), but rather the explanation must make some essential appeal to the agent’s direct causal involvement. What it means for the agent to directly cause an action is for the agent to (literally) be the causal relatum that stands in causal relation to the action. So, instead of saying that Angelica’s freely making an utterance is explained in terms of an event causing Angelica’s making an utterance, the agent-causalist contends that Angelica herself caused the event of Angelica’s making an utterance. What we have here is a case of a substance standing in causal relation to an event.

Substances, for the agent-causalist, are said to correspond roughly to our conception of ordinary objects, and further described as “concrete particulars.” So, for example, cups, chairs, trees, tigers, and, of course, people, would all count as substances in this picture. Most proponents of agent-causalism do not provide a deeper analysis of what makes some complex entity a substance, as opposed to a mere aggregate, though a number of proposals are available (see, for example, VanInwagen,  1990; Markosian, 1998; Merricks, 2001). Whatever criteria agent-causalists may come down on, they are at least committed to the position that agents are substances.

It is important to note that the agent-causalist need not posit any new kind of causation. Formally speaking, the causal relation that an agent stands in to an event is no different than the causal relation one event stands in to another event; it is the very same kind of causal relation that obtains when one event causes another event. The only difference is that the causal relata differ in each case (Clarke, 2003, p. 186).

When the agent directly causes an action, this is said to occur at least partly in virtue of the agent’s possessing “a causal power, fundamentally as a substance, to cause a decision” (Pereboom, 2014a, p. 51, emphasis added). This causal power is typically referred to as an agent-causal power. Early agent-causalists seemingly thought that no deeper analysis of agent causation was available, remarking that agent causation was “primitive” (Bishop, 1983, p. 74), “undefined” (Chisholm, 1976, p. 70), and even “mysterious” (Taylor, 1992, p. 53). More recent defenders of agent causation, however, have been more optimistic about giving such an account.

This optimism seems to have arisen in light of the resurgence of so-called neo-Aristotelian metaphysics, which often provide special priority to the notions of substance and causal power. Within neo-Aristotelian metaphysics, causal powers are taken to be a class of properties whose essences are linked to particular ends (e.g., Mumford, 1998; Mumford and Anjum, 2011; Bird, 2007; Lowe, 2003). That is, it is part of the nature of causal powers that they help to “bring about, or probabilify, certain effects” (Armstrong, 2005, p. 312). An example of a causal power might be “charge.” It is part of the nature of charge that particles that possess it have the capacity to repel other particles (cf. Kuykendall, 2019, p. 340). It is typical, then, for powers theorists to explain all causation in terms of the exercise of causal powers like this.

While powers theorists are divided about what the fundamental causal relata are, most are quite friendly to the idea of substance causation, since substances are taken to be the bearers of causal powers. Given all of this, it is no surprise that agent-causalists have looked to neo-Aristotelian metaphysics for support. Jacobs and O’Connor (2013) and O’Connor (2021) are perhaps the clearest cases where this commitment is explicitly made. On this kind of picture, the agent-causal power will only differ from other causal powers in terms of the kind of effect that it is directed at. Whereas, say, charge is directed at repelling other particles, the agent-causal power is directed at causing actions.²

It is sometimes thought that talk of agent (or substance) causation is merely elliptical for talk of a certain kind of event causation. For instance, (Thomson, 2003) distinguishes between a non-fundamental “cause” and a fundamental “CAUSE.” She then maintains that only events can be CAUSES, but allows that non-events, such as agents, can be causes. She writes that, when x causes y, “[i]f neither x nor y is an event, then for x to cause y is for some event appropriately related to x to CAUSE some event appropriately related to y” (emphasis added). On such a construal, agents can play a kind of causal role, though this causal role is ultimately derivative of the more fundamental causal role of some events.³

Clarke (2017) similarly suggests that “[c]ausation by substances is non-fundamental if it can be reductively analyzed in terms of causation by things of one or more other categories.” More specifically, he suggests:

Reductive analysis: Substance S caused event e₂ just in case there was some event, e₁, such that e₁ involved S and e₁ caused e₂.

The relation of involvement can be interpreted in various ways, such as the “event’s being a change undergone by that substance,” or perhaps even the substance’s simply being the subject of the event.

Furthermore, the “just in case” idea expressed in Reductive Analysis is that the events standing in causal relation are both conceptually and ontologically prior to the substance causing an event, such that it is the event’s standing in causal relation that explains why the substance stands in a causal relation to the event (Clarke, 2017). Put another way, the idea is that substance causation is non-fundamental if substance causation is grounded in—i.e., made true by—event causation. We might also add, then, that event causation is non-fundamental if event causation is grounded in substance causation.

The key takeaway of this discussion is that a pertinent distinction between event- and agent-causalism is the ontological and conceptual priority given to each causal relata (at least in the context of free actions). The event-causalist maintains that events have ontological and explanatory priority with respect to any substance causation—i.e., that events are the fundamental causes of free actions—while the agent-causalist maintains that substance causation has ontological and conceptual priority with respect to any event causation—i.e., that substances are the fundamental causes of free actions.⁴ (In what follows, all talk of causation refers to fundamental causation, unless stated otherwise.)

3 A false dichotomy—and a causal pluralist proposal

The boiler-plate dichotomy just laid out is almost as common in discussions of free will as the distinction between compatibilism and incompatibilism. Nevertheless, it is a false one. According to event-causalism, an action is free only if it is caused by certain mental events. And, according to agent-causalism, an action is free only if it is caused by an agent. Two positions have emerged out of this dialectic: either we go in for an event causation story of free will, or we go in for a substance causation story of free will. But, whichever we go in for, the two are mutually exclusive; we cannot allow for both in our theory of free will.⁵

The options, however, are not mutually exclusive. Free will theorists have unwittingly accepted that, assuming mental events (such as reasons) cannot be causes of free actions (as does the agent-causalist), no events of any kind could be causes of free actions. But, mental events do not seem to exhaust the relevant possibilities—there appears to be at least another option that deserves consideration. To see this, however, some setup is required. So, in the remainder of this section, I will explain why there is another kind of event that deserves consideration as being the cause of free actions. Furthermore, I will use this insight to begin to sketch out how this may open up the possibility of a view on free will that allows for both event causation and substance causation.

So far, we have been relying on a pre-theoretic conception of events—roughly understood as things that “happen” or “occur”—but, it will be helpful to home in on a particular conception of events. To be more specific, I will understand events in the sense explicated by Kim (1973). While Kim’s conception of events is by no means the only conception of events on offer, it seems to be the most widely assumed conception of events, especially in the free will literature. Below we will return to the question of how much of our discussion hinges on this particular conception of events.

With that said, Kim defends a property exemplification model of events. According to this model, an event is the exemplification of a property, P, by substance, x, at a time, t. Schematically, then, all events have the structure of: x’s exemplifying P at t. So, for example, our previously mentioned event of the tree’s falling would be short-hand for something like: the tree’s exemplifying the property of falling at midnight.

We need only three ingredients to have an event on this model: a substance, a property, and a time. What we are looking for is a candidate event that can be the cause of a free action. Candidate substances and times are fairly straightforward. The substance will be the agent who is acting freely, and the time will be whenever the proposed event occurs. All that is left is a candidate property. One such property is the aforementioned agent-causal power. This property seems like a plausible candidate, since agent-causalists maintain that the exercise of such a property is a necessary condition for acting freely. The agent, then, will always exemplify this property when acting freely. Since we are also looking for a candidate event that will always be present in the occurrence of a free action, the agent-causal power seems like a good fit.

With that laid out, we have the following candidate event which might play the causal role of bringing about a free action: an agent’s exercising her agent-causal power at a time.

The proposal is that events of this type deserve consideration as potential causes of free actions. Just as traditional event-causalists maintain that free actions are caused by certain mental events, such as a person’s having certain reasons, the suggestion here is that free actions could be caused by a distinct sort of event—namely, the agent’s exercising her agent-causal power at a time. However, in contrast to standard agent-causalist accounts, the agent is not directly causing the action, but is doing so by means of the event just described.

This insight can be used to generate a further proposal. Call the view that freedom is compatible with both event and substance causation Causal Pluralism. On the Causal Pluralist proposal, free will is wholly explicable in terms of event causation. But, the event which may be the cause of a free action is not the typical mental event, such as the agent’s reasons. Rather, it is an event directly involving the agent—the agent’s exercising her agent-causal power at a time. However, the Causal Pluralist does not maintain that event causation is necessary for free will. For the Causal Pluralist, free will is equally compatible with substance causation. An action may be free if it is directly caused by the agent, in virtue of exercising her agent-causal power. How we might distinguish between these two types of causes of free actions is discussed in the next section:

• A. The agent’s exercising her agent-causal power at a time.

• B. The agent, in virtue of exercising her agent-causal power at a time.

It is important to clarify that Causal Pluralism is not a commitment to something like a dual aspect view, where both substance and event causation are really one and the same, but looked at, or described, in two different ways. Rather, the idea is that our being free does not necessarily turn on whether the causal relata involved in free action turn out to involve substances or only events. The Causal Pluralist may remain neutral about such matters—that is, neutral about whatever the fundamental causal relata turn out to be. She need only maintain that substance and event causation be of the right kind if an action is to be free.

To sum up briefly, then, Causal Pluralism maintains that free will is compatible with both event and substance causation (given suitable characterizations). On the one hand, if the agent directly causes an action in virtue of exercising her agent-causal power at a time, then the action may be free. But, on the other hand, if an action is caused by an event of the sort: an agent’s exercising her agent-causal power at a time, then the action may also be free. This is not to say that a free action must be caused by both a substance and an event, but merely that, if either causal story turned out to obtain, our free will might still be secured.⁶

To be sure, none of this is proof that this distinctive form of Causal Pluralism is true. That it is possible, though, and even appealing, seems a particularly important thing to appreciate, given how worked-over and seemingly intractable debates have been about free will. There may yet be a way forward for convergence amongst event- and agent-causalists.

4 A challenge for agent-causalists

Bringing the underlying machinery of agent causation to the surface and contrasting it with a property-exemplification model of events forces us to think more critically about the differences between substance and event causation. In the previous section, I introduced the Causal Pluralist view and distinguished between two types of potential causes of free actions within the framework. While some have made similar suggestions (see, for example, Nelkin, 2011, pp. 75–79), here, I hope to develop these ideas in more detail and provide further motivation for them. In particular, I do this by examining more closely the nature and metaphysics of events. Most writers—especially agent-causalists—seem to assume an overly simplistic account of events when contrasting event causation with substance causation—which, I think, has made it easier for them to be dismissive of event causation (in the context of free actions). In this section, then, I provide further motivation for thinking that both kinds of causes may be consistent with our being free. I start by providing a way of distinguishing between these two types of causes. After that, I consider a challenge for why agent-causalists in particular should take seriously the possibility that both causes are compatible with free will (as the Causal Pluralist maintains).

If all agent causation involves the exercise of an agent-causal power at a time, one might initially wonder why agent causation is not simply event causation by another name. Indeed, Clarke recognizes this concern and writes:

If a substance causes an event, it does so in virtue of having some causally relevant property$\ldots$The accommodation, however, comes perilously close to acknowledging that it is the substance’s having the property at the time in question that is the cause. (Clarke, 2003, pp. 201–202)

While the concern about whether substance causation collapses into event causation has not garnered much attention, some authors have briefly weighed in on the matter.⁷ For instance, in response to Clarke, Pereboom (2014a, pp. 54–58) contends that it is at least conceivable that an agent cause an effect independently of any causally relevant property. Since we can conceive of this kind of substance causation, then it really must be distinct from event causation. Though helpful, Pereboom’s point does not provide us a with concrete way of demarcating substance causation from event causation, and so it may be worthwhile to consider alternative proposals.

Ann Whittle has also responded to Clarke’s concern, but has gone further than Pereboom in providing a concrete means of demarcation. She writes:

[W]hile the views sound similar, they are not equivalent. To say that an effect occurred ‘in virtue of’ the laws does not thereby make the laws causes of the effect. Similarly, saying that an effect occurred ‘in virtue of’ a substance having a property at a time does not render the having of that property the cause$\ldots$we may insist that the ‘in virtue of’ relation holding between substances and their powers need not be understood in terms of what the substance’s properties cause. Rather, properties are responsible for the fact that substances cause things. (Whittle, 2016, p. 16)

According to Whittle, substance causation differs from event causation with respect to the role that the causally relevant property (in our case the agent-causal power) plays. The idea is that the ‘in virtue of’ relation—in the claim an effect occurred ‘in virtue of’ a substance having a property at a time—can take on at least two meanings, one causal and one non-causal. On the causal reading, a substance’s exemplifying a power is the cause of an effect. On the non-causal reading, a substance is the cause of an effect, with the exemplification of the power playing a non-causal role in explaining how the substance caused the effect.

Whittle stops short of explicating how to understand this non-causal ‘in virtue of’ relation that she invokes. While she suggests that there is an analogy with the laws of nature, it is not entirely clear what the details are supposed to look like. Part of the problem is that there are a number of different ways to understand what laws of nature are—some of which threaten circularity in the present circumstances.⁸

One natural interpretation of Whittle’s non-causal ‘in virtue of’ relation is that of metaphysical dependence, or grounding. It is typically thought that one thing (fact, etc.) F grounds another thing (fact, etc.) G when (inter alia) G’s obtaining (non-causally) depends on F’s obtaining, and F’s obtaining (non-causally) explains G’s obtaining (Clark & Liggins, 2012). To take an example, it is commonly thought that the singleton set $\{$Socrates$\}$ is grounded in the person Socrates, since the existence of the singleton set depends on, and is explained by, the existence of the person Socrates.

Although the grounding interpretation has some plausibility, it is unclear if Whittle would be willing (or would have the need) to take on board all of the commitments that come along with metaphysical grounding. For that reason, I will leave it open what exactly the non-causal ‘in virtue of’ relation amounts to—although, it seems to me that metaphysical grounding still provides a useful approximation for trying to understand what the relation is.

To return to the main thread of our discussion, Whittle’s proposal provides us with a means of demarcating substance causation from event causation. In particular, the distinction looks something like this: event causation occurs when the substance’s exemplifying (or exercising) the causally relevant property at a time stands in causal relation to an effect. And, in contrast, substance causation occurs when the substance stands in causal relation to the effect, but the substance does so in virtue of (in some non-causal sense) the substance’s exemplifying (or exercising) some causally relevant property (at a time).

Whittle’s suggestion provides us with a helpful way of bringing out the potential differences between event and substance causation. The suggestion also seems to be in line with what current agent-causalists maintain about the metaphysics of agent causation (cf. Jacobs & O’Connor, 2013). So, there is good reason to think that the proposal captures a plausible way in which agent-causalists might try to explicate the difference between event and substance causation.

Let us grant that Whittle’s foregoing distinction provides a way of demarcating event causation from substance causation. While we might dispute which story provides us with the best metaphysics of causation, I believe a more important issue deserves attention.

The issue is this: granting the distinction between substance causation and event causation, is the difference between the two a freedom-relevant difference? Suppose Angelica makes an utterance. Could her acting freely in this case turn on which of in virtue of exercising?

A$^{*}$.:

Angelica’s exercising her agent-causal power at a time.

B$^{*}$.:

Angelica, in virtue of (in some non-causal sense) her exercising her agent-causal power at a time.

It is not entirely clear that it would, as it is not clear that there is any freedom-relevant difference between the two causes.

In light of this, if the agent-causalist is correct that substance causation is compatible with our being free, then we need some principled reason for maintaining that only substance causation is compatible with our being free. That is, we need some principled reason for ruling out event causation as being compatible with free will. But, it is not clear there is any such reason for thinking this—that there is some freedom-relevant difference. Because of this, we should take seriously the proposal that both substance and event causation are compatible with our being free.

Freedom-relevant differences come in many different shapes and sizes. Here are some examples that have been proposed throughout the years: the causal history of an action, the presence of indeterminism, the ability to do otherwise, and the reduction of agency to mental events. In the present circumstances, the onus is on the agent-causalist to show that there is more than a mere difference between substance and event causation. They must show that the difference between the two captures some freedom-relevant feature. And, as the list just provided indicates, there are several ways to do that. The important point here, though, is simply that arguing about potential freedom-relevant differences is a mainstay of the free will literature, and so it should not be seen as unusual to ask for one in this context.

It may be worth reminding the reader at this point that the foregoing challenge is meant to apply to both compatibilist and incompatibilist (i.e., libertarian) agent-causalists. Thus, if one is to propose a freedom-relevant difference between substance and event causation, it is important to be clear who the interlocutor is. For instance, suppose one thinks that the problem of luck shows that indeterminism (of the sort postulated by libertarians) undermines free will.⁹ If the Luck Objection is presented to a libertarian agent-causalist, this might show that they ought to give up being a libertarian, but not necessarily that they ought to give up being a compatibilist agent-causalist. Perhaps there are freedom-relevant differences that apply equally well to both compatibilists and incompatibilists. That is all well and good. The important point here is that we need to be careful who is being addressed when considering potential freedom-relevant differences.¹⁰

While I have stressed the significance of freedom-relevant differences, one might also wonder what role non-freedom-relevant differences should play in this discussion. As noted earlier, there are a variety of options when picking one’s preferred causal relata—even within a neo-Aristotelain framework. Buckareff (2011, 2017) and Mumford and Anjum (2011) maintain that powers are causes. Kuykendall (2019, 2021), Whittle (2016), and Skow (2018) all maintain that substances are causes. And most of analytic philosophy over roughly the past century has worked under the assumption that events are causes. Each of the foregoing camps will have considerations and arguments in favor of their choice of causal relata. So, what should we say about these divides?

I think that these considerations will be important in picking one’s preferred causal relata, but I also believe that they don’t make a difference to free will. Potential considerations are likely to include things like: being the most metaphysically plausible, having the best fit with our best scientific theories, cohering with one’s preferred theory of causation (e.g., a counterfactual account), and so on. While I take these all to be legitimate considerations, it is important to keep in mind that they don’t show that free will specifically requires one or another causal relata. For example, suppose one prefers a counterfactual analysis of causation, and suppose further that one believes that this account of causation requires that events be causes. While this would give one reason to believe that all causes are events, it would not show that free will requires that events be causes. I take the same point to apply to all considerations that don’t qualify as freedom-relevant differences.

I believe a two-fold challenge for agent-causalists can be drawn from our discussion. The first challenge is to show that substance causation does not collapse into event causation. The second challenge is to show that the reason why substance causation does not collapse into event causation is due to some freedom-relevant difference. While I have suggested one plausible way of addressing the first challenge, I have also explained why I think this is not enough to address the second challenge. Since Causal Pluralism avoids the foregoing concerns, the agent-causalist has special reason for taking the proposal seriously.

Before moving on to the next section, it is worth pausing to highlight an important advantage of the Causal Pluralist view. In particular, I would like to suggest that Causal Pluralism provides a way of securing central aspects of agent-causalism, without the same ontological stringency.

Causal Pluralism is able to capture much of what agent-causalism offers in part by accommodating much of its metaphysics. Causal Pluralism takes as its starting point the agent-causal story, with irreducible agents and their agent-causal powers, and builds it into its own proposal. However, motivated by the fact that there is not a clear freedom-relevant difference between substance and event causation (given suitable characterizations), the Causal Pluralist also allows for event causation. But, it is important to notice that even the event allowed for in Causal Pluralism is not far off, ontologically speaking, from the substance-cause postulated by agent-causalism. In both cases, we have irreducible agents exercising their agent-causal powers. By allowing for such features no matter what the causal relata are, Causal Pluralism captures a core and guiding tenet of agent-causalism. Furthermore, it does this without restricting itself to fundamental substance causation.¹¹

5 Events reconsidered

One might naturally wonder whether the plausibility of Causal Pluralism, especially the challenge raised in the preceding section, turns on a specific conception of events—namely, the property-exemplification model. And, while I think it would be an interesting and important point if free will turned out to be compatible with both substance causation and only one form of event causation, I think there are ways of extending our remarks to other accounts of events. So, the goal of this section is to motivate the point that Causal Pluralism does not stand or fall with the property-exemplification model of events.

With that said, a few brief remarks are in order before moving on to the proceeding discussion. The first is that I will assume that substances (such as agents) and causal powers (such as an agent-causal power) are taken to be fundamental constituents in our ontology. I will then briefly canvas three different accounts of events and motivate the claim that agents and their agent-causal powers may be so intimately involved with these various construals of events that it is not clear that, if any of them should turn out be correct, our being free would be undermined. My suspicion is that intuitions may vary from account to account for various readers, so that Causal Pluralism may seem to be more or less plausible depending on the account of events under consideration. It is important to state upfront, then, that it is not my aim to show that Causal Pluralism is equally plausible on every construal of events, but simply that there exists at least some flexibility when it comes to what theory of events a Causal Pluralist might endorse.

5.1 States of affairs

Start first with a conception of events that diverges the least from Kim’s. Chisholm (1990, 1992, 1994) defends an account where events are construed as states of affairs.¹² Chisholm construes states of affairs as the exemplification of properties by substances. Chisholm, then, departs from Kim only insofar as he thinks the times are not constituents of events. To provide an example for concreteness, a fully specified Kimian event would be something like: Angelica’s waving at 12am, whereas a full Chisholmian specification of the same event would simply be: Angelica’s waving (which has the property of occurring at 12am).

If a Kimian event is compatible with Causal Pluralism, then so is a Chisholmian event. The Chisholmian event-cause of our free actions would simply be the agent’s exercising her agent-causal power (which occurs at a particular time). The only modification that would be needed to make room for this kind of event would be to drop the requirement that times are constituents of the events, and there’s no obvious reason to think that this move would carry any complications that would threaten our acting freely.

5.2 Spatiotemporal regions

Now consider an account of events that departs further from Kim’s. This account construes events as spatiotemporal regions (Davidson, 1985; Quine, 1985). We can get an intuitive grasp on spatiotemporal regions by first thinking of the universe in four-dimensionalist (or eternalist) terms, where the totality of the universe—past, present, and future—exist “all at once” in a four-dimensional block. Events can then be picked out by taking “slices” or subregions of the universe. For instance, an event such as Angelica’s waving at 12am will occupy some subregion within the block, and that subregion will be the event.

Note that, while it is easier to get a grasp on the idea of spatiotemporal regions within four-dimensionalism, the Davidsonian need not be committed to such a view. To give just one example, some presentists—i.e., those who say that only the present exists—allow that the present consists of a (much) smaller, and constantly changing, four-dimensional block. This would allow a presentist to account for the existence of spatiotemporal regions in much the same way as the eternalist.¹³

At face value, such an account might seem highly antithetical to the Causal Pluralist project, given how far the account departs from our initial conception of events. However, things are less problematic than they initially seem. As long as we are content with saying that agents are physical substances that are located in time and space, we can say that the relevant events (i.e., spatiotemporal regions) involving free actions are simply delimited by the location of the agent (cf. Bennett, 1988). When an agent acts freely, the agent exercises her agent-causal power at a time and place. The event of the agent exercising her agent-causal power would then be the exact—no more, no less—spatiotemporal region where the agent exercises her power. Conceiving of events as delimited by the substance involved would seemingly “blur” the distinction between substance and event causation (cf. Simons, 2003).

5.3 Properties of spatiotemporal regions

Consider one last account of events—one that departs even further from Kim’s account than the one just considered. On this alternative conception, events are a special class of properties—they are properties of spatiotemporal regions (Lewis, 1986; cf. Bennett, 1988). This account builds on the previous one by incorporating spatiotemporal regions into its analysis of events, but does not go so far as to reduce events to those spatiotemporal regions.

Conceiving of an event as a property is prima facie puzzling. What kind of property is this? Lewis himself admits to only giving constraints for when a property is “formally eligible” for being an event, though the constraints provided are not particularly helpful in this context.

To get a better fix on how we might understand Lewis’ claim that events are a certain class of properties, it may be helpful to consider some of what Richard Montague has said on the matter. Montague has defended a nearby version of Lewis’ account—one according to which events are properties of times, rather than properties of spatiotemporal regions. My interest is not in Montague’s own account, but in the following comments he makes:

A third possibility, and one that seems to be the only reasonable suggestion, is to take as the event corresponding to a formula the property expressed by that formula. Thus the event of the sun’s rising will be the property of being a moment at which the sun rises, and events in general will form a certain class of properties of moments of time. (Montague ,1969, p. 160, emphasis added)

The general strategy proposed here appears to be to, first, take what seems (pre-theoretically) to be an event—such as the sun’s rising—and then, second, make a property out of it—such as the property of being a moment at which the sun rises.

Since Lewis admits to making a similar kind of proposal, it seems we could apply Montague’s remarks to Lewis’ own account. This would mean that events correspond to properties such as the property of the sun’s rising; but, instead of attributing such properties to times, we would simply attribute them to spatiotemporal regions. While this does not give us a principled way of individuating events, it at least should give us an intuitive grasp on the proposal that Lewis is making.

It may be worth addressing that Lewis takes properties to be sets, and so ultimately reduces events to sets of regions spread across different concrete possible worlds. We may not want to adopt all of Lewis’ machinery, so it is worthwhile to flag that options are available. For instance, Bennett (1988) proposes an account of events that takes them to be tropes that are had by spatiotemporal regions, where tropes are said to be “abstract particulars” or “property instances.” Since Bennett takes events to be properties of sorts (which are had by spatiotemporal regions), he can be understood as proposing a variation of Lewis’ account. I see no reason why, then, one could not also adopt a different account of properties on this matter.

We have said that, according to this account, events are properties of spatiotemporal regions. For example, Angelica’s waving is an event and a property. It is an event because it is a specific sort of property. And this property is instantiated at a region—namely, the region where Angelica waves. So, for Lewis, there is an event at that region because this property is instantiated at that region.

The same can be applied to cases where an agent acts freely. If an agent acts freely, she thereby exercises her agent-causal power to freely cause some action. The exercising of her agent-causal power may be described as a property: the property of exercising her agent-causal power. We may thereby also characterize that property as an event. The agent’s exercising her agent-causal power is a property that is instantiated at the very region where the agent exercises her agent-causal power, and it is this kind of event that may be thought to cause our free actions.

Before concluding, it is worth taking a moment to make a few comments. First, the foregoing accounts by no means exhaust all of the available accounts of events and the possible permutations.¹⁴ Such a task would take us far beyond the purposes of our present one. (I leave the task to the reader to plug in their preferred theory of events and check the results.) Our present task has been to show that Causal Pluralism does not stand or fall with a property exemplification model of events. There exist other accounts of events one may endorse while also endorsing Causal Pluralism. To lay my cards on the table, Chisholmian events and events conceived of as spatiotemporal regions seem just as consistent with Causal Pluralism as Kimian events. I am less confident about events conceived of as properties of spatiotemporal regions, though this may be due to my uneasiness about talk of properties as causes more generally. Of course, intuitions may vary.

It seems to me that the spatiotemporal regions theory is more in line with how philosophers of science tend to think of events, so it may be of special interest how well it meshes with Causal Pluralism. And, on this score, I think Causal Pluralism does well enough. So long as we grant that substances and powers are fundamental constituents in our ontology, we can construct events using these materials in much the same way that a Kimian constructs events—that is, out of substances, properties, and times.

All in all, I hope to have motivated the point that the plausibility of Causal Pluralism does not stand or fall with the Kimian conception of events.

6 Concluding remarks

Causal Pluralism gives us a new perspective on a number of important issues pertaining to free will. While I do not have the space to fully address all such issues here, I would like to stop and take note of one issue in particular: agency reductionism. Space constraints preclude a fuller exploration of agency reductionism; but, before concluding, I would like to (i) sketch out how I think Causal Pluralism could address the matter, and (ii) draw out an interesting upshot of having the Causal Pluralist view on the table.

Agency reductionism concerns whether “the causal role of the agent in all agential activities (from nonintentional to autonomous action) is reducible to the nondeviant causal activity of appropriate mental states and events involving the agent” (Franklin, 2018, p. 15). It is standard, then, to characterize traditional event-causalist views as adhering to agency reductionism, since they maintain that all actions (including free actions) are caused by mental events. And, it is also standard to characterize traditional agent-causalist views as adhering to agency nonreductionism, since they maintain that at least some actions are caused by substances.

A natural question is whether Causal Pluralism is best characterized as adhering to agency reductionism or nonreductionism. Under the Causal Pluralist view, the question divides itself: (a) if the (fundamental) causes of free actions turn out to be substances, is Causal Pluralism committed to agency reductionism? And (b), if the (fundamental) causes of free actions turn out to be events (of the sort specified earlier), is Causal Pluralism committed to agency reductionism?

I think that, if we start by considering (a), Causal Pluralism will straightforwardly classify as agency nonreductionist. This is because the view—under the assumption that the fundamental causes of free actions are substances—simply adopts the same metaphysical story as traditional agent-causalist views. The more interesting question, it seems to me, is what we should say about (b).

I believe that Causal Pluralism is also best classified as adhering to agency nonreductionism even if the (fundamental) causal relata of free actions turn out to be events. The reason for thinking this is that the kind of events allowed for by Causal Pluralism are crucially and relevantly different from the kind of events postulated by traditional event-causalist views. The guiding idea behind agency reductionism, I take it, is that the agent (or agential activity) can, in some sense, be wholly decomposed into the workings of the subcomponents or subsystems of the agent (see, for example, Velleman, 1992; cf. Kane, 1996, pp. 193–194). This is what makes traditional event-causalist views reductionistic about agency. Mental events postulated by traditional event-causalism, such as beliefs and desires, are events that occur within, or inside, the agent. Traditional event-causalism, then, reduces the agent (or agential activity) to the workings of the mental subsystems of the agent.

The kinds of events that Causal Pluralism would allow for are not events that take place within the agent. Rather, the events postulated are pitched at the level of the agent (qua substance). This means that there is no attempt to reduce the agent (or agential activity) to the workings of the subcomponents of the agent. What this indicates is that Causal Pluralism is not committed to agency reductionism, whether the first causal relata are agents or events (of the sort previously specified).¹⁵

Though not stated explicitly, authors sometimes seem to talk as if agent causation is the only way to avoid agency reductionism (cf. Velleman, 1992; Franklin, 2018). But, an important upshot of the preceding discussion is that the two seem to come apart: one can be an agency nonreductionist while still allowing for event causation. This is because it is not the adoption of event causation per se that makes an account agency reductionist, but rather, the adoption of certain kinds of events.

Given the foregoing, Causal Pluralism looks to have at its disposal the resources to develop a promising line of reply to formidable objections to standard event-causalist accounts, such as Pereboom (2014a, 2014b) Disappearing Agent Objection and Franklin’s (2016, 2018) It Ain’t Me Argument. Arguments like those developed by Pereboom and Franklin aim to undercut traditional event-causalist accounts by showing that the mental events that are purportedly the cause of free actions fail to supply what is needed for free will. In the case of the Disappearing Agent Objection, it is contended these mental events cannot account for an agent’s settling a free action; and, in the case of the It Ain’t Me Argument, it is contended that they cannot account for an agent’s self-determining a free action. Both authors maintain that agent causation presents the most promising solution to their respective arguments.¹⁶

We do not need to get into all of the details of the arguments here. What is important for our purposes is concentrating on what lies at the heart of these objections. It seems to me that these arguments get their traction by leveraging the distinction between an agent (conceived of as a substance) and the agent’s mental events. In doing so, it is easy to get an intuitive grip on why one might think that the agent’s mental events cannot account for their settling a free action, or for their self-determining a free action. These arguments, it seems to me, provide lucid ways of articulating the concern raised by Taylor (1992, p. 51): “If I believe that something 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 believe that I was the cause of that external or internal event.”

Now, with all that laid out, it’s not clear to me that these arguments are going to have the same kind of pull when applied to Causal Pluralism (qua its event-causal specification). The kind of event allowed for by Causal Pluralism is pitched at the level of the agent. There is no effort to reduce the agent’s causal contribution in free action to the causal contribution of the agent’s mental events (or mental subcomponents). The concerns that the agent disappears from the picture, or that it really “ain’t” the agent determining the action, consequently don’t have the same kind of force when targeted at Causal Pluralism.

Furthermore, recall that traditional agent causation is taken as a solution to both authors’ arguments. Since Causal Pluralism adopts much of the same underlying metaphysics in its event-causal specification, it seems to me that both arguments are going to have a difficult time consistently maintaining that traditional agent causation is a solution to their respective arguments, but that Causal Pluralism in its event-causal specification is not.¹⁷

Let us now step back and take stock of some of the central ideas that have been put forth in this paper. First and foremost, I have tried to introduce a novel view on the issue of free will—namely, Causal Pluralism. Causal Pluralism maintains that free will is compatible with both fundamental event and substance causation. Thus, the view allows one to stay neutral about what the fundamental causal relata are, so long as the relata are of the right sort.

Second, while I think Causal Pluralism deserves wider consideration, I have also tried to show that the demand to consider Causal Pluralism is all the more pressing for extant agent-causalists. Insofar as the move to adopt Causal Pluralism is more natural for the agent-causalist, agent-causalists ought to reconsider the fact that they have unnecessarily wedded themselves to substance causation, thereby making their account more ontologically demanding than it needs to be. Causal Pluralism may then be a way of securing the goods of agent-causalism without all the ontological demands.

Third, I hope to have shown that, independent of whether one is ultimately attracted to the view, having Causal Pluralism on the table helps us to see a number of issues in a new light, such as: (a) there are overlooked ways to address the dichotomy when choosing between event and substance causation in a theory of free will, and (b) event causation may come apart from agency reductionism.

No doubt further issues remain that deserve consideration. For instance, a fuller and more systematic treatment of how well Causal Pluralism fares against its competitors is in order, as is a closer look at the nature of agency reductionism. Furthermore, although the focus has been on the event-substance divide in the context of free will, there remains the possibility of extending the considerations made here to the nature of action more broadly. This seems to me an interesting possibility which I am partial to. However, it is one that I’m not fully ready to embrace without a closer examination of the nature of action. Regardless, such treatments will have to come at a later time. For now, agent-causalists who wish to resist Causal Pluralism would do well to reconsider the relation between substance and event causation, as well as what substance causation secures that event causation cannot.