In Causation and Free Will, Carolina Sartorio advances an actual-causal-sequence theory of free will: freedom in acting (or in omitting to act) is grounded in, and only in, the causal history of one’s conduct (and in whatever, if anything, further grounds that causal history). The view is offered as a reasons-sensitivity theory, one on which the relevant sensitivity is construed not modally or counterfactually but in terms of actual causation. The key to understanding reasons-sensitivity in this way, Sartorio maintains, is to recognize that the causal histories of conduct are far richer than is commonly thought. In particular, when we act freely, our conduct is caused not just by reasons to act as we do but also by numerous absences of reasons to do otherwise. The causal influence of these absences of reasons partly constitutes our sensitivity to reasons, and thus partly grounds our freedom.

My comments will focus on the following issues: what it is to act (as Sartorio puts it) on the basis of an absence of reasons; how causation by reasons and their absences relates to the reasons-sensitivity that is relevant to free will; what role absences of reasons play in Frankfurt cases; and what (if anything) grounds causation by absences of reasons. At some points, my aim is to invite further detail in Sartorio’s theory; at others, I’ll raise challenges to the non-modal conception of reasons-sensitivity that she advances.

1 Acting on the basis of absences of reasons

Responding to reasons, when one’s conduct is motivated by those reasons, is commonly referred to (by action theorists) as acting for reasons. Sartorio generally follows this practice. When it comes to the relevant phenomenon of responding to absences of reasons, she writes of acting on the basis of these absences. About both phenomena she takes a causal view: when an agent responds to a reason (or to an absence of a reason), the reason (or the absence) causes the agent’s conduct, and causes it in “the right way”—some way that is not deviant or wayward.

Still, the extent to which the two phenomena parallel each other depends, in part, on what practical reasons are. A Davidsonian tradition regards them as psychological states, notably desires for certain ends paired with beliefs about how to achieve those ends. In a standard case, when one acts for a certain reason, some such desire-belief pair, which jointly rationalize one’s action, non-deviantly cause one’s conduct. A rather different view sees practical reasons—at least normative ones—as facts. If it is raining, then the fact that it is raining might be a reason for me to take an umbrella with me when I go out. The fact can be such a reason whether or not I believe that it is raining (and whether or not I believe that there is anything to be gained by taking my umbrella). My acting for this reason will, presumably, involve my somehow mentally registering—I’ll say “recognizing”—the fact that it is raining; it might even be said to require my recognizing that fact as a reason to do a certain thing. A causal account will hold that some such recognition plays an appropriate causal role in my conduct; and if facts can be causes, the account might say, further, that my recognition must be appropriately caused by the recognized fact.

On familiar Davidsonian accounts, acting for a reason is not said to require either recognizing the psychological state that is the reason or recognizing that psychological state as a reason. (Proposals for ruling out causal deviance do not generally require possession of any such state of recognition.) A Davidsonian view of acting on the basis of the absence of a reason might similarly be said not to require recognition of that absence; it might require only causation (in a non-deviant way) by the absence of a desire, or by the absence of a belief.

On the alternative view of what reasons are, how should we understand acting on the basis of the absence of a reason? There are at least two options: it might be said that this requires recognition of the absence of the reason; or it might be said that it can suffice (provided the causal route is appropriate) that there is an absence of recognition of the reason. The first option is significantly more demanding, and on this construal, we act on the basis of few absences of reasons, for commonly when we act we don’t recognize many absences of reasons to do otherwise. Yet if reasons are taken to be facts, the first construal is the one that more closely matches, in form, acting for a reason.

All the same, an agent can be in an important respect sensitive to what reasons there are if her lacks of recognition of reasons are (appropriately) caused by absences of reasons, even if the absences of reasons do not cause recognition of those absences. If the lacks of recognition in turn cause conduct (in a non-deviant way), perhaps we can have all the sensitivity to absences of reasons that free will requires. Recognition might not play the role in acting on the basis of an absence of a reason that it plays in acting for a reason.

Sartorio sometimes (e.g., 126, 144) calls certain desires and beliefs reasons, and she sometimes (e.g., 125, 144) refers to reasons in ways in which we can refer to facts.Footnote 1 But she says that she will “remain as neutral as possible” on the nature of reasons, while assuming that “we have an intuitive grasp on what reasons are, and of what it is for agents to recognize reasons” (111). The mention here (and other places, where responsiveness is being discussed) of recognizing reasons seems to favor the alternative view over the Davidsonian one, though I don’t read much into this. In any case, providing a more detailed account of acting on the basis of an absence of a reason might require abandoning some degree of neutrality about what reasons are.

2 Reasons-sensitivity and free will

Our position on the nature of reasons should accord with the roles reasons play in our theories. One question to consider: is free will adequately conceived in terms of sensitivity of one’s conduct to reasons and absences thereof, when reasons are understood to be certain psychological states of an agent? There is reason to think that it is not.

Consider an agent whose view of things is utterly insane, someone incapable of appreciating the world for what it is. Her conduct might be sensitive to presences and absences of psychological states of herself—including desires for ends and beliefs about how to achieve those ends—while these states in turn remain insensitive to what is worth wanting or what is really the case. It is at best questionable whether we should attribute free will to someone so cut off from reality. (The point has been emphasized by Wolf (1990), and Fischer and Ravizza's (1998) reasons-responsiveness account also requires some degree of contact with reality.)

All the same, there are limitations on sensitivity to facts (and absences of facts) that seem not to detract from free will. The fact that it is raining, or the absence of this fact, might be inaccessible to me, and my behavior might be insensitive to it, due not to any deficiency on my part but to some external interference in the flow of information. Lack of causal influence in this kind of case (by the fact or by the absence of the fact) does not seem to diminish my freedom in any way. I’m incapable of being rightly informed about some matter, but I might be as free as can be to choose (poorly informed as I am) between my options.

In some other cases, lack of actual causation by reasons (construed as facts) or absences thereof seems not to undermine either free will or a relevant kind of reasons-sensitivity. I might act negligently when some (actual) fact is sufficient reason not to do what I do. In my negligence (let us suppose) I fail to notice it. This reason to do otherwise doesn’t cause my action. There might be a wide range of absences of other sufficient reasons to do otherwise that likewise don’t cause my conduct on this occasion, since (we may imagine), given my negligence, I wouldn’t have noticed these reasons had they been present. Yet I might act freely in such a case. I am capable of noticing the sufficient reason to do otherwise that is present, and I’m able to act on my recognition of it. I might be capable also of recognizing many other sufficient reasons to do otherwise that are absent, should they be present, and able to act on them, even though, in my negligence, I wouldn’t. These capacities and abilities are important facets of the freedom with which I act on this occasion; they constitute what may fairly be called a relevant kind of sensitivity or responsiveness that I have to these reasons. But this reasons-sensitivity is not a matter of causation of my conduct by these reasons and absences of reasons. (Similar things may be said about cases of perversity or akrasia, with the difference that in these cases I might recognize the reasons to do otherwise that are present.)

Sartorio’s theory requires, for one’s conduct to be free, that there be an appropriate range of reasons to which one is sensitive, whose presence or absence causes one’s conduct. And that might be found in the case I’ve just considered. Thus, her theory might yield the right verdict about the agent’s freedom in this case. Still, my capacity to notice and ability to respond to the reasons that, in my negligence, I do and would fail to notice is a central aspect of my freedom in this case. My sensitivity to these reasons is a modal property of me, one relevant to the freedom of my conduct on this occasion, but one that is not a matter of causation of my conduct by these reasons (or absences of reasons). There is reasons-sensitivity relevant to free will that, it appears, outstrips the actual causal history of conduct.

3 Frankfurt cases

In a Frankfurt case of action, an agent (Frank, let’s say) performs an action on his own, without being forced by anything or anyone to do that thing, but there is someone else (a nefarious neuroscientist) around who is monitoring Frank and would have intervened and ensured that Frank do what he did had Frank not done it on his own. In fact, the neuroscientist didn’t intervene at all; there was no need for her to do so. As do many writers, Sartorio holds that there can be cases of this kind in which the monitored agent acts freely. The basic argument (32–33): imagine a case that is as similar as possible to a certain Frankfurt case except that the neuroscientist is only observing and has no intention of intervening; call this the normal case. There can be a normal case of this kind in which Frank acts freely; take this to be the one we are considering. Since the neuroscientist doesn’t actually intervene in the Frankfurt case, the causal history of Frank’s conduct in that case is relevantly similar to that of his conduct in the normal case. Now, Sartorio maintains, freedom supervenes on (relevant elements of) the actual causal history of conduct. (It is grounded in, and only in, the causal history of the conduct in question [and in whatever, if anything, further grounds that].) Since Frank acts freely in the normal case, and since the causal histories of his conduct are relevantly similar in the two cases, he acts freely as well in the Frankfurt case.

As I’ve said, on Sartorio’s view, the freedom-grounding causal histories of free actions include much causation by absences of reasons. Is causation by absences of reasons unaffected when we move from a normal case to a Frankfurt case? If not, this line of argument for the freedom of agents in Frankfurt cases won’t go through.

Sartorio points out that there will be some differences in absence causation between the normal case and the Frankfurt case. Absence causation is prolific. One absence that causes Frank’s action in the Frankfurt case is the absence of a space alien who stands ready and able to destroy Frank if any neuroscientist intends to interfere with Frank. The absence of such a being is not a cause of Frank’s action in the normal case. But these elements of the causal histories are (intuitively) irrelevant to free will. Differences in causation by absences of reasons, however, are another matter.

Suppose that reason R is sufficient reason for Frank not to do what he did, and that R was absent. Did its absence cause Frank’s action? If absences of reasons can be causes, then in the normal case this absence might be a good candidate for a cause of Frank’s action. It’s not so clear that (even assuming that absences can be causes) this is so in the Frankfurt case.

Causation by absence is often characterized in terms of counterfactuals of causation. Accounts of causation of an (actually occurring) event e by the absence of an event of kind C employ the counterfactual: had there occurred an event of kind C, it would have prevented e (by causing some event incompatible with e, or by causally interacting with, and thereby cutting short, a process that in fact culminated in e). Satisfaction of the causal counterfactual is sometimes said to be necessary and sufficient for causation by the absence, sometimes said to be necessary, and sometimes said to be sufficient.Footnote 2

None of these views counts the absence of R in the Frankfurt case as a cause of Frank’s action (the absence doesn’t satisfy what is said, on some of the views, to be a sufficient condition). And some of them rule that absence out as a cause of the action (the absence doesn’t satisfy what is said, on these views, to be a necessary condition). For in the Frankfurt case, had R been present, it would not have prevented the actually performed action from occurring (that action would still have been performed).

Sartorio doesn’t advance an account of absence causation. She does offer what she says is a necessary condition for causation, which she argues can be satisfied by absences of reasons to do otherwise in Frankfurt cases. A cause, she maintains, is a difference-maker, not in the sense that had it been absent, its effect would not have occurred, but in the sense that had it been absent, its absence would not have caused that same effect (94). The absence of the absence of R would (it seems) be R, so to determine whether the absence of R satisfies the proposed requirement for being a cause of Frank’s action, we consider whether R, if present, would have caused that same action.

Imagine that R is the very sign that the neuroscientist is looking for to tell her that she needs to intervene. Had it been present, it would have caused her recognition of it, which, let’s suppose, would have caused her to prevent Frank’s recognition of R. Not recognizing R, Frank would have acted as he in fact acted. The neuroscientist’s intervention would have been a cause of Frank’s action, and R would have been a cause of the intervention. Would R have been a cause of Frank’s action?

If causation is always transitive, the answer is affirmative. Sartorio argues that causation is not always transitive. And perhaps her view is that transitivity fails in the (alternative) scenario we are considering here. It might be said to be an instance of threat-cancellation: the presence of R is a threat to Frank’s performing the actually performed action, and that action results in the alternative scenario, where R is present, only because something—the neuroscientist—counteracts the threat. Threats that are cancelled in this way, Sartorio argues, are commonly not causes of the outcomes that they threaten, even if they cause interventions which in turn cause those very outcomes.

I see the appeal of this argument, but I’m not sure what to make of it. But note that ruling out R as a cause of Frank’s action, had R been present, gives us only a necessary condition for counting the absence of R as a cause of the action in the actual scenario. The absence is not ruled out, but it’s not ruled in, either.

There is a different way in which Sartorio sometimes characterized difference-making that might be thought to give us grounds for counting the absence of R as a cause of Frank’s action. Causes, she says, “make a contribution that their absences wouldn’t have made” (94). If the absence of R makes a causal contribution to Frank’s action that R would not have made, then perhaps a sufficient condition for the absence’s counting as a cause of the action is satisfied.

But whether the absence of R makes any causal contribution to Frank’s action, in the Frankfurt case, is just what’s at issue. The question of whether it is a cause is, it seems to me, unresolved.

There is one aspect of Sartorio’s discussion of this question that I find puzzling. She mentions in a note (131 n. 23) that she remains tempted to affirm the following: whereas an action can cause an outcome even if the outcome would still have occurred in the absence of the action, an omission cannot cause an outcome if the outcome would still have occurred in the absence of the omission. [She had advanced this principle in an earlier paper (Sartorio 2005)]. She says that she does not think that the asymmetry extends to presences and absences generally. But why would it hold only for actions and omissions?

If practical reasons are facts, then the fact that someone has performed a certain action (offering Frank a large reward for doing a certain thing) can be a reason to act. When such an act is omitted, such a reason is absent. Why would the omission of the action and the absence of the reason differ in their causal powers, such that the former cannot cause an overdetermined effect, while the latter can? Indeed, since omissions themselves seem to be absences (of actions), the distinction is curious.

Consider the following case, which Sartorio calls Accomplice:

I witness a man being robbed and beaten but I decide not to call the police. The phone lines were in working order when I made the decision. However, unbeknownst to me, the robber’s accomplice was keeping an eye on the neighbors who were witnessing the attack. Had I picked up the phone, he would have immediately cut the phone lines off (81–82).

About this case, Sartorio says that my failure to try to call the police does not cause their not being called. That seems right. Given the would-be intervention by the accomplice, the absence of an attempt by me is causally irrelevant to the outcome. But the absence of the fact that I make such an attempt is equally causally irrelevant. The omission and the absence seem on a par here.

Perhaps Sartorio is wrong to hold onto the asymmetry that she had previously advanced, or perhaps there is justification for restricting it as she suggests in her book. I think more would need to be said to establish the latter.

4 Causation by absence

As I’ve stated Sartorio’s theory (and as she often states it), reasons-sensitivity is partly a matter of causation by absences of reasons. It might thus be thought that the theory is committed to the possibility of absence causation. But that is not so. The commitment, Sartorio explains, is weaker, to there being some relation—or plurality of relations—in which conduct can stand to both presences and absences, such that freedom is grounded in conduct’s being so related. Although she thinks that causation is the grounding relation, she allows that, if absences cannot be causes, some other “quasi-causal” relation might do some of the grounding work.Footnote 3

Even this qualified commitment implies an interesting position on absences: there are things that are absences of things. For if there are not, then there’s nothing that is an absence of a reason and stands in any relation (causal, quasi-causal, or what have you) to anything. With the theory so committed, one question that arises is: what kind of thing is an absence of a thing? I don’t see that Sartorio offers an answer.

Some theorists (e.g., Lewis) maintain that absences of things aren’t things of any kind (there is nothing that is an absence of a thing) and yet there is absence causation. Causation, in these cases, is non-relational. We might say that in absence causation we have a “quasi-relation.” A qualified version of Sartorio’s commitment, then, might require some relation, quasi-relation, or plurality of one or both, as grounds for freedom.

However we state the commitment, we might ask what, if anything, grounds causation (or quasi-causation) by absences. For Sartorio accepts that any grounds of this are among the grounds of free will.

Primitivism—nothing more fundamental grounds causation (or quasi-causation) by absences—is an option, though I am not aware of anyone who takes this view. The proposals that I mentioned in Sect. 3, couched as they are in terms of counterfactuals of causation, suggest that absence causation (or quasi-causation) is grounded at least partly in what (if anything) serves as truthmakers of these counterfactuals.

Consider: why might it be that, if some reason R that is in fact absent were instead present, R would causally interact with conduct-generating processes within me such that, acting for reason R, I would not do what I actually do but would instead do otherwise? This might be so because I have certain powers—a capacity to recognize R and an ability to act on that recognition—powers that, were R present, would be manifested, in part because they would not be masked by other powers that would be co-present (no nefarious neuroscientist would interfere). If powers of things serve as truth-makers of the relevant counterfactuals of causation, these powers ground causation (or quasi-causation) by absence. And if freedom is grounded partly in causation (or quasi-causation) by absences of reasons—and by whatever, if anything, further grounds this—it is grounded partly in modal properties of agents, in their capacities and abilities. Reasons-sensitivity, even if it can be characterized in terms of actual causal histories, remains something modal, for some of what grounds the causal histories is modal.

In Sect. 2 I raised the question of whether an agent’s sensitivity to reasons, on some occasion, is equivalent to the causation of her conduct then by reasons and absences of reasons. The further question here is whether, even if there is such an equivalence, which, if either, of these phenomena is more fundamental. If many of the causes (or quasi-causes) of conduct count as causes (or quasi-causes) because of the agent’s sensitivity to certain reasons, the latter might be both explanatorily and metaphysically more basic. An actual-causal-sequence theory of free will might then not be incorrect, but it will leave free will’s deeper metaphysical ground unrevealed.