Sublating the free will problematic: powers, agency and causal determination

Abstract

I argue that realism about causal powers (which, loosely following Brian Ellis, I refer to as ‘anti-passivism’) sublates the passivist, Humean-inflected free will problematic. In the first part of the paper I show that adopting what I call ‘powers-non-determinism’ reconfigures the conceptual terrain with respect to the causation component of the contemporary problematic. In part two I show how adopting ‘powers-non-determinism’ significantly alters the nature of the discussion with respect to the agency component of the problematic. In part three I compare ‘powers-non-determinism’ to an otherwise- Humean agent causal position.

My aim in the analysis to come is to show that, and how, the contemporary debate in Anglo-analytic philosophy over free will is radically reconfigured if the passivist account of causation that is built into it is replaced with a powers-based account. More precisely, it is sublated by such a substitution. Sublation is a Hegelian term. To sublate is to re-frame, such that apparently irreconcilable, ostensibly independent claims may be shown to be misconceived partial elements of a framework that is itself coherent. The structure of my argument is given by the structure of the free will problematic itself, which, as a matter of conceptual architecture, is composed of two categories, causation and agency, which are presumed to be dichotomous. In the first part of the paper I show how assuming the existence of real causal powers alters the causation component of the ostensible dichotomy. In the second part I demonstrate how doing so relates to the conceptualization of free agency. Both elements of the presumptive dyad are altered when anti-passivism is assumed, as is the problematic as a whole. One consequence of the fundamental change in the terms of debate is a bolstering of the position known as agent-causation. I therefore end the paper by articulating the merits of a comprehensively anti-passivist version of agent-causation as compared to those of an agent-causal view married to an otherwise-passivist metaphysics.

1 Powers and causation

Let me say just a word about what I am calling a powers-based approach. While it is not my intent in this paper to defend a developed first-order account of the nature of powers (mine or anyone else’s), I do want to be clear about what the general parameters are for the type of position that, for present purposes, will qualify as a powers-based approach. This step is important because rejoinders from passivists often depend upon assuming that a power is something that is precisely the opposite of what an anti-passivist thinker, regardless of the details of their view, takes a power to be. The most efficient way to get at the difference between anti-passivism and passivism by thinking of a children’s animated flipbook, the sort that one makes by drawing a number of pictures, each slightly different, on separate pieces of paper. When one fans through the pages quickly, it looks as though the drawings are doing things. The example of a flipbook allows us to distinguish between two competing views. The first view is that real activity, real doing, is not at all like flipbook-style doing. Specifically (says the proponent of this view), real activity is not composed of static “stills.” I call this position anti-passivism, taking a cue from Brian Ellis. The second view—call it passivism—is that real doing and flipbook-style doing are in fact exactly alike (in the relevant respect). Real-world activity seems to be something other than a succession of static “stills,” but in actuality it is no different, metaphysically, from flipbook activity. These two positions are contraries. Anti-passivists believe that the world contains a phenomenon the existence of which is denied by passivists, namely, just that phenomenon in virtue of which (says the anti-passivist) real activity and flipbook-style-activity are not just different, but opposites (in the relevant respect): one is active, the other is not-active. Traditionally, all parties referred to the posited phenomenon (the existence of which some affirmed and others denied) by the term ‘power,’ and so will I. For a pre-theoretical sense of what the term in its traditional sense is meant to pick out, it may be helpful to think of the common distinction made between verbs and nouns. Powers, we might say—Thomas Reid did say, and G. E. M. Anscombe too—are why we need verbs, active ones in particular. In a more theoretical register, I shall use the term power to refer to a potential, had by that which is powerful, to engage in activity or doing of a given kind.¹ This definition is meant to be general enough that it could be accepted by contemporary anti-passivist thinkers ranging from Roy Bhaskar, Rom Harre, E. H. Madden and Nancy Cartwright to Brian Ellis, Stephen Mumford and Rani Lill Anjum, although this is not an exhaustive list. It is meant to be restrictive enough, however, to be explicitly at odds with the position defended by Alexander Bird, for example, wherein powers are simply properties with “If ...then” identities, whose natures are essential to them, as well as with any other overtly or covertly passivist usage of the term power (Cartwright 1983; Chakravartty 2008; Ellis 2001; Ellis and Lierse 1994; Harré and Madden 1975; Mumford 2004; Mumford and Anjum 2011).²

The contemporary free will problematic does not hinge upon belief in the designs of God(s) or fate, as older iterations of it did. It presupposes only that causation amounts to a rubric of nomological necessitation, such that determinism obtains. This seemingly benign assumption gives rise to a disquieting bit of logic: if it is true that all events follow deterministically from existing conditions plus invariant laws of nature, then it cannot be the case that some events (i.e., those involving agents) do not. The situation is summed up nicely by the title of David Lewis’ paper “Are We Free to Break the Laws?” Compatibilists say no, but stipulate that while the laws are inviolable, we may (a) define the concept of ‘free will’ in such a way that what it refers to is something other than having desires, intentions and behaviors that are set by us, rather than by laws of nature; and/or (b) relax the meaning(s) of ‘necessitate,’ ‘law’ and/or ‘cause,’ so that an environment that is characterized by contingent regularity will count as deterministic, as David Hume did (or more recently, Kai Neilsen³). Hard determinists simply accept with equanimity the implication that we do not have free will. Libertarians, finally, contend either (a) that free actions are spontaneous, and therefore uncaused; or (b) that free actions are indeterministically caused, consistent with governance by probabilistic rather than deterministic laws; or (c) that free actions involve an entirely different kind of causation than that which applies to all other events.⁴ Event-causal libertarians opt for (a) or (b); agent-causalists for (c). All passivists face the added challenge, with respect to the theorization of agency, of holding the view that nothing actually has any productive power to do anything, but that is not the difficulty around which the debate is standardly framed.

The question that I want to pose first is this: what happens to the free will problematic if one rejects a nomological account of causation, and adopts a powers-based approach instead?⁵ Causation, says the anti-passivist, is not a matter of lawful regulation at all. Causation is about the display of the powers of powerful things – activity that may or may not occur, and which, if it does, may or may not issue in any given outcome. Laws, if and where they hold, and assuming that one countenances law-talk at all, are merely descriptions of such displays, i.e., descriptions either of the activity of powerful things or of the outcomes of that activity, depending upon the specifics of a given powers-based account. Agency, therefore, has nothing to do with breaking them or otherwise getting out from under their rule. Indeed, ‘metaphysical freedom,’ from this perspective, is just a name given to the remarkable second-order causal powers had by human beings (and, one might think, by other sentient substances), powers that allow us to, among other things, direct, display or refrain from displaying various of our first-order powers. It should be immediately clear, even at this level of generality, that changing the underlying metaphysics matters significantly. For instance, what I have just said is that the covariance between causation and free will is positive, not negative. The greater the degree of causal determination (by an agent), the greater is the degree of freedom.⁶

But we can be more specific than this. First, consider the types of powers that one might take to exist, and the sorts of law-statement that each type would underwrite, should one believe in the existence of the requisite type of power, and be concerned to account for law-statements. These are: (a) powers the display of which would underwrite non-probabilistic law-statements (or would do so assuming non-interference), if one accepts such statements; (b) powers the display of which would underwrite probabilistic law-statements; and (c) powers the display of which do not underwrite any meaningful law-statements at all. Second, note that there are three possible views that one could have about which of these types of powers do exist. One might hold that the only powers that exist are those the display of which is perfectly regular (if only under artificially induced, experimental conditions)—a situation that would certainly give rise to the appearance of nomological necessitation. I will call this position p-determinism: “p” for powers. At the other end of the continuum, one might contend that there are no powers the displays of which are regular, not even in a controlled environment. We can call this p-indeterminism. Most natural, I think, would be to claim that powerful things vary in the regularity of their behavior. Call this view p-non-determinism. It will be open to the p-non-determinist to say that the powers borne by some kinds of things give rise to regular conjunctions of events (or do so absent interference), thereby underwriting the deterministic law-statements affirmed by those who believe in deterministic laws, while the powers borne by other kinds of things, do not. Displays of the latter types of power can at best be described probabilistically—or, in the case of the behavior of human agents, arguably cannot be well described in terms of laws at all. All three types of display will be thought to be aspects of the same underlying reality, one that includes powerful particulars of different kinds, having qualitatively different kinds of causal powers.

Now, it is not my intention to make a case for one or another version of anti-passivism, or to settle any disputes related to the ontology of powers. My claim, the claim that I do mean to defend, is meta-theoretical: viz., that belief in the existence of powerful particulars sublates the free will debate. The point of identifying the stance that I am calling p-determinism was only to show that the anti-passivist has something to say about the appearance of nomological necessitation. P-non-determinism, in turn, is simply the view that not all behavior is regular, though some may be. Indeed, it is a strength of my argument, if it is correct, that it does not hang on the details of one’s anti-passivism. This said, I will add that in my view it is powerful things, and not their properties, that can or cannot do this or that—and that behave, accordingly, in ways that can be described, if they can be, by non-probabilistic rather than probabilistic laws (if one counts any laws as non-probabilistic) or that do not admit of adequate description in terms of laws at all. It would be a bad reification, from my perspective, to think that it is powers themselves that engage in activities of different kinds.⁷

In any case, it is the third position, p-non-determinism, that is of interest. At the most basic level, it does away with the dichotomy between causation and agency, since agency, from this perspective, is merely a species of substance causation, i.e., of an entity being a cause. To show that, properly understood, an ostensibly dichotomous dyad was never actually dichotomous in the first place is to sublate the (ostensible) dichotomy, rather than to resolve it on its own terms. But a sublation also preserves whatever truth there may have been in the original apparently incompatible alternatives, as they were originally construed. Here, p-non-determinism (in contrast to p-indeterminism) allows its proponents to explain the appearance of uniform regularity that nomological determinism describes. P-non-determinism also preserves the idea—supported by event-causal libertarians and compatibilists as well as by otherwise-passivist agent causalists—that agents have free will. Contra event-causal libertarians and compatibilists alike, the p-non-determinist is entitled to presume that metaphysically free agents are causally self-determining. Contra otherwise-passivist agent causalists, she holds that agency is causal in the same sense in which all causal processes are causal. The upshot is that the p-non-determinist does not have to reconcile agential self-determination with either nomological determinism or with gaps in, or the suspension of, causation. Rather, she has only to affirm the unexceptional fact that human beings are neither omnipotent nor the only causal bearers on the scene. Our doings as agents occur in a field of activity involving other powerful particulars, ranging from electrons to bacteria to the international banking system. Some things have powers that can thwart ours; others do not. Some—sugar, water, proteins, the sun—have powers the display of which we need in order to survive. There is nothing surprising about any of this.

And we can be even more specific yet about the sublation. If we look now at the various positions available to passivists with respect to free will, we will in each case be able to re-describe the stance from a powers perspective. Hard determinism will look to be an unwarranted generalization from the regularities associated with behavior of only one class of powerful particular, viz., those entities that behave in ways that are well described by non-probabilistic law statements (if one admits of such statements, as hard determinists tend to do). Passivist compatibilists, meanwhile, inasmuch as they are precluded by their rejection of substance causation and powers alike from theorizing agential activity robustly construed, may be viewed from a powers perspective as raising interesting ancillary issues, but the issues necessarily will be beside the point, for the anti-passivist, with respect to the metaphysics of agency. The anti-passivist will want to know about the causal powers of human substances; the passivist compatibilist has nothing to say on the topic since s/he denies that substances have causal powers. Event-causal libertarianism, in turn, is, from a powers perspective, a vain if valiant attempt to establish a genuine agential efficacy that, by definition, is required to be a-causal (since if it were causal, it would, as per event-causal libertarianism, be deterministic and thus at odds with free action). Finally, otherwise-passivist agent causalists are shown to be needlessly involved in special pleading for agents—and, moreover (as I shall argue below), to be undermined in their very ability to do so by their more general commitments.

In the past I have called the stance that I am recommending (when it is combined with a view of agents as emergent, non-dualistic entities) ‘Aristotelian compatibilism,’ a label meant to capture the idea that agency is indeed compatible with causation. But it could equally well be described as a re-conceptualized agent-causal libertarianism, one that does not require the invocation of a special, ad hoc kind of causality pertaining only to persons. Whichever name one prefers, what matters is that it is possible, if one is a p-non-determinist to (a) make good hermeneutic sense of the available positions within the conceptual space of the standard problematic; whilst (b) recognizing the problematic itself to rest upon false metaphysical premises. A passivist might object that even if it were true that agency is a matter of the display of the real causal powers of sentient substances, there would still a problem because those powers would themselves be deterministically triggered by the antecedent display of the powers of other things. One obvious response to this thought is that even if a power is expressed, there is no guarantee of a given effect occurring (the effect here being the triggering of ostensibly sui generis agential powers), since the activity of some other powerful particular could easily interfere with the outcome.⁸ Furthermore, it is in the nature of the case that even if the display of a second-order agential power were to have been deterministically triggered by the prior expressed power of something external to the agent, the triggered power is precisely a second-order power of choosing. It is the agent who, having just that second-order power, decides what, if anything, she will do, not the powerful thing that (by stipulation) had the power to spark the display of her agential powers. Thus the person who says that powers cannot help with free will because what the supposedly free agent will do is deterministically caused by other powers is, at best, like our p-determinist. To be a p-determinist, however, is not to raise a problem faced by the p-non-determinist. It is just to deny the reality of agential powers. The question is whether or not agents do have such powers—not whether or not our behavior would still be deterministically governed by the expression of other powers if we did.

I want to stress yet again that it is powers as conceived by anti-passivists—the kind of powers that until very recently have been rejected by Humeans as animistic or occult—that make a difference for the free will problematic. Alexander Bird, for instance, believes in powers (“potencies,” he calls them) that are a sub-class of passivist dispositional property, except for that unlike their Humean-style analogue they have their own identities (or natures).⁹ Bird is right to say, as he does, that an account such as his does not help to secure the possibility of free agency.¹⁰ And this will be true of any use of the term ‘power’ that refers us, in the end, only to the fact of order, be the order a fixed arrangement of properties in this world, or fixed sequences of event in all possible worlds. The claim is that rejecting passivism changes things, not that using the word ‘power’ does. Nor, for that matter, will it suffice to allow real powers, but only the type affirmed by the p-determinist, the type had by things such as water, sugar and automatic nervous systems. Neither, finally, will it do to believe in the existence of electrons, for example, and by extension in their non-regular powers, but not to believe in the existence of agents, or selves (as opposed to bodies), and therefore not to believe in agential powers.

Let me make one additional point before moving on to consider the agency component of the free will problematic: rejecting passivism also does away with the putative contradiction between a belief in free will and modern science. Here the apparent conflict derives from the passivist assumption that since causation amounts to the fact of order, the aim of science is to identify lawful regularity. Call this a nomological conception of science. If one subscribes to such a view, then any account of agency according to which agents, rather than laws, are the determining source of their own actions will appear to run afoul of the norms of scientific explanation – as will any event-causal libertarian account according to which metaphysically free acts are not caused at all. This line of thinking gives an epistemic face to the metaphysical problem of luck, or control, in that the putative acts in question are construed as not just metaphysically random, but in principle inexplicable. It follows that belief in the existence of such phenomena is patently anti-scientific. Anti-passivists reject the nomological conception of science outright. It is the powers of things that ground law-statements (and laws, if laws exist), says the anti-passivist; it is therefore the discovery of powers that is the fundamental aim of scientific inquiry. Anti-passivist philosophers of science maintain that such an account of natural science is descriptively superior to nomological alternatives. Nancy Cartwright and Brian Ellis, for example, have long held such a view, and Roy Bhaskar preceded them. With respect to the social sciences, where the practical commitment to passivism is perhaps stronger than in the natural sciences, anti-passivists sometimes add that the relative paucity of social science may be explained by its practitioners’ efforts to isolate constant conjunctions rather than to identify causal powers.¹¹ In any case, the point is that the p-non-determinist will not recognize the supposed dichotomy between what Wilfred Sellars famously called our manifest image (or self-image) and scientific theory. Agency is a species of causal power, the p-non-determinist will say, science the study of such powers generally.

2 Powers and free will

Discussions of free will in Anglo-analytic philosophy have come to involve one or more of the following themes or terms: (1) being the cause of; (2) intelligibility, control or luck; (3) moral responsibility; (4) the possibility of having done otherwise; (5) acting for a reason; (6) psychic harmony, unity and/or authenticity; (7) being “fully” or “genuinely” free. Assuming that p-non-determinism changes what one will say with respect to each of these points. Above all, the p-non-determinist (unlike the passivist) is not required, by her metaphysics, to talk about something other than a power had by agents (we might call it a power of conscious or intentional action, though other words might also do) when she talks about the nature of agency. The central preoccupations of passivist efforts to theorize free will are therefore either redundant or largely irrelevant to the precise task at hand, from an anti-passivist perspective.

• (1) Being the cause of

From an anti-passivist perspective, causal processes involve the display of a power or powers. The passivist, by contrast, does not conceptualize causation in generative terms at all. It is important to be clear about this fact. In “Compatibilism Without Frankfurt: Dispositional Analyses of Free Will,” for example, Bernard Berofsky ostensibly addresses the possibility that realism about causal powers could bolster what he calls “conditionalist compatibilism.”¹² In so doing he explicitly references Stephen Mumford and Brian Ellis. Berofsky concludes that it cannot. What is important, however, is not the conclusion—nor Berofsky’s offhand assertion that “if free will is a disposition, it is evidently not basic and would, therefore, more plausibly rest on some sort of causal basis”¹³—but rather the fact that, having contrasted realism about causal powers with conditional analyses, he then blithely glosses over the fundamental difference between them as accounts of causation. Indeed, the very notion of “some sort of causal basis” that he takes Mumford and Ellis-style agential powers to require, he presumes the anti-passivist may simply lift from Lewis.¹⁴

With respect to the notion of “being the cause of,” the implications of affirming a bona fide anti-passivism are appreciable. First, efficacy now attaches directly to the causal bearer. The agent really is “the cause of.” She is “the cause of” in virtue of being the bearer of the causal powers that are involved in the production of the outcome. Thus causation and agential control do not come apart, for the anti-passivist, as they do for the standard event-causal libertarian or compatibilist. The agent is not a link in a chain of events; nor is she connected in some to-be-specified way to phenomena that are either uncaused or only probabilistically caused. Rather, she just is the cause, in virtue of having the properties that she does. To be sure, this much the anti-passivist shares with the agent-causalist who otherwise endorses non-productive event-causation. But—and this is a second point—all causes are causes in the same sense, for the p-non-determinist. In all cases, to be the cause of x is to be the powerful particular that brought x about, or that did so in concert with other causes. That agents are causes therefore follows straight-forwardly from the metaphysics of causation, for the consistent anti-passivist; it is not an ad hoc claim to be independently established. These two points together yield a response to one sort of objection to agent-causation, voiced, e.g., by Laura Ekstrom, viz., that even if we grant that an agent is the cause of x, the agent-causalist cannot say what caused the agent to cause x (let alone to cause it when she did).¹⁵ Here is Ekstrom: “But how can I, as a persisting entity, make something happen (or come to exist)? Normally, when something happens, something else happened previous to it to cause it to occur. ... But I do not happen; I simply exist.”¹⁶ The anti-passivist will respond that the incredulity is unwarranted. It is nothing other than persisting powerful particulars that do make things happen, not just in the case of agents but in all cases. This is the view held by many powers theorists, including E. J. Lowe, for example.¹⁷

• (2) Intelligibility, control or luck

The p-non-determinist does not face the so-called intelligibility problem, viz., the difficulty of showing how it can be that agents are able to influence or direct the causal processes in which they supposedly play a role. I have elsewhere referred to this as the ‘traction’ problem,¹⁸ and it also often comes under the headings of ‘luck’ or ‘control.’ For obvious reasons, the problem is one that is faced by libertarians who hold that a free act has no cause at all. But it is also faced by compatibilists, and by those libertarians who hold that free acts are caused, but only caused probabilistically. The compatibilist has to show that agents have causal control over their acts despite the acts being caused not by them but by antecedent conditions plus laws of nature. The libertarian who believes that causation is nomological but probabilistic, meanwhile, faces challenges on both sides: on the one hand, she has to show that agents have control even in those cases in which the probabilistic regularity does not occur (i.e., during the exact gap in causation that is said to allow for there to be free will); on the other hand, she too, like the compatibilist, has to show that the agent has causal control even though the agent is not herself the cause.

By contrast, the traction/luck/control problem does not get started for the anti-passivist. Here’s why: from a powers perspective, causation is not something that occurs independently of the powerful particulars whose properties are displayed in any given causal process, such that a way then has to be found to connect those same powerful particulars to those processes. On the contrary, powerful particulars just are causes, in virtue of being that which is able to engage in productive activity. This is so in all cases, not just in the case of agency. It follows that in the case of agency the p-non-determinist will not face the challenge of (a) attaching causation itself to substance-causes, here agents; and/or (b) connecting agents to acts that are supposedly theirs, yet are presumed (i.e. by the event-causal passivist) to be either uncaused or caused by antecedent conditions rather than by the agent. The anti-passivist will be at a similar advantage in defending any position that trades upon the notion of source-hood (e.g., source incompatibilism).¹⁹ Any worry that this just pushes the problem back a frame, by introducing a causal relationship between entities and their powers, is unwarranted. The relationship between a powerful particular and its own essential powers-to-do is not itself causal: it is constitutive.²⁰ Salt does not cause its power to melt ice. Rather, salt has that power, as part and parcel of being salt.²¹

Human beings, one might think, have second-order powers (i.e., powers over our other powers), in addition to our first-order powers.²² Brian Ellis refers to such powers as “meta-causal.” As he has it, we can intentionally acquire new dispositions so as to be able to respond to our environment in new ways. Ellis describes this capacity for self-transformation as “a power to produce a causal power.”²³ I am inclined to count as second-order not only the power to gain new powers, but the power to exercise or refrain from exercising currently-had powers. Certainly both our first and second-order powers afford us a type of control over what we do that salt does not enjoy. But, as with salt, vis-à-vis the powers borne by members of its kind, members of our kind just have the power, if we do, to intentionally expand upon and/or exercise or refrain from exercising our first-order powers. It is in the nature of the case that the powers had by an agent in virtue of being the kind of substance that she is are had by her essentially; her having them is not a matter of luck. Notice too that, for the p-non-determinist, there is also no problem of attaching to the agent the doing of this or that particular act, at this or that particular time t. Being able to decide to do this or that at time t just is the second-order power that is had by powerful particulars that (or who) are sentient. What the agent’s motivations, objectives and/or considerations may have been, such that she did in fact do whatever it was that she did at 10am on a given Tuesday, are not pertinent, metaphysically, from this perspective. This said, the fact that control comes for free, as it were, entails neither that an agent will be able, in any given case, to exercise the powers that she chooses to exercise, nor, if she is so able, that she will succeed in bringing about any given outcome.

• (3) Moral responsibility

It is prima facie plausible that being metaphysically free is a necessary condition for being blame- or credit-worthy. In relation to this point, the focal question in the contemporary debate has become: “Is someone who could not have done otherwise morally responsible for her acts?”²⁴ I address the topic of alternative possibilities in the next section. Here I am interested only with: (a) the collapse of the category of metaphysically free agency into that of moral agency, and (b) the relationship between being the cause of x and being morally responsible for x.

With respect to (a), the anti-passivist will not be tempted to identify the idea of “being the cause of” (being a metaphysically free agent), with that of moral agency. Berofsky, who distinguishes the question of free will from questions about moral agency for reasons unrelated to anti-passivism, expresses this point nicely. He writes, “I believe that the free will problem as a problem generated by the prospect of determinism is a metaphysical problem. It cannot, therefore, be solved just by examining the concept of responsibility and the varied social and moral dimensions of agential evaluation.”²⁵ Indeed, in the context of setting out the metaphysics of agency, a focus on most (if not all) of the phenomena that, for the event-causal passivist, end up doing the metaphysical work that would otherwise be done by a real power of intentionality—such a focus will be regarded, by the anti-passivist, as misplaced. The salient metaphysical question for the anti-passivist will be whether or not human beings really do have a certain type of second-order causal power.

With respect to (b), there are three possible scenarios in relation to which, from a powers perspective, we may ask about the attribution of blame or credit, relative to the metaphysics of causation: first, one in which we assume that the agent has either caused or undertaken to cause a given outcome x; second, one in which we assume that the agent did not cause x (and did not undertake to do so); third, one in which we assume a powers-based version of determinism, i.e., what I called above p-determinism. In the case of (1), the anti-passivist will separate the question “Which powers, borne by which causal bearer, were involved in the causal display that brought about x?” from the question “Who or what is morally responsible for x having occurred?” Again, unlike the Kantian, for instance, who will say that free will just is the ability to act in accordance with moral law, p-non-determinist will distinguish between efficacy as such (which is what she will mean by ‘metaphysically free agency’ in the case of human actors, regardless of whether an act is done on purpose or by accident) and behavior that is motivated by moral concerns in particular. However, if an anti-passivist did want to say that causing x renders an agent morally responsible for x, either as a rule or in a given case, the dissolution of the luck or traction problem would facilitate the ascription of credit or blame (inasmuch as it does away with the need to “attach” the act to the agent at the level of metaphysics). The passivist, by contrast, might well find herself in the position of saying that moral responsibility is attached to “being the cause of,” yet be unable to attach causation itself to agents.

Scenario (2), meanwhile, invites us to think about transitivity. Here, by stipulation, the agent did not cause x. But perhaps she is blame or credit-worthy in virtue of being the indirect cause of x. The main point to be seen in this case is that the anti-passivist will not be able to invoke the notion of an unbroken causal chain in order to establish that Agent A is an indirect cause of x. On the contrary, it would have to be A itself—not some intermediary powerful particular, with powers other than those had by A—that has the power to bring about x before the anti-passivist will be able to say that A is the powerful particular that, or who, caused x. This aspect of an anti-passivist metaphysics makes indirect causation harder to theorize; it may be, for example, that what is called indirect causation is really co-causation. I do not want to comment upon whether or not an anti-passivist metaphysics allows for causal transitivity. At a minimum, however, if one thought that it did, a powers-based transitivity would be punctuated: stopping and then starting afresh with each qualitatively different power borne by each causally contributing powerful particular.²⁶ I say this because if there is a chain of outcomes—b then c then ... x—it will be an open question, at each juncture, what kind of thing is capable of producing the effect in question. It would not follow from the fact that A’s can produce B-outcomes that A’s have what it takes to produce x-outcomes. The need to re-conceive or even reject transitivity may set the bar higher for establishing instances of indirect causation, but it should not prevent the p-non-determinist from assigning credit or blame where either is thought due.

Scenario (3) is especially informative. (3) shows us that the realist about powers who is a p-determinist will have difficulty ascribing blame or credit to agents because it will be difficult for her to affirm the existence of agents, as distinct from bodies. To defend determinism in a powers-based environment, recall, is to reject the existence of powers to behave spontaneously, be they purported to be had by sentient or non-sentient entities, of which second-order agential powers are a type. But a putative “agent” with no powers to behave spontaneously would be gratuitous, ontologically; such an entity would have only and precisely those powers had by bodies. To put it differently, the powers proponent who is a p-determinist will have already answered, in the negative, the question of whether or not there exist powerful particulars who, being agents, bear powers of a sort that are not borne by entities that are not agents. The same will hold at the level of properties. Reductive or non-emergent mental properties—i.e., mental powers that are said to be, in the end, physical powers (be this via reduction, supervenience or function)—will show up as redundant in an anti-passivist ontological inventory, mere rhetorical artifacts designed defray the cost of identifying the mental with the physical. The notion of a non-reductive but nonetheless deterministic mental property, meanwhile, will be hard to make sense of. What would such a power be a power to do? What we learn from scenario (3) ultimately, then, is that since p-determinism allows for entities that are bodies but does not allow for entities who are agents, the p-determinist may not consistently talk about agents (rather than bodies) whose actions are causally necessitated. From a powers perspective, such an idea is a category mistake. Of course, an anti-passivist who believes in the existence of entities with agential powers will have thereby already denied p-determinism.

• (4) The possibility of having done otherwise

If standard nomological determinism is true, and at all times t there is only one possible future, then it appears as though an actor will never have been able to have done otherwise than as she did. And it seems plausible, as I have said, to think that someone who could not possibly have done otherwise is not morally responsible for her behavior. Harry Frankfurt’s now-classic case of Jones, the hapless fellow who could not have done otherwise and yet appears to be morally responsible, was designed to avert such a conclusion by showing that moral responsibility does not, in fact, require alternative possibilities. In Frankfurt’s version of the story, someone named Black wants someone named Jones to do x. But to quote Frankfurt, Black “prefers to avoid showing his hand unnecessarily.”²⁷ Thus, so long as Jones is on course to choose to do x on his own, Black will not get involved. But if Jones should decide against doing x, Black is prepared to go to whatever lengths necessary in order to ensure that he, Jones, ends up doing x after all. We can fill in the blank as to how Black will pull this off, says Frankfurt.²⁸ The only thing that matters is that Jones will, if it comes to it, be unable but to do x. As it happens, there is no need for Black to intervene. Jones does x all on his own. Therefore, says Frankfurt, Jones is morally responsible for having done x, even though in point of fact he could not have done otherwise.

What, if anything, changes about Frankfurt cases, if we transpose them into an anti-passivist metaphysical register? The assessment that will be made of Jones’ actual behavior is not hard to make out. Jones has, of his own accord, exercised his agential powers. If one thinks that agents are morally responsible for un-coerced displays of their causal powers, then there is no question but that Jones is responsible for his having done x. But what is going on with counterfactual-Jones? It cannot be that the same powers of the same entity are being expressed in the counterfactual scenario, since that situation is supposed to provide a fail-safe, relative to the expression of the properly agential powers of actual-Jones. Who, then—or what—bears the fail-safe powers? Black does: Black is determined, as it were, to get Jones to do x. To be sure, counterfactual-Jones is also involved. But how? Counterfactual-Jones (unlike actual-Jones) is not the author of the actions that are ostensibly his (else there would be no fail-safe, relative to actual-Jones). For the p-non-determinist, this fact suggests that counterfactual-Jones lacks the power in virtue of which entities count as being agents, rather than as being some other kind of powerful particular.²⁹ Counterfactual-Jones is not simply a living body that lacks consciousness, though. He lacks the defining powers of an agent, but he has powers that a zombie-like body that lacked the power of consciousness would not have. For one thing, he believes himself to have genuine agential powers. And he is an entity that is capable of having such a belief. Still, we know better. For again, if counterfactual-Jones were really in possession of agential powers, then actual-Black would have no back-up plan. It might be that we should think of counterfactual-Jones as an extremely sophisticated tool, vis-à-vis both counterfactual-Black and actual-Black. The bottom line, however, is that from a powers perspective, the counterfactual scenario is superfluous, featuring—as it does—a creature who (a) is not Jones and (b) is arguably not an agent at all. There is nothing to be learnt about actual-Jones’ moral culpability in acting as he does, let alone about his metaphysical standing qua agent, from knowing the degree of Black’s resolve to exercise his own agential powers in order to secure the end that he, Black, desires. The anti-passivist does not need to appeal to alternative worlds—or even to elaborate counterfactual scenarios—to affirm that while Jones is the cause of his own behavior, he is not, and would not be, the cause of Black’s behavior.³⁰ But were she to indulge the fiction of the dastardly Black, she would analyze it in the way that I just have.

Admittedly, the gist of said analysis has been claimed by non-powers theorists too, in the form of arguments related to causal histories and source-hood.³¹ If counterfactual-Jones isn’t really Jones, then even for the nomological event-causalist, the case is not one in which Jones himself could not have done otherwise. What difference does it make, then, to take a powers-based approach to Frankfurt cases? One answer is this: realism about causal powers settles issues of causal history, identification and source-hood by default. Just as the p-non-determinist does not have to find a way to attach causation to entities in order to secure causal control, she does not have to perform an added genealogical operation in order to establish whether or not the powers of a given powerful particular are its own powers. Jones’ powers qua agent are his necessarily; Black’s are Black’s. Moreover, even if the passivist has been able to sort out the causal histories so as to establish that actual-Jones is not counterfactual-Jones, she will still have to address the attachment or control problem, and the anti-passivist will not.

• (5) Acting for a reason

The concept of acting for a reason enters the free will debate first as a response to the intelligibility problem. Carl Ginet, for example, suggests that because uncaused acts are explicable via reference to reasons, they are not random.³² Others have made similar arguments to the effect that probabilistically caused actions attach in the right way to agents because such actions are caused in whole or in part by the having of reasons.³³ Randolph Clarke has suggested that even a viable agent-causal theory would have to be augmented by an event-causal account of acting for reasons.³⁴ Cognition is obviously a core component of the ability, if there is one, to spontaneously and intentionally exercise other powers. The concept of acting for a reason can therefore be expected to play an important role in any powers-based account of free will. Certainly this is so in the two classic versions published to date, those of Tim O’Connor and Jonathan Lowe. O’Connor argues that agents cause intentions-to-do-x-for-reason-R (where reasons are defined in internalist terms).³⁵ Lowe has it that agents cause x by way of willing to do so, in response to reasons(s) R (where reasons are construed along externalist lines).³⁶ Variations in underlying powers-based ontologies correlate with other powers-based accounts of agency. Brian Ellis, for example, maintains that powers are the essences of what he calls dynamic universals, or process kinds. He also holds that human beings are the bearers of second-order, “meta-powers.”³⁷ A fully worked-out Ellis-style approach might well feature agency as a process kind, the essence of which is precisely the type of 2nd-order power that Ellis attributes to fully functioning human substances. Similarly, an elaborated Mumford and Anjum-style theory might include the claim that agential powers, like other causal powers, require manifestation partners. All p-non-determinists will agree, however, that the ability to reason is an essential power of agents. This said, the phenomenon of acting for a reason plays a different theoretical role in the context of p-non-determinism than it does in passivist accounts of free will. Specifically, the p-non-determinist will deny both (a) the event-causal libertarian claim that reasons plus uncaused or only probabilistically caused events add up to agency; and (b) the event-causal compatibilist claim that free will can be defined a-causally, by reference to the presence or absence of certain kinds of motivating beliefs and/or desires or patterns thereof.

• (6) Unity, hierarchy, harmony

Plato suggested that the person who is driven by appetites for things other than goodness is not free. This idea shows up in contemporary debates as the view that an agent enjoys free will if her internal psychological desire-structure meets certain self-reflexive standards. Frankfurt, for instance, holds that an agent has free will, in virtue of which she is an agent and not what Frankfurt calls a “wanton,” just insofar as she is able to “have the will [she] wants.”³⁸ Frankfurt expresses the idea in terms of a postulated hierarchy of desires: to act freely is to do that which is consistent with one’s second-order desires concerning the content of one’s first-order desires. Gary Watson refers to Plato explicitly.³⁹ One acts freely, Watson says, when one’s actions are consistent with one’s values, rather than with one’s desires (if these motivational systems should conflict). Watson’s strategy is thus qualitative rather than quantitative, we might say, but the underlying idea is the same—and there are of course many examples of this type of approach. What becomes of the concern for psychic harmony if one rejects passivism? On the one hand: nothing. While the p-non-determinist has the advantage of being able to talk in a realist way about psychic forces or drives should she wish to, she will not be led by her metaphysics to weigh in on the dynamics of psychological well-being, let alone to defend any particular psychological theory if she does. On the other hand: something, and for precisely the same reason. What I mean by this is that the achievement of psychic harmony is not part of the definition of metaphysically free agency, from a powers perspective. To have free will at the level at which our having it has been called into question (whether by the p-determinist or by the passivist hard determinist), says the p-non-determinist, is to be the bearer of a certain kind of causal power, viz., the power to intentionally undertake to display one’s other powers. Whether or not agents do have such a power is an altogether different question from whether or not one has been able to avoid the psychological turmoil unfortunately made possible by the having of it.

• (7) Liberty

John Stuart Mill opens On Liberty by alerting his readers that “[t]he subject of this Essay is not the so-called Liberty of the Will so unfortunately opposed to the misnamed doctrine of Philosophical Necessity; but Civil, or Social Liberty.”⁴⁰ I have argued elsewhere that in fact it is not possible to draw a neat line between social and political philosophy and metaphysics.⁴¹ Still, Mill is right that the conundrum about free will and the matter of defining social-political liberty are not the same subject, as he puts it. The metaphysical question concerns the fate of agency, given a certain account of causation. What we post-Aristotelian moderns really want to know, if we are honest about it, is whether or not agency is actually possible in the face of causality, once the latter has been conceived in terms of nomological order, rather than in terms of activity. The social-political question, by contrast, has to do with a particular kind of situation in which we may or may not find ourselves. What is the nature of the good that we call “liberty,” and/or what are the essential characteristics of a social-political setting in virtue of which it may be counted as exemplifying and/or allowing for that good? Some argue that it is the absence of impediment; others maintain that it is the ability of actors to realize personal and/or collective aspirations. Isaiah Berlin, who may or may not have accurately portrayed Mill’s own views, gives us the terms “negative” and “positive” liberty, respectively, to denote these alternatives. A further consideration is whether or not liberty at this level presupposes any other social-political state(s) of affairs as a condition of its possibility—or any psychological one(s), for that matter. Answers tend to vary directly: the more robust the conception of liberty, the more will be thought to be required in order for it to obtain. Indeed, one who already held a positive conception of liberty might think that what I have parsed as requisites are not pre-conditions of liberty at all, but simply additional constitutive features of liberty.

The metaphysical question and the social-political question(s) are not the same, but they are related. If nothing else, any account of liberty at the social-political level will presuppose the metaphysics necessary to sustain it. Some combinations of social-political theory and metaphysics will be ruled out. Others will be logically possible, but less likely. The proponent of negative liberty, for example, has several options. She could be a (standard) determinist: the fixed order of the world, she would say, just happens to include both a given agent’s deterministically caused desire for x, and the fact of there being no impediment to her pursuing it. She could also be a standard compatibilist—although if she were the sort of compatibilist who maintains that metaphysical freedom is the absence of constraint relative to the pursuit of de facto desired ends, regardless of how those ends came to be the desired ones, she would be obliged to say that agents who lack the social-political liberty to do as they like are thereby stripped of their free will.⁴² Or she could be a metaphysical libertarian, insisting that it is only metaphysically unconstrained agents who stand to enjoy social-political liberty, even in the negative terms in which she has defined the latter. Other possible combinations may be less coherent. For example, while the concept of positive liberty does not logically preclude a commitment to (standard) hard determinism or to (standard) compatibilism, it nevertheless arguably sits most comfortably with standard metaphysical libertarianism. Just as she rejects the idea that one enjoys social-political liberty regardless of how limited one’s opportunities are (so long as nothing stands in one’s way), a proponent of positive liberty, one might imagine, will be unlikely to think that an agent may be counted as metaphysically self-determining if she is able to do and to desire only that which has been determined by laws of nature and antecedent conditions.

Once again the question is whether or not any of this changes if one denies passivism, and if so how. As before, the answer is: yes and no. On the one hand, a powers-based metaphysics entails neither a negative nor a positive conception of liberty. In principle, the powers theorist may adopt either model. On the other hand, entailment is not the only test. One point to appreciate, as a matter of good sense if not of deductive requirement, is that both the event-causal compatibilist and the event-causal libertarian are primed to conceive of political liberty in negative terms, even if they are not logically compelled to do so. The compatibilist will find it easy to think that there is no need for her to say anything that she has not already said when the discussion turns from metaphysics to social-political liberty: one is free if one is not-coerced, period. The libertarian, meanwhile, has already gone a step further. In her view, to be metaphysically free is to be not-even-coerced-by-causation. Thus for her too it will be easy to think that social-political freedom amounts to the absence of impediment.

Second, we have seen that the anti-passivist conceives of causation as a matter of things displaying their powers. What will be of interest with respect to human beings, from this perspective—whether or not we choose to call it liberty—is the exercise of our own powers, including our uniquely self-reflexive powers. Now, as I have said, the anti-passivist need not believe that agential powers exist. It is open to her to be a (powers-based) physicalist and/or a (powers-based) determinist. But if she does believe that human beings have agential powers—a capacity for self-conscious self-determination, we might say—then any discussion that it would make sense to call a discussion of liberty, or social-political freedom, will likely center on the cultivation and expression thereof. And relative to such a discussion the difference between p-non-determinism and passivism will matter. At the most basic level, the capacity in question (followers of Nussbaum might use the term “capability”) will be thought to be a real power or cluster of powers, something that, in virtue of its reality, really can be either fostered or thwarted.⁴³ By contrast, the passivist will claim that what appears to be a ‘power’ is in fact a sequence of states of affairs. In addition, as I noted in Sect. 2, the anti-passivist will not face any metaphysical traction or source-hood problems. Finally, as a matter of metaphysics, the contrast between positive and negative liberty will not be a sharp one, from a powers perspective. This is an important point. In both models of liberty, what is of metaphysical interest to the anti-passivist is the ability of the agent to act. The only thing that differentiates the two accounts with respect to action as such is the placement of a particular normative marker—a bar that distinguishes “mere” human efficacy from human flourishing. Should it be set low, at the absence of constraint? Or higher, so as to take in enabling conditions and/or the achievement of any specified outcome(s)? Regardless of how she answers, the p-non-determinist, unlike the passivist, will regard discussions of socio-political liberty as tracking the same causal power of self-determination that gives content to metaphysical freedom.

I argued in Part 1 that a powers-based metaphysics alters the causation component of the free will problematic by re-casting nomological phenomena as epiphenomenal products of the real causal powers of propertied things. I also suggested that one could already see how such an ontology might reconfigure the entire problematic. Having looked now at the agency “side” of the problematic, we may add that realism about causal powers (a) resolves important problems faced by event-causal defenders of free will; (b) reveals a range of non-causal formulations of free will to be digressions; and (c) simultaneously (i) clarifies the difference between the concept of free will and that of social-political liberty; and (ii) blurs the distinction between positive and negative liberty by connecting both to the agential power(s) of self-determination. We can now see even more plainly than we could before the sense in which adopting a p-non-determinist stance reconfigures the problematic as a whole. It is not just that some powers generate an appearance of nomological determinism but others do not. It is also that free will, from an anti-passivist perspective, just is causal determination. As I said at the outset: the greater the degree of causal determination (by the agent), the greater the degree of freedom.

3 Conclusion: powers and agent-causation

Otherwise-passivist proponents of agent-causation are in a position to say at least some of the things about agency that the p-non-determinist can say. It will be helpful, therefore, to identify the specific gain associated with a comprehensive powers-based approach. When I observe that the proponent of an otherwise-passivist agent-causal account is able to say much of what the realist about powers can say, I have in mind the kind of position associated with Roderick Chisholm, for example—one in which it is assumed both (a) that agents exist; and (b) that agents are productive causes; but also (c) that in all other situations, causation consists of sequences of events rather than in the display of real powers borne by substances. This “hybrid” model, as I shall call it, is the one that is rejected by those who dismiss agent-causation on the grounds that it posits a mysterious, “not-normal” kind of causation, supposedly operating along-side the regular, passivist kind.

The question, then, is this: how does the hybrid model of agent-causation compare to a comprehensive anti-passivist model? Exactly which lines of argument that are open to the p-non-determinist are available to the passivist who makes an exception for agents? There is some overlap. If it is true that agent-causation resolves issues of control, source-hood and the like, then any agent-causalist will be better able to handle the whole set of traction problems than will the event-causal libertarian. In order to secure the advantage it will be enough to think that agents have real causal powers, even if nothing else does. As soon as she is called upon to defend the existence of such powers, however, the hybrid agent-causalist will be hindered by her default commitment to passivism. I have argued elsewhere that a coherent Aristotelian metaphysics allows one to meet the challenge to non-reductive physicalism posed by Jaegwon Kim, for example, in particular the problem of over-determination that follows if one grants the causal closure of the physical.⁴⁴ Here what I want to show is that the hybrid agent-causalist will be at a disadvantage in responding to a critic such as Randolph Clarke.

Clarke has argued persuasively that event-causal libertarians cannot deal with control and luck issues, but he rejects agent-causation because it presumes substance causation, which he believes falls to an objection of the following sort.⁴⁵

To clarify, suppose we take for granted a view of events on which a typical event is an object’s o’s possessing a property P at a time t (Kim 1976). The object o is a constituent of the event, as are the property P and the time t. ... Now suppose that properties are what ground causal powers. Suppose that some property P grounds a power to cause an effect of a certain sort. Imagine that a certain substance s comes to possess P, and that until s acquires P the chance of the effect in question is very low. The occurrence of the event s’s acquiring P, or the obtaining of the state of affairs s’s possessing P, will typically raise the chance of the effect’s subsequence occurrence. There is nothing of this sort left for the substance s to do; the event or state of affairs takes care of it!⁴⁶

A consistent anti-passivist will be able to respond to Clarke as follows. First, s/he will deny that an event is equivalent to an object’s possessing a property at a time t. Ellis, for example, has it that events are instances of process-kinds; they are causal displays the essential natures of which are given by the powers of which they are the expression.⁴⁷ Possessing a property at time t, by contrast, is simply a fact regarding that which does indeed have the property in question at time t. Second, s/he will not agree that a power must be grounded by some further property. These are preliminaries, though. Third, and most important, the anti-passivist will be in a position to see that Clarke presupposes event-causation from the start. Insofar as s’s acquiring a power P (a non-essential power, or s would already have had it) makes it be more likely that a P-related outcome will occur, there is no reason to conclude, as Clarke does, that it is the acquiring of P that is where all of the (passivist) metaphysical “action” is, leaving s with nothing to do. Contra Clarke, the anti-passivist will conclude that s, having gained an additional power P, is now even more causally efficacious than it was before. Of course, the fact that the anti-passivist can say this does not settle the dispute. But it shows that Clarke has not settled it either. He has simply asserted that causation is as event-causalists believe it to be. And the point is that the hybrid agent-causalist is not well equipped to respond to such an assertion, since she too believes that causation as such is event-causal. I should add that there is also an advantage enjoyed by the anti-passivist that is not subtle at all. Specifically, the otherwise-passivist agent-causal theorist is clearly vulnerable to the objection noted above regarding the introduction of a second, non-standard type of causality that applies only to agents. Clarke raises this objection along with the one I have been discussing, and he is hardly alone in doing so. The p-non-determinist avoids the problem altogether. She claims only that many different kinds of substance exist, bearing many different kinds of causal power(s).⁴⁸

Above all, the hybrid model leaves the free will problematic intact. A comprehensive anti-passivist approach, by contrast—what I have called p-non-determinism—stands to both (a) make true and (b) reveal as ultimately superficial and/or ill-conceived the various opposing claims that comprise it (i.e., the problematic), both within each of its conceptual components (i.e., causation and agency) and between them. As we saw in Part 1, p-non-determinism dissolves the dichotomy between standard determinism and standard indeterminism. When we add in the analysis of Part 2, we see exactly how, in addition to better securing agent-causation, a powers-based ontology also allows both for what is true about event-causal compatibilism (viz., that agency is not at odds with causation) and for what is true about event-causal libertarianism (viz., that agents with free will have the ability to initiate causal processes)—thereby dissolving the contradiction between these terms too.

If this set of deliverances does not amount to a sublation of the free will problematic, it is hard to know what would.