Indirectly Free Actions, Libertarianism, and Resultant Moral Luck

Abstract

Martin Luther affirms his theological position by saying “Here I stand. I can do no other.” Supposing that Luther’s claim is true, he lacks alternative possibilities at the moment of choice. Even so, many libertarians have the intuition that he is morally responsible for his action. One way to make sense of this intuition is to assert that Luther’s action is indirectly free, because his action inherits its freedom and moral responsibility from earlier actions when he had alternative possibilities and those earlier directly free actions formed him into the kind of person who must refrain from recanting. Surprisingly, libertarians have not developed a full account of indirectly free actions. I provide a more developed account. First, I explain the metaphysical nature of indirectly free actions such as Luther’s. Second, I examine the kind of metaphysical and epistemic connections that must occur between past directly free actions and the indirectly free action. Third, I argue that an attractive way to understand the kind of derivative moral responsibility at issue involves affirming the existence of resultant moral luck.

Libertarianism is the view that free will is incompatible with causal determinism and that human beings at least sometimes act freely.¹ A prominent family of libertarian views that Randolph Clarke (2000, p. 23) identifies as “action-centered” are distinguished by the claim that an agent acts freely only if she has alternative possibilities open to her at the moment of choice.

Daniel Dennett (1984) has famously offered alleged counterexamples to action-centered libertarianism. For example, Martin Luther refuses to recant his theological views by saying “Here I stand. I can do no other.” Supposing that Luther spoke truly, he lacked alternative possibilities at the moment of choice, but, intuitively, he is morally responsible for his action.² Thus, there are cases with respect to which action-centered libertarians appear to be committed to three jointly inconsistent claims: (i) Luther is morally responsible for the action, (ii) Luther’s being morally responsible for the action requires that it is a free action, and (iii) Luther does not act freely because he could not have done otherwise at the moment of choice.

Action-centered libertarians (hereafter, libertarians) can respond in two basic ways. On the one hand, they can deny (i). That is, they can deny the intuition that Luther acts freely and morally responsibly.³ On the other hand, they can deny (ii) or (iii). In other words, they can accommodate the intuition that Luther is at least morally responsible for his refusal either by affirming that people can be morally responsible for certain kinds of actions that are not free actions (van Inwagen 1989) or by affirming that certain kinds of free actions do not involve alternative possibilities at the moment of choice (Kane 1996b).⁴ What these latter two views have in common is that they trace Luther’s moral responsibility for the refusal to earlier character-forming actions when he had alternative possibilities at the moment of choice. I focus on the Robert Kane-style version of this tracing response, and return to the disagreement between him and Peter van Inwagen in the next section. Making sense of Luther’s free and morally responsible refusal, then, involves distinguishing between directly and indirectly free actions. An agent performs a directly free action only if she has alternative possibilities at the moment of choice. And an agent performs an indirectly free action only if she has performed directly free actions in the past, and those past directly free actions have cultivated character for which she is morally responsible such that her will is set one-way in the present circumstance.

Libertarians such as Kane (1996b), Timothy O’Connor (2009), Timothy Pawl and Kevin Timpe (2009, 2013), Timpe (2011, 2014), van Inwagen (1989), and Dean Zimmerman (2011) embrace the basic distinction between what I am calling directly and indirectly free actions (although at least some of them think that what I am calling ‘indirectly free actions’ are more aptly labeled ‘indirectly morally responsible actions’). Refer to this group as Indirect Libertarians.

Nevertheless, Indirect Libertarians to date have developed only partial accounts of indirectly free actions. This is surprising precisely because they assign indirectly free actions a central role in their accounts of morally responsible agency. Additionally, an account of indirectly free actions is important for theistic libertarians who believe that freely developing character before death is vital to explaining why we can act freely and morally responsibly in heaven, a state in which it is impossible to do wrong (Brown 2015; Pawl and Timpe 2009, 2013; Sennett 1999; Timpe 2014; Zimmerman 2011).

In this paper, I provide a more developed account of indirectly free actions. The contribution I make is to compile disparate and partial accounts into a fuller account, refine some of the partial accounts to make them more plausible, and fill in gaps left by the compilation of these partial accounts. I proceed as follows. First, I explicate relevant parts of the metaphysical nature of an indirectly free action. Second, I offer an account of the metaphysical contribution that an agent must have made through her past directly free actions to perform an indirectly free action. Third, I provide an account of the epistemic connection an agent must have had between her past directly free actions and her indirectly free action. Fourth, I argue that an attractive way to understand the degree of an agent’s derivative moral responsibility for an indirectly free action requires accepting resultant moral luck.

Three preliminary points are in order. First, for the purposes of exploring the nature of indirectly free actions, I assume that directly free actions are unproblematic. Thus, I set aside well-known difficulties such as the luck objection to libertarianism (see Levy 2011; Mele 2006; van Inwagen 2002). Second, because I assume that directly free actions are unproblematic, I leave their metaphysical nature unspecified; I do not enter intermural debates between agent-causal and event-casual libertarians. Third, although I intend the account of indirectly free actions that I develop to be for libertarians, it can apply mutatis mutandis to history-sensitive compatibilists such as John Martin Fischer and Mark Ravizza (1998, p. 50), who also endorse the idea that indirectly free actions are important in a general account of moral responsibility.

1 The Nature of Indirectly Free Actions

I begin with some terminology. I use the term practical necessity to refer to a cluster of dispositions of sufficient strength such that the agent with those dispositions must act or omit acting in a particular kind of way in certain circumstances. A practical necessity is roughly the same idea as Harry Frankfurt’s (1988, p. 86) “volitional necessity” and Bernard Williams’s (1993, p. 59) “moral incapacity.”⁵ I use the term practically necessitated action for an action that an agent must perform in certain circumstances because of her practical necessity. Lastly, I define the term indirectly free action to refer to any practically necessitated action for which an agent is praiseworthy, blameworthy, or neutral-value-worthy. The difference between practically necessitated actions and indirectly free actions is that only the latter must have the kind of history that confers praiseworthiness, blameworthiness, or neutral-value-worthiness to them from earlier directly free actions.

One theme in Frankfurt and Williams is that practical necessities are importantly related to an agent’s character and reasons. Frankfurt (1988, p. 86) asserts that Luther’s reasons were “too good” to do anything else, and Williams (1993, p. 65) maintains that Luther’s reasons were “totally decisive” (cf. Dennett 1984; Kane 1996b; O’Connor 2009; Pawl and Timpe 2009; Stump 1999; Timpe 2014; Zimmerman 2011). Although it would be interesting to analyze the metaphysics of exactly how the agent’s reasons practically necessitate the indirectly free action, I do not take on that project in the interest of exploring other questions. I follow Timpe’s (2014) appeal to van Inwagen’s (1989, p. 407) principle that describes a conditional relationship between certain kinds of character and reasons, on the one hand, and a practically necessitated action, on the other:

If X regards A as an indefensible act, given the totality of relevant information available to him, and if he has no way of getting further relevant information, and if he lacks any positive desire to do A, and if he sees no objection to not doing A (again, given the totality of relevant information available to him), then X is not going to do A (italics in original).⁶

Two points are worth highlighting. First, van Inwagen (1989, p. 409) equates the force of ‘X is not going to do A’ in the consequent with ‘X cannot perform A’. Second, this principle is about what a person omits doing, but the same point applies to what a person does: so, if X regards A-ing as the only defensible act (given the totality of available information and there is no way to get more), if X desires to A, and if X has no objection to A-ing (given the totality of available information), then X must perform A. For example, a mother is offered a nickel to cut up her daughter into tiny pieces; she refuses the offer and could not have accepted it (Stump 1999, p. 323). Since the mother (i) believes refusing the offer is the only defensible action (given all of her information and believes that is no way to gather more), (ii) desires to refuse the offer, (iii) believes that there is no objection to refusing the offer (again, given all of her information), she must refuse the offer. Supposing that we understand reasons as pairs of beliefs and desires, the conditional illustrates the way in which X’s reasons constrain X’s action (Timpe 2014).

Given that practically necessitated actions are determined by the agent’s reasons and character and that libertarians think that free will is incompatible with determinism, is an indirectly free action a misnomer? Van Inwagen (1989, p. 418) believes that what I have called a ‘directly free action’ is the only kind of free action there is. Other philosophers who agree with van Inwagen are nevertheless happy to refer to indirectly free actions as actions that are “in a sense” free (O’Connor 2009, p. 122) or “freely undertaken” (Zimmerman 2011, p. 177). Yet other philosophers including Kane (1996b) and Pawl and Timpe (2009, p. 400) suggest that there are different kinds of free actions, and so are glad to countenance indirectly free actions as a bona fide kind of free action. Plausibly, the issue here is whether to define free action in terms of the control required for moral responsibility or in terms of a more robustly metaphysical notion concerning alternative possibilities. I side with Kane, Pawl, and Timpe, but those who follow van Inwagen can substitute ‘indirectly morally responsible actions’ for ‘indirectly free actions.’ In the next section, I consider the following question: what kind of metaphysical contribution must the agent have made via past directly free actions to be morally responsible for a present practically necessitated action?

2 The Metaphysical History Condition

In general, Indirect Libertarians do not say much about the kind of metaphysical contribution that the agent must have made to her character via past directly free actions. Van Inwagen (1989, p. 420) recites the platitude about having performed past directly free actions that influenced the practical necessity. Pawl and Timpe (2009), Timpe (2014), and Zimmerman (2011, pp. 176–177) say little more. Kane (1996a, b), however, makes explicit two historical conditions for indirectly free actions that other Indirect Libertarians tacitly assume. After I present Kane’s two conditions (one negative and one positive), I offer two more.

Kane’s (1996a, p. 82) negative condition is that the agent’s directly free actions need not be the only cause of the practical necessity: “Of course, for Luther to be responsible for his character and motives does not mean that he must have been their sole and complete cause, as we said, or created them ex nihilo.” Thus, the agent need not be morally responsible for the whole practical necessity for it to generate an indirectly free action. I suggest also that to be fully morally responsible for an indirectly free action, one need not be morally responsible for the whole practical necessity from which it issues. In terms of Thomas Nagel’s (1979, p. 28) categories, being fully morally responsible for the indirectly free action is compatible with the practical necessity’s being “constitutively lucky” to some degree; constitutive luck occurs when an agent has a disposition, capacity, or temperament that is at least in part non-voluntarily acquired.⁷ So, at least one whole disposition or disposition part need not be formed by directly free actions in a disposition cluster that constitutes the practical necessity that gives rise to an indirectly free action.⁸

Kane’s (1996a, p. 77) positive condition is that past directly free actions or omissions must have made a difference to the agent’s character such that the practical necessity counterfactually depends on having performed those free actions or omissions:

For you to be responsible for your will’s already being set the way it is, some of your past actions must have ‘made a difference’ in the sense that they were parts of the actual web of causes and conditions that resulted in your will’s being set the way it is, and if these actions of yours had not been parts of that web of causes and conditions, it would not have resulted in your will’s being set the way it is.

So, the agent must have performed past directly free actions that have “made a difference” to a disposition that eventually helped to form the practical necessity (cf. Kane 1996b, pp. 35, 39–40); if the agent had not directly freely acted or omitted in the way that she did, she would not have formed the practical necessity that necessitates her indirectly free action.

While I accept both of Kane’s conditions, they do not adequately discriminate the degree of a person’s praiseworthiness or blameworthiness for an indirectly free action. I appeal to two case pairs to motivate the need for further conditions.

First, consider Chuck and Instant Chuck. Alfred Mele (2006) introduces Chuck as someone who is wholeheartedly murderous. In his youth, Chuck had the bad constitutive luck to enjoy torturing animals. Even so, he had the good constitutive luck to feel guilty before, during, and after the torturing event, which provided many occasions for him to reconsider his ways. In part because he valued being unfettered by conventional morality, he set out to squelch that guilty disposition by torturing animals and by bullying vulnerable persons. After enough time on this reformation program, we may suppose that Chuck is one action away from developing a practical necessity to murder in certain circumstances. Additionally, suppose that God creates Instant Chuck, who is a distinct person that comes into being at that instant, as a perfect duplicate of Chuck (but with fake memories). So, Instant Chuck is also one action away from developing a practical necessity to murder. Both Chuck and Instant Chuck perform the action, develop the practical necessity, and indirectly freely murder someone soon after. They both satisfy Kane’s metaphysical conditions for indirectly free actions. They perform directly free actions that make a difference to their practical necessities and their practical necessities counterfactually depend on their directly free choices. As O’Connor (2009, p. 123) would rightly point out, however, there is a moral difference between them.⁹ I indicate what this difference is after the next pair of cases.

Second, consider Revamped Joe and Slow-Developing Joe. Revamped Joe is an average human being, and is not particularly generous. He has an opportunity to take a pill that would create in him a kind of wholehearted generosity. He is torn. But he takes it, and is revamped. Subsequently, he finds himself with the practical necessity to be generous and he indirectly freely gives generously. Slow-Developing Joe, however, develops a qualitatively identical practical necessity to be generous in the hard-won way—namely, through choosing to be generous and omitting stingy actions over a long period of time. Both Revamped Joe and Slow-Developing Joe contribute to their character through directly free actions and their practical necessities counterfactually depend on those directly free actions. Again, it is intuitive that there is a moral difference between them.

What kind of moral difference obtains in these cases? My diagnosis is that even though Chuck and Instant Chuck are equally bad persons, Chuck deserves more blame than Instant Chuck both for his practical necessity and his indirectly free action (cf. Hartman 2017, pp. 124–139). Likewise, although Slow-Developing Joe and Revamped Joe are equally good persons, Slow-Developing Joe deserves more praise than Revamped Joe both for his practical necessity and his indirectly free action.

Let us consider first the claim about Chuck’s being more blameworthy for his practical necessity to murder and Slow-Developing Joe’s being more praiseworthy for his practical necessity to give generously. Why think that claim is true? Chuck’s and Slow-Developing Joe’s practical necessities are more of their own doing than the practical necessities of their counterparts. Chuck performed many actions to develop his practical necessity, whereas Instant Chuck performed only one such action. Chuck also had more opportunities than Instant Chuck to avoid having that practical necessity. Likewise, Slow-Developing Joe performed many more actions and omissions to bring about his practical necessity in comparison with Revamped Joe; he also expended greater moral effort to develop the practical necessity. The general lesson is that the degree to which a practical necessity is of the agent’s own making has implications for her praiseworthiness or blameworthiness with respect to the practical necessity.

If we grant these differential praiseworthiness and blameworthiness attributions for practical necessities—as I think we should—does it follow also that Chuck and Slow-Developing Joe are respectively more praiseworthy or blameworthy in virtue of their indirectly free actions? I think it does. As O’Connor (2009, p. 123) recognizes in the context of directly free actions, the degree of a person’s freedom and responsibility for an action is partially a function of the degree to which the factors that influence her choice are of her own making. This is one reason why, in general, properly functioning teenagers are less free and responsible than properly functioning adults, and it is one reason why, in general, properly functioning adults without manipulated character are more free and responsible with respect to their actions than properly functioning adults who have their character manipulated to various degrees. This idea plausibly applies also to indirectly free actions; the degree of a person’s indirect freedom and responsibility for a practically necessitated action is partially a function of the degree to which her practical necessity is of her own making.¹⁰ So, Chuck’s practically necessitated murder is more indirectly free and blameworthy than Instant Chuck’s practically necessitated murder, because constitutive luck plays less of a role in generating Chuck’s indirectly free actions.¹¹ The same applies mutatis mutandis to Slow Developing Joe but with respect to circumstantial luck, which is the kind of luck that occurs when the morally significant challenges or opportunities a person has are determined by factors beyond her control (Nagel 1979, p. 28).

Here, then, is a third metaphysical condition: the degree of a person’s praiseworthiness or blameworthiness for an indirectly free action is at least partially a function of the agent’s history of forming the practical necessity, which includes (i) the quantity and quality of her directly free actions and omissions that make a difference to acquiring the practical necessity and (ii) the quantity and quality of her opportunities to avoid acquiring the practical necessity.

Nevertheless, these three conditions do not exhaust all that needs to be said about the historical metaphysics of indirectly free actions. For we do not think that just any directly free action or omission on which a practical necessity counterfactually depends should count toward her moral responsibility for her practical necessity and practically necessitated actions. For example, consider Jan who chooses to become courageous for its own sake. Later in life, however, she develops a practical necessity to murder, and her courage is taken up as part of the bundle of dispositions that constitute that practical necessity. Jan*, however, is a qualitative duplicate of Jan with only the historical difference that she is courageous as a matter of constitutive luck. It follows, then, that Jan’s practical necessity is more of her own doing than Jan*’s practical necessity. So, the third condition appears to imply that Jan is more blameworthy for her murderous practical necessity and practically necessitated action, which is the wrong result.

What we need, then, is a fourth metaphysical condition that adequately specifies the kinds of directly free actions (courageous, lascivious, just, etc.) that can transfer praiseworthiness and blameworthiness to a certain kind of practical necessity and can partially fix the degree of a person’s praiseworthiness and blameworthiness for an indirectly free action.¹²

One might think that the fourth condition is that directly free actions cannot be related by luck to the practical necessity. Since Jan’s courageous directly free actions performed earlier in life are related by luck to the later developed practical necessity, those courageous directly free actions are not relevant to determining how blameworthy she is for her indirectly free action. But this proposal has a costly implication. Because virtually all directly free actions that eventually make a difference to acquiring a practical necessity do so at least partially as a matter of resultant luck—that is, luck in the way one’s action or omission turns out (Nagel 1979, p. 28)—this proposal implies that virtually no directly free actions count toward the agent’s moral responsibility for her practical necessity, which implies that there are virtually no indirectly free actions. Why think that almost all directly free actions that make a difference to having a practical necessity do so partially as a matter of resultant luck? Consider the way in which a person’s action or omission affects her dispositions is typically insufficient to yield a practical necessity. It is only after many such choices and external influences are diachronically compounded that a person’s dispositions become sufficiently strong to constitute a practical necessity. Thus, it is almost always at least partially a matter of resultant luck whether a particular directly free action or omission changes her dispositions in a way that eventually gives rise to a practical necessity.

Another proposal is that the directly free actions that should contribute to the degree of moral responsibility for a practical necessity are those that are sufficiently like the practical necessity. The relevant kind of similarity involves at least moral valence consistency. So, only morally bad (and neutral) directly free actions can contribute to morally bad practical necessities and indirectly free actions, and only morally good (and neutral) directly free actions can contribute to morally good practical necessities and indirectly free actions. Directly free actions in which the agent aims to acquire courage for its own sake have a good moral valence. Thus, according to this baseline consistency criterion, the courageous directly free actions are not candidates for the relevant kind of counterfactual dependence and do not count toward the degree of Jan’s blameworthiness. This proposal, then, nicely solves the problem.

This similarity condition, however, requires further specification. I propose that intrinsic features of the practical necessity should determine the parameters of further specification. An agent is practically necessitated to murder, for example, only in a certain kind of circumstance and in a certain kind of way—not always and everywhere. If the practically necessitated murder is stably motivated, for example, not only by an interest in violence but also in greed and envy, then violent, greedy, and envious directly free actions that make a difference to having the practical necessity transfer moral responsibility to it and to the indirectly free action. Thus, the kinds of directly free actions that are relevant to moral responsibility transfer are those that adequately resemble the intrinsic nuances of the practical necessity itself.¹³

In summary, Indirect Libertarians should endorse four historical metaphysical considerations on indirectly free actions. The first two considerations are necessary conditions on acting indirectly freely. An agent acts indirectly freely only if her practical necessity counterfactually depends on past directly free actions, and those directly free actions bear an adequate degree of similarity to the practical necessity. The second two considerations set parameters on the degree of praiseworthiness and blameworthiness an agent earns for her indirectly free action. The agent need not have generated the practical necessity ex nihilo by directly free actions to be fully praiseworthy or blameworthy for the indirectly free action that issues from it, and the agent’s degree of praiseworthiness and blameworthiness for an indirectly free action is determined in part by (i) the quantity and quality of her directly free actions and omissions that make a difference to acquiring the practical necessity and (ii) the quantity and quality of her opportunities to avoid acquiring the practical necessity. In the next section, I examine the required epistemic connection between directly free actions and indirectly free actions.

3 The Epistemic History Condition

In general, Indirect Libertarians fail to consider the kind of epistemic connection that must obtain between an agent’s directly free actions, practical necessity, and indirectly free action. Kane (1996a, b), for example, entirely omits discussion of it. The terrain here, however, is better covered than in the previous section due especially to the work of Fischer and Neal Tognazzini (2009), Seth Shabo (2015), and Timpe (2011).

Let us begin by considering Manuel Vargas’s (2005, p. 274) Epistemic Condition on moral responsibility, which I reword to focus exclusively on indirectly free actions.

VEC: An agent acts indirectly freely only if that outcome was reasonably foreseeable at an earlier time at which she performed the directly free actions that eventually contributed to the relevant practical necessity.

Vargas (2005, p. 271) argues that VEC is satisfied less often than we suppose by generalizing from several examples (cf. McKenna 2008, p. 33).¹⁴ I focus on his case Jeff the Jerk: Jeff is a middle-manager who is commissioned to lay off his employees, but does so in a gratuitously rude and insensitive manner. Suppose that Jeff’s jerkiness practically necessitates firing his employees in that way, because he has a practical necessity to behave like a jerk in circumstances in which he can get away with it unscathed. To ascertain whether Jeff’s practically necessitated action was reasonably foreseeable at the time of the relevant directly free actions, we must consult his history. Fifteen-year-old Jeff was much less successful than he liked in getting the attention of his attractive classmates, and he noticed that people who we might describe as jerks were having more success. So, he decided to inculcate jerk-like attitudes and behaviors in himself to get their attention. “With surprisingly little effort, he succeeded. In fact he more than succeeded—it didn’t even take the whole academic year for him to go from being jerk-like to being a full-on jerk” (Vargas 2005, p. 276). But when he decided to become a jerk and behave in jerky ways, he never conceived that it would result in the despicable firing of his employees. So, when Jeff performed directly free actions, it was not reasonably foreseeable that he would be practically necessitated to fire those employees in a rude and insensitive way. Since VEC implies that no one is morally responsible for an outcome that they could not reasonably be expected to foresee, Jeff is not morally accountable for his practically necessitated action, and so does not act indirectly freely.

How might Indirect Libertarians respond to this challenge that there are very few indirectly free actions on account of our limited foresight? One prominent response advocated by Fischer and Tognazzini (2009) is to explain how Jeff’s practically necessitated action was reasonably foreseeable.¹⁵ In their estimation, Vargas’s case seems compelling only due to the specificity of Jeff’s future practically necessitated action—namely, Jeff could not reasonably have been expected to foresee the consequence of that despicable firing of those future employees at that company. But VEC does not specify the level of descriptive generality. Consider, for example, three levels of description offered by Fischer and Tognazzini (2009, p. 537):

(Level 1) “Jeff fires those employees who work for that company on that precise day in that precise manner.”

(Level 2) “Jeff fires some of his employees at some company or other at some point in the future in a despicable manner as a result of his jerky character.”

(Level 3) “Jeff treats some people poorly at some point in the future as a result of his jerky character.”

Fischer and Tognazzini (2009, pp. 537–538) plausibly contend that although Jeff’s action at description Level 1 and Level 2 is not reasonably foreseeable, his action is reasonably foreseeable at description Level 3. Furthermore, a plausible version of VEC requires reasonable foreseeability at only description Level 3, because what matters is not the exact details of the foreseeable outcome but rather its general morally significant features. When we construe the outcome in VEC at Level 3, our practically necessitated actions often satisfy that necessary condition, and so are often candidates for being indirectly free actions.

Timpe (2011) supports Fischer and Tognazzini’s response by providing an “improved epistemic condition” (IEC) that further explains their judgments. I explicate only the part of Timpe’s (2011, pp. 19–20) IEC that concerns the nature of indirectly free actions and seems at least minimally plausible for libertarians.¹⁶ I also reword IEC to use the terminology in this paper and rename it as Timpe’s Epistemic Condition.

TEC: An agent acts indirectly freely only if at the earlier time when she performed directly free actions, she had at least reasonable and merely dispositional beliefs that performing those directly free actions might lead to performing an action of a similar kind in the future (or was culpable for failing to have those beliefs).

Five features of TEC merit explication. First, the modifier ‘reasonable’ refers to being sufficiently justified by evidence, but TEC leaves open which account of justification is best to use. Second, the ‘might lead to’ relation should be understood as stronger than ‘possibly leads to’ and as weaker than ‘would lead to’ and ‘must lead to,’ but the precise probability that is required for moral responsibility transfer is determined contextually (cf. Fischer and Ravizza 1998, p. 50n21). An important part of that contextual determination is how serious the outcome is (Timpe 2011, p. 18). For example, if the practically necessitated action would be very harmful, only a very small probability would be required to ground a moral responsibility transfer.

Third, it is unnecessary that the agent have knowledge or an occurrent belief about the relationship between her directly free action(s) and the practically necessitated action, because only a dispositional belief is required to preserve moral responsibility transfer (cf. Peels 2017, p. 177). It is, however, worth pointing out that whether the belief is occurrent or merely dispositional may affect her degree of moral responsibility. An agent would be more blameworthy for performing a morally wrong action with the occurrent belief that the action is wrong than she would be with a merely dispositional belief, all other things being equal.

Fourth, Timpe (2011, pp. 22–23) does not commit to an account of culpable ignorance. He points toward a non-tracing account of culpable ignorance, because he asserts that a problematic circularity would follow if TEC, a tracing condition, itself appeals to tracing. But I do not see what motivates that circularity claim. Consider the way in which a generic tracing view of culpable ignorance would apply. An indirectly free action traces back to the agent’s past culpably ignorant directly free actions that in turn trace back further to a time at which she should have improved her epistemic condition but fails to do so in a way suitable to incur blameworthiness. At this level of generality, we should be able to see the problematic circularity, and there is none to be found. For this reason, I suggest that both tracing and non-tracing accounts of culpable ignorance are consistent with TEC, but leave it open which account Indirect Libertarians should adopt.

Fifth, TEC does not address questions about belief content. First, how general should the description of the future action be? It appears that Timpe has description Level 3 in mind, and this seems plausible. Second, must Jeff merely believe that his performing jerky actions will make him more likely to become a jerk and perform jerky actions in the future, or must Jeff also believe that if he performs enough jerky actions that his future jerky actions might be practically necessitated? Given Level 3 generality, it seems plausible that the required belief need be only that his jerky actions might lead to his becoming a jerk and performing future jerky actions. Third, must Jeff actually believe the proposition expressed in Level 3 description, or is it enough that he would immediately draw the conclusion if he were asked a question about it given his other beliefs? To satisfy the non-culpable ignorance clause of TEC, he needs to believe it. After all, if Jeff does not even merely dispositionally believe the Level 3 generality proposition, he is ignorant. Even so, he could still satisfy the culpable ignorance clause of TEC.

With this explication of TEC in hand, let us apply it to Jeff the Jerk. It is plausible to think that Jeff had occurrent beliefs that becoming a jerk might lead to his treating people badly in the future due to his jerkiness, but, even if he did not, he very plausibly had at least merely dispositional and reasonable beliefs about it. And because intentionally acquiring bad character is a very serious affair due to its high likelihood of producing future harm, he need not even think it was very probable that such current jerky actions would lead to performing more in the future. Therefore, it is very plausible to suppose that Jeff satisfies TEC, and that he is morally responsible for the jerky way in which he fires his employees, at least with respect to the epistemic condition.

Jeff’s case is a case of a person who sets out to become a certain way, but, in the more common kind of case, people do not aim to become the kind of people they do become. Let us consider whether TEC can handle this latter kind of case.

Shabo’s (2015, pp. 996–999) Greg the Greedy is tasked to recommend budget cuts. His superiors expect that he will recommend deep cuts that will result in the termination of employees, and, in return for those cuts, it is insinuated that Greg would receive a fiscal bonus. Greg’s options are to (i) disregard the expectation of his superiors and fail to get the bonus, (ii) recommend minimal layoffs and get the bonus, or (iii) exceed their layoff expectations and hope for a slightly larger bonus. Greg’s greed and inability to identify with the needs and interests of his social inferiors practically necessitate his choice to exceed their layoff expectations.

Does Greg act indirectly freely? To answer this question, we must consult his history. Greg moved to a new school in seventh grade (2015, p. 996–997). He was naturally insecure, and “instinctively emulated” the cool kids, who acted like jerks. He never decided to do this. After receiving “positive reinforcement,” he continued to “gradually and unreflectively adjust his demeanor to theirs.” He knew that his disdain and ridicule made others angry, but this did not trouble him. As he became more confident later in high school and college, he relied on this jerky behavior less, but he continued to divide up the world into “big shots” and “nobodies.” On his first job out of college, he was impressed by his aloof and greedy superiors, and he consciously emulated them. He became focused on acquiring wealth.

Was Greg’s firing more employees than expected a reasonably foreseeable outcome of those earlier directly free actions? Shabo (2015, pp. 998–999) argues that the answer is plausibly ‘no’ by canvassing candidate actions and omissions: (i) Greg’s decision in high school to act obnoxiously toward a nobody, (ii) Greg’s omission to reevaluate his behavior in light of being blamed by nobodies, (iii) Greg’s omission to reevaluate his behavior in light of perceiving in himself a clear pattern of obnoxious behavior, (iv) Greg’s latest belittlement that culminates in a tendency to be unmoved by the needs and interests of perceived social inferiors, (v) Greg’s decision in college to stop acting obnoxiously for social utility but ignoring the ethical reasons to do so, (vi) Greg’s inability to be moved by the interests of nobodies becomes settled, and (vii) Greg’s conscious decision to emulate his superiors at work. Shabo (2015, p. 999) concludes that “there seems to be no single episode such that Greg could have been expected to recognize that he was risking significant hardship or deep-seated indifference to others … by his behavior then.” Shabo (2015, p. 997), then, holds that Greg’s practically necessitated action was not reasonably foreseeable, which presents a clear challenge to the Fischer and Tognazzini strategy.¹⁷

What I want to highlight is that TEC is well-suited to handle this kind of case. Recall that TEC does not require that there is a single episode such that Greg occurrently believed or knew that his actions risked developing indifference to nobodies. Rather, TEC requires that Greg had at least merely dispositional and reasonable beliefs that performing the directly free actions might lead to being unmoved by the needs and interests of perceived social inferiors and might lead to imposing needless hardship on nobodies.

It seems plausible that Greg had those beliefs at, for example, (iii)—that is, his omission to reevaluate his behavior in light of perceiving in himself a clear pattern of obnoxious behavior. His perceiving this clear pattern was occurrent, and it is very plausible that this occurrent recognition generated at least the merely dispositional belief that the pattern would continue into the future unless he decided to change his ways, which is something he omitted to do. Thus, at the very least he had the relevant merely dispositional belief when he performed jerky actions and omissions after this time. The probability assessment about performing future jerky actions could even have been low due to the significant harm they project.

If, however, Greg did not have even those beliefs (and he was not culpably ignorant for failing to have them), Greg’s directly free actions that developed the tendency to be unmoved by the needs and interests of his perceived social inferiors would not contribute to his degree of responsibility for his indirectly free action. Nevertheless, Greg would still be blameworthy to some lesser degree for his indirectly free action. After all, the practical necessity that issues in the action counterfactually depends also on his greedy actions at (vii), and he plausibly had at least a merely dispositional belief that his conscious choice to emulate his greedy superiors at work might lead to treating people badly in the interest of personal gain in the future, especially if he observed his superiors doing this. In this case, then, Greg’s indirectly free action would trace back to his emulating his bosses at work, and so TEC can preserve the moral responsibility transfer from previous directly free actions to Greg’s indirectly free action. Of course, Greg would be less blameworthy for his indirectly free action than he would have been if he had the relevant beliefs during (iii) as well. I conclude that TEC is fit to handle a broad range of cases, and is a plausible position for Indirect Libertarians to take. In the next section, I explore the nature of indirectly free actions further by considering the nature of derivative moral responsibility.

4 Derivative Responsibility and Resultant Moral Luck

No Indirect Libertarian, as far as I am aware, has taken up the question about how to understand the nature of derivative moral responsibility of indirectly free actions with respect to whether resultant moral luck exists.¹⁸Resultant moral luck occurs when the consequence of an agent’s directly free action or omission is at least partially beyond her control, and that consequence positively affects how much praise or blame she deserves. This kind of moral luck is relevant to indirectly free actions precisely because it is at least partially a matter of resultant luck whether a directly free action or omission changes her dispositions in a way that eventually gives rise to a practical necessity.

Return to the Luther case. On the supposition that the resultant luck in the Luther case is resultant moral luck, Luther is more praiseworthy overall in virtue of his practical necessity and his indirectly free action than the degree to which he is praiseworthy for earlier directly free actions relevant to developing the practical necessity. In particular, Luther is more praiseworthy overall for acting indirectly freely than he would have been if he had the practical necessity but had not performed the indirectly free action. This moral evaluation of Luther’s action seems to me to be exactly right; thus, my own view is that the most plausible account of indirectly free actions is committed to the existence of resultant moral luck.¹⁹

It is important to recognize, however, that accepting extant resultant moral luck is not required by the nature of indirectly free actions. The most plausible way to deny resultant moral luck and yet affirm indirectly free actions is to make a distinction between the “degree” and “scope” of moral responsibility (Zimmerman 2002, pp. 560–561; cf. Peels 2017, p. 224). The degree of moral responsibility is about how much praise and blame the agent deserves, and this kind of moral responsibility cannot be affected by luck. But the scope of moral responsibility concerns merely the events for which an agent is morally responsible; this kind of moral responsibility may be affected by luck. Apply this distinction to the Luther case. Since it is at least partially a matter of resultant luck whether Luther’s previous directly free actions would eventually form a practical necessity (after all, he might have been stuck by lightning after performing them and died having never acquired the practical necessity), he is responsible for his practical necessity and his indirectly free action in the scope sense only (cf. Simkulet 2015). Thus, Luther’s indirectly free action does not contribute to his overall praiseworthiness; his earlier directly free actions exhaust how much praise he deserves. He is, of course, morally responsible for his practically necessitated action, and so the denial of resultant moral luck is compatible with acting indirectly freely.

Even though endorsing indirectly free actions does not require accepting resultant moral luck, I do think that their conjunction is plausible. In particular, I contend that accepting resultant moral luck is theoretically motivated for Indirect Libertarians in part because they tend to accept the existence of circumstantial and constitutive moral luck.

Circumstantial moral luck occurs when it is outside of the agent’s control whether she faces a morally significant challenge or opportunity, and it positively affects her praiseworthiness or blameworthiness. Constitutive moral luck occurs when an agent’s dispositions or capacities are not voluntarily acquired, and they positively affect how much praise or blame she deserves for a trait or an action. One way to see why Indirect Libertarians tend to accept these kinds of moral luck is to show they cannot or would not want to deny them. Indirect Libertarians cannot deny these kinds of moral luck by factoring out all luck from moral responsibility as Neil Levy (2011), Galen Strawson (1994), and Bruce Waller (2011) suggest that they do, because the result would be that no one would be positively praiseworthy or blameworthy for anything, which is incompatible with the libertarian commitment that we at least sometimes act freely and responsibly. Indirect Libertarians would not want to deny these kinds of moral luck by allowing people to be morally responsible in virtue of what they would directly freely do in non-actual circumstances and with non-actual constitutive properties as Michael Zimmerman (2002) suggests that they do, because at least most Indirect Libertarians deny that such counterfactuals can be true (O’Connor 1992; van Inwagen 1997; Zimmerman 2011; cf. Hartman 2017, pp. 71–80).²⁰ Thus, Indirect Libertarians tend to accept extant circumstantial and constitutive moral luck, because they do not endorse the main ways to deny these kinds of moral luck.

Accepting circumstantial moral luck, however, provides good analogical evidence for endorsing resultant moral luck (Hartman 2017, pp. 105–111).²¹ I sketch the argument concretely in terms of three examples. No Start, our first character, gets into her car to drive home in an angry mood, but her car does not start. As a result, she has to call a cab to take her home. Merely Reckless, our second character, gets into her car to drive home in the same angry mood. But because her car starts, she drives homeward in anger, and she luckily makes it there without incident. The comparative case of No Start and Merely Reckless is a standard example of circumstantial luck. Given that the Indirect Libertarian accepts circumstantial moral luck, Merely Reckless is more blameworthy than No Start in a way that is partially determined by luck, because luck in opportunity allows only Merely Reckless to sustain her intention into overt action. Killer, our third character, also gets into her car to drive home in the same angry mood, but her reckless driving ends up killing a pedestrian. Suppose that Merely Reckless would have killed a pedestrian in the same way if there had been a pedestrian on the road for her to kill. The case of Merely Reckless and Killer is a standard example of resultant luck.

Notice three ways in which the circumstantial luck case of No Start and Merely Reckless, on the one hand, is analogous to the resultant luck case of Merely Reckless and Killer, on the other. First, luck is the salient difference between the agents in both case pairs—whether the car starts and whether there is a pedestrian on the road. Second, the event for which the agent in each case pair might be more blameworthy depends at least in part on that person’s own agency. Merely Reckless’s driving depends on her agency, and Killer’s causing the death of a pedestrian depends on her agency. Third, the agent who might be more blameworthy in each case pair had at least a fair opportunity to avoid being additionally blameworthy in that way. In the case of No Start and Merely Reckless, Merely Reckless had a fair opportunity to avoid driving recklessly, because she could have directly freely chosen to avoid driving while she was angry. In the case of Merely Reckless and Killer, Killer had a fair opportunity to avoid killing a pedestrian, because she also could have directly freely chosen to avoid driving while she was angry; and killing a pedestrian is a foreseeable consequence of reckless driving. Plausibly, similarities of these kinds are what led David Enoch and Ehud Guttel (2010, p. 376) to assert that “The problem of moral luck seems to be the very same problem whether it is luck in consequences or in circumstances, and is typically so treated in the literature.”

Thus, given the differential desert that Merely Reckless is more blameworthy than No Start in a way partially determined by luck and these three similarities between the case pairs, we have good analogical evidence that Killer is more blameworthy than Merely Reckless in a way that is partially determined by luck. The Indirect Libertarian who accepts circumstantial and constitutive moral luck has a built-in reason to affirm extant resultant moral luck.

In summary, although Indirect Libertarians can consistently deny resultant moral luck, I suggest that the most intuitive evaluation of Luther’s indirectly free action involves accepting resultant moral luck, and that Indirect Libertarians have an independent reason to affirm resultant moral luck based on the analogical argument. The Indirect Libertarian may thus plausibly regard Luther as more praiseworthy overall for performing the indirectly free action than he is merely for performing the directly free actions that formed him into that kind of person.

5 Conclusion

I have compiled various partial accounts of indirectly free actions into a single account, refined some of those partial accounts to make them more plausible, and filled in gaps left by the compilation of these partial accounts. In particular, I added two metaphysical historical conditions, made some clarifications to TEC and applied it to a new kind of case, and argued that an attractive view of indirectly free actions involves accepting resultant moral luck.

Here is the account. An agent acts indirectly freely if and only if she has performed past directly free actions such that (i) she had at least reasonable dispositional beliefs at the time of the directly free actions that those actions might lead to performing an action of a similar kind in the future (or was culpable for failing to have those beliefs), (ii) her practical necessity counterfactually depends on those directly free actions, and (iii) those directly free actions bear an adequate similarity to the practical necessity. Furthermore, she need not have generated the whole practical necessity by directly free actions to be fully praiseworthy or blameworthy for the indirectly free action, and how praiseworthy or blameworthy she is for the indirectly free action is determined in part by (i) the quantity and quality of her directly free actions and omissions that make a difference to acquiring the practical necessity, (ii) the quantity and quality of her opportunities to avoid acquiring the practical necessity, and (iii) whether the reasonable belief about the future outcome of the directly free actions is occurrent or merely dispositional.²²