Omissions, Moral Luck, and Minding the (Epistemic) Gap

Abstract

This paper warns of two threats to moral responsibility that arise when accounting for omissions, given some plausible assumptions about how abilities are related to responsibility. The first problem threatens the legitimacy of our being responsible by expanding the preexisting tension that luck famously raises for moral responsibility. The second threat to moral responsibility challenges the legitimacy of our practices of holding responsible. Holding others responsible for their omissions requires us to bridge an epistemic gap that does not arise when holding others responsible for their actions—one that we might often fail to cross.

1. Introduction

There is a growing literature on omissions and our failures to act more broadly, and an increasing number of philosophers are now recognizing the impact that omissions and failures to act have on questions of moral responsibility. This paper serves as a lighthouse, warning of two threats to moral responsibility that arise when accounting for responsibility for our omissions. The first threatens the legitimacy of our being responsible, in that it expands the preexisting tension that luck famously raises for moral responsibility originally raised by Nagel (1979). In particular, this paper argues that, given some plausible assumptions about responsibility for our omissions, constitutive and circumstantial luck are rampant for our omissions and can arise in a different kind of way from the way they arise for our actions. Hence, if we want to understand fully what constitutive and circumstantial luck are and the threats they raise for moral responsibility, we must consider the impact of our omissions. The second threat to moral responsibility that arises from our omissions challenges the legitimacy of our practices of holding responsible. Holding others morally responsible and accountable for their failures to act requires us to bridge an epistemic gap that does not arise when holding others responsible for their actions—and there is good reason to think we fail to cross this gap a nontrivial amount of the time.

This paper proceeds by first highlighting the importance of abilities for moral responsibility and by then noting a plausible asymmetry between actions and omissions with respect to agents’ abilities and moral responsibility. This leads to the identification of what I call asymmetric-abilities cases. Such cases only arise for failures to act, and they raise the two previously discussed challenges for theories of moral responsibility. Since these two challenges only arise for our non-doings and not for our actions, they also suggest that developing a good theory of moral responsibility for omissions goes far beyond simply making minor tweaks to the extant theories of moral responsibility for actions.

2. Abilities, actions, and omissions

Let’s begin by considering the importance of abilities for moral responsibility. This is not the place for a complete defense of ability-based theories of moral responsibility as superior to all possible rivals. Instead, I hope to show why abilities form a plausible foundation for theorizing about responsibility. Moral responsibility is commonly thought to have two central conditions: the control condition, which requires that we have sufficient control over what we do, and the epistemic condition, which requires that we are (or should be) sufficiently aware of what we do. ¹ Note that abilities are important for both conditions: what we have control over crucially depends on what we are able to do, and being relevantly aware often requires awareness of our abilities. Consider also that basing moral responsibility primarily on abilities has become increasingly popular in the moral responsibility literature. ² Furthermore, many other theories of responsibility that are not explicitly ability-based still rely on or heavily involve abilities. ³

Abilities, of course, come in many varieties and flavors, so it’s worthwhile to say a bit more about which abilities are relevant to moral responsibility for actions and omissions. Drawing from this literature on abilities and moral responsibility, the relevant abilities are specific abilities, which encompass both general abilities and opportunities. Roughly, general abilities are agents’ internal capacities based on skills, know-how, coordination, and so on, whereas opportunities roughly are agents’ external circumstances that permit them to exercise their relevant abilities. ⁴

Now, let’s examine what the relevant specific abilities for responsibility are—what must agents specifically be able to do? Here, I contend there is an asymmetry between actions and omissions with respect to the ability to do otherwise:

Asymmetry: Direct ⁵ moral responsibility for actions requires possessing the specific ability to perform the relevant action that was actually performed but does not require the specific ability to do otherwise. In contrast, direct moral responsibility for omissions requires possessing the specific ability to perform the omitted act and thereby does require the ability to do otherwise.

There also is a relevant corollary of Asymmetry that is highly plausible on its own:

Corollary: An agent can’t be directly responsible for an omission if she wasn’t able to perform the omitted act.

Different versions of this asymmetry thesis have been contested and defended in the moral responsibility literature, so rather than try to give a full defense of it, I will try to show why it is a plausible assumption.

To justify Asymmetry, let’s first consider a common starting point for thinking about the abilities relevant to responsibility: the Principle of Alternative Possibilities. The Principle of Alternative Possibilities has been defended by various philosophers within the free will literature, and it holds that we are only responsible for an action or omission if we were specifically able to do otherwise. ⁶ The Principle has historically faced controversy due to apparent counterexamples because of Frankfurt-style cases in the vein of those raised in Frankfurt (1969). However, I contend that such counterexamples only arise for actions, not omissions. Frankfurt-style action cases involve agents who are intuitively responsible for their actions despite not being able to do otherwise. For instance, Assassin decides to shoot Victim, but if Assassin had started to decide not to shoot Victim, a neuroscientist monitoring Assassin’s brain would have made Assassin decide to shoot Victim. As it turns out, Assassin decides on her own to shoot Victim, and the neuroscientist does not intervene. Although some dispute Frankfurt-style action cases, many accept them and would hold that Assassin is morally responsible for deciding to shoot Victim even though Assassin was unable to decide otherwise. ⁷ This is not the place for a full discussion of Frankfurt-style action cases, but they arguably show that responsibility for actions does not require the ability to do otherwise, which is the first half of Asymmetry.

Despite facing these challenges for actions, the Principle of Alternative Possibilities arguably gets omission cases right: responsibility for omissions arguably does require the ability to perform the omitted act (and responsibility for omissions thereby requires the ability to do otherwise). To see why, first consider the standard analysis of Fischer and Ravizza’s (1998) classic Sharks case:

Sharks: John sees a child drowning in the ocean and decides not to jump into the water to save him. As it turns out, there are sharks in the water that would have eaten John if he had jumped into the water.

According to the standard, widely held analysis of Sharks, John is not responsible for an omission, not saving the child, because he couldn’t have done otherwise (i.e., he lacked the specific ability to save the child). Note that holding that John is responsible for his omission to save the child also entails holding that Corollary is false: one would have to hold that John is responsible for omitting to do something that he can’t do, which seems implausible. Now consider that if we alter the case so that the sharks are not there to stop John, he then seems responsible for omitting to save the child, which suggests that the difference in his specific ability to perform the omitted act is central to his responsibility for his omission.

As further support of the idea that moral responsibility for omissions requires the ability to do otherwise, note that the reasoning used in support of Frankfurt-style action cases does not straightforwardly extend to Frankfurt-style omission cases. In a Frankfurt-style omission case, an agent is supposedly responsible for omitting to perform a certain action despite not being specifically able to perform it; attempts to give such cases have been met with resistance. ⁸ One challenge is that some apparent Frankfurt-style omission cases conflate actions with omissions. Suppose Sam decides not to rescue a drowning swimmer when he was unable to decide otherwise. It might thereby seem like Sam is responsible for an omission even though he lacked the ability to do otherwise. However, deciding not to do something is an action, not an omission. Hence, this is actually a Frankfurt-style action case, not a Frankfurt-style omission case (and it just resembles the earlier action case of Assassin deciding to shoot Victim). To see what a Frankfurt-style omission case would look like, consider a variation where Sam fails to make any decision at all. In this version, Sam fails to decide to save the swimmer—which is an omission—and he was unable to make that decision. The problem is that it is no longer clear that Sam is blameworthy for omitting to decide to rescue the swimmer; it violates Corollary and strikes me as quite unintuitive to hold that Sam is responsible for failing to decide something when he was unable to make that decision. In contrast, when we blame Sam in the action version of the case for deciding to do something when he was unable to decide otherwise, note that Sam still is able to make that decision (and he, in fact, exercises that ability).

Thus, there are cases where an agents’ responsibility for an omission arguably depends on whether he was able to perform the omitted act, and the kinds of reasoning used to defend Frankfurt-style action cases do not clearly extend to Frankfurt-style omission cases. This makes it plausible to hold that responsibility for omissions requires the ability to perform the omitted act (and therefore the ability to do otherwise), which is the second half of Asymmetry.

In short, I hope to have shown why abilities—in particular, specific abilities—are central to issues of responsibility for actions and omissions, and although I have not provided a full defense of Asymmetry, I hope to have shown why it is at least a plausible assumption. ⁹ I will proceed under the assumption that moral responsibility for omissions requires the specific ability to do otherwise, but responsibility for actions does not. Let’s now use this to identify what I call asymmetric-abilities cases.

3. Asymmetric-abilities cases

Consider the following pair of cases:

Epilepsy: Ephrem sees a child drowning in the ocean, but he decides to sunbathe on the beach rather than rescue the child. Unbeknownst to him and although he has never had a seizure before, Ephrem has epilepsy and would have had an epileptic seizure if he had tried, or even started to try, to rescue the child.

No Epilepsy: Nina sees a (different) child drowning in the ocean, but she decides to sunbathe on the beach rather than rescue the child. Nina does not have epilepsy, and she was able to rescue the child.

Ephrem and Nina each fail to save a drowning child. However, their abilities to save the relevant child differ: only Nina possessed the specific ability to save the child. ¹⁰ It also seems quite plausible that Ephrem is not eligible to be morally responsible for his failure to save the child, and we can use the reasoning from the previous section to explain why. Given Asymmetry, responsibility for an omission requires the ability to perform the omitted act, and Ephrem lacked the specific ability to rescue the child, so he is not responsible for omitting to rescue the child. Relatedly, it would violate Corollary to hold that Ephrem is blameworthy for failing to do something that he was unable to do. In contrast, Nina plausibly is responsible for her omission to save the child (barring other excusing or exempting considerations, such as epistemic considerations if Nina was unaware that the child was drowning) because she was specifically able to rescue the child, and there is no corresponding violation of Corollary. Hence, Nina and Ephrem’s responsibility for their respective failures to rescue the drowning child tracks whether they possessed the specific ability to rescue the child. More broadly, two agents fail to perform the same action of saving a drowning child, but only one is able to do so. And, on the basis of such a difference in their abilities, only the agent who is able to perform the omitted act is morally responsible for her omission. Epilepsy and No Epilepsy thereby form a pair of what I call asymmetric-abilities cases—cases with a difference in agents’ responsibility for the same behavior that arises because of a difference in their respective abilities.

Before exploring asymmetric-abilities cases further, let’s clarify a few points here. First, there are different projects and accounts of omissions, so there might appear to be disagreement about whether an agent’s failure to perform an act that she was unable to perform qualifies as an omission. ¹¹ I remain neutral on such debates; all that matters for the purpose of this paper is whether the relevant agents are responsible for their failures to act, not whether those failures to act also qualify as omissions. Hence, every mention of “omission” can be replaced with “failure to act,” depending on the reader’s preferences.

Second, it might seem tempting to try to simplify the assessments of responsibility here by saying that Ephrem and Nina are both equally blameworthy in virtue of being equally bad people—they each wouldn’t have saved the child regardless of their respective abilities—so we don’t need to worry about who is responsible for which particular behavior. However, although the moral evaluation of persons as a whole is undoubtedly an important part of our moral responsibility practices, so too is the moral evaluation of agents’ particular behaviors. And evaluating someone’s moral character is importantly distinct from evaluating her particular responsibility for her actions and omissions. This is why many philosophers hold that agents can be responsible for their particular behaviors.

Relatedly, there are many competing conceptions of responsibility, and it is important to clarify the type of responsibility at stake here. For instance, consider Watson’s (1996) distinction between attributability responsibility and accountability responsibility. Roughly, attributability responsibility involves properly attributing bad behavior to an agent that reflects bad moral character and “deep self.” Accountability responsibility involves properly holding an agent accountable for that particular behavior in a way that warrants the full range of the reactive attitudes like blame and resentment because she freely and knowingly performed it. ¹² It might seem tempting to blame Ephrem for his omission to save the child by thinking of responsibility as attributability on the grounds that his omission to save the child is properly attributable to him and reflects bad moral character on his part.

However, this paper is not about responsibility as attributability because the requisite control and responsibility at stake for issues like moral luck and our practices involving the reactive attitudes go beyond what is required for attributability. To illustrate, it is possible for someone’s bad actions to evince poor moral character and be correctly attributable to her and accurately reflect a bad deep self, even though she only has that character and deep self as a result of persistent childhood abuse. Following Wolf (1990), Watson (1996), Levy (2005), and others (see, e.g., Talbert 2019), more is required to justify the full range of the reactive attitudes than making these proper attributions of behavior. In particular, a deeper level of control is required: a level of control that would justify the reactive attitudes, warrant considerations for punishment by others, and make it fair to hold an agent to account for her behavior. This is the kind of control involved in accountability responsibility. Hence, this paper focuses on responsibility as accountability in the vein of Watson (1996), Levy (2005), Shoemaker (2011), McKenna (2012), and many others.

Thus, while we can assess agents’ moral character and their attributability responsibility, it is still important to assess agents’ accountability responsibility for particular behaviors—for their actions and omissions. Even if one judges that Nina and Ephrem are equally bad people or that their behaviors reveal equally bad character, there is still an important philosophical question about whether they are equally accountability-responsible for their respective omissions. I have argued that they are not, and as we will see in the coming sections that this has important results for the problem of moral luck and for our practices of holding agents responsible for their omissions.

As a third clarification, one might argue that Nina and Ephrem are equally accountability responsible for their respective omissions to save the drowning child (even though Ephrem was unable to save the child) because they are each able to decide not to save the child, and they each exercise that ability, so their responsibility for their omission traces back to their decision. ¹³ However, the responsibility at stake when thinking about the ability to do otherwise is direct responsibility, not indirect responsibility that can trace back to earlier behaviors. To illustrate, Frankfurt-style action cases and the part of Asymmetry that concerns actions would be uninteresting and uncontroversial if the relevant responsibility were indirect because, for instance, drunk-driving cases already straightforwardly show that a drunk driver can be indirectly responsible for causing an accident, even though she was unable to avoid doing so given her drunken state. The interesting contention of Frankfurt-style action cases and the first part of Asymmetry is that an agent can be directly responsible for an action despite lacking the ability to do otherwise. Similarly, the interesting contention of the second part of Asymmetry and Corollary is that an agent cannot be directly responsible for omitting to do something that she was unable to do. The relevant issue for Epilepsy and No Epilepsy is therefore whether Nina and Ephrem are directly responsible for their omission to save the child. Additionally, we can clarify Epilepsy and No Epilepsy to make the relevant omitted action a basic mental action (a decision) rather than a more complicated bodily action (rescuing someone). Ephrem would have had a seizure if he had tried or even started to try to rescue the child, and we can clarify that starting to try includes deciding to rescue the child. Hence, Ephrem would have had a seizure if he had decided or started to decide to rescue the child, so he was unable to decide to rescue the child. In contrast, Nina was able to decide to rescue the child. Hence, they both omit to decide to rescue the child, but only Nina is responsible for her omission to decide, whereas Ephrem is not. We therefore still have a pair of cases with differences in agents’ responsibility for the same behavior that results from differences in their respective abilities, so we still have a pair of asymmetric-abilities cases. For clarity, what I say going forward can be read as applying to “whole” omissions but starts, more basically, with mental omissions.

With these clarifications in place, we can now turn to the features of asymmetric-abilities cases. According to Asymmetry, there is no ability to do otherwise requirement for moral responsibility for actions, but there is such a requirement for omissions. Hence, to determine whether an agent is responsible for an omission, we have to inquire into her unexercised abilities. However, we do not need to do this for actions (we need only look into her exercised abilities). Asymmetric-abilities cases arise when agents have different unexercised abilities. Therefore, the first key feature of asymmetric-abilities cases is that they only arise for omissions, not actions.

Next, there are different kinds of asymmetric-abilities cases because there are multiple ways to lack the specific ability to do something: lacking the general ability to do it, lacking the opportunity to do it, or both. Some asymmetric-abilities cases arise when two agents have different internal abilities and the same external opportunities, like Epilepsy and No Epilepsy. There also are versions of asymmetric-abilities cases in which the agents have the same internal abilities but differ in their external opportunities:

Child: Heathcliff sees a child drowning in the ocean, but he decides to sunbathe on the beach rather than rescue the child.

No Child: At a beach many miles away, Catherine is also sunbathing. She mistakenly thinks that there is a child drowning on the beach near her, and she decides not to rescue that child. However, in fact, there are no nearby drowning children, and Catherine lacked the opportunity to rescue a drowning child.

Heathcliff is blameworthy—at least as far as abilities and opportunities go—for his failure to save the child in a way that Catherine is not because Heathcliff had the opportunity to save the child, whereas Catherine did not. There also can be cases that involve a difference in responsibility that results from a combination of differences in abilities and opportunities. ¹⁴ For clarity’s sake, I will group cases with asymmetric abilities, asymmetric opportunities, or a combination of both under the label “asymmetric-abilities cases.” As we will see in the next two sections, these cases raise two dangers to theories of moral responsibility.

4. Asymmetric abilities and moral luck

4.a Types of moral luck

The first reason that asymmetric-abilities cases matter for moral responsibility is that they are important for moral luck: they show that constitutive and circumstantial luck are rampant for omissions in a way that they are not for actions in a way that has not yet been fully appreciated. To see this problem, let’s clarify what the problem of moral luck is in general and what constitutive and circumstantial luck are in particular. In general, the threat of moral luck arises from two competing sets of intuitions about moral responsibility. First, in what is commonly referred to as the Control Principle, there is the deeply intuitive thought that someone should only be morally responsible for her behavior to the extent that her behavior depended on factors that were under her control. ¹⁵ Second, there are cases with intuitive differences in responsibility that seem to derive from features that are outside of an agent’s control: two people might drive home drunk and swerve onto a sidewalk, but only one driver hits a pedestrian. The driver who actually hits a pedestrian is often held to be more morally responsible than the one who does not, but this difference is due to luck in whether a pedestrian is present on the relevant stretch of sidewalk, which is obviously outside either driver’s control. The problem of moral luck is, therefore, how to resolve the tension between these deeply held but conflicting sets of intuitions.

Following Nagel’s (1979) traditional categorization, there are four subspecies of moral luck that differ in the relevant source of the luck: resultant luck, causal luck, circumstantial luck, and constitutive luck. Resultant luck is the luck in the outcomes of one’s actions, and examples include the aforementioned set of drunk drivers: both drivers perform the same risky action with similar intentions and lack of regard for the safety of others, but the outcomes of those actions are significantly different. ¹⁶ Causal luck is the luck in the distant causes of one’s behaviors. ¹⁷ If determinism is true, then factors in the distant past that are outside of our control ultimately cause our behaviors. ¹⁸ Of special interest to us will be the remaining two forms of moral luck, circumstantial luck and constitutive luck, so let’s discuss them in greater detail.

Circumstantial luck is the luck in one’s circumstances. In particular, Nagel (1979) highlights how circumstantial luck involves the luck in which moral tests one faces. An individual who, in fact, grew up in the United States and lived an ordinary life might have willingly become an officer in a concentration camp if he had grown up in Nazi Germany, and an individual who grew up to become an officer in a concentration camp might have lived an ordinary life had he been born in the United States. However, if we hold that these two people are morally responsible to different extents for their behaviors, then we have a difference in moral responsibility that derives from factors of the agents’ circumstances that are outside of their control.

Constitutive luck is the luck in one’s mental and physical traits and dispositions. Someone might seem blameworthy for being indifferent to another’s suffering, but her indifference might result from the nonvoluntarily acquired trait of being coldhearted, and being born with that trait is outside of her control. ¹⁹ If we blame her for having this trait, we are blaming her for factors outside of her control.

4.b Rampant constitutive and circumstantial luck for omissions

Having clarified what constitutive and circumstantial luck are, we can use asymmetric-abilities cases to show why these kinds of luck are rampant for omissions. If we accept the Control Principle, it should not be possible for there to be the differences in responsibility that we find in asymmetric-abilities cases since whether or not an agent possesses the relevant ability or opportunity in such cases is often outside of her control. Nonetheless, I have argued that there are good theoretical reasons backing up the differences in responsibility in asymmetric-abilities cases and, as I have contended, such differences are intuitively plausible and reflect plausible principles like Corollary. Recall Epilepsy and No Epilepsy: Nina, the nonepileptic sunbather, is morally responsible for her failure to save the drowning child in a way that Ephrem, the unknowingly epileptic sunbather, is not. With Asymmetry, I have argued that this intuitive difference in responsibility is backed by reasonable assumptions about responsibility for omissions, and it results from differences in their abilities to save the child. However, it is outside either agent’s control that Ephrem has epilepsy and Nina does not. Thus, if there is any difference in Ephrem and Nina’s responsibility (and it would violate Asymmetry and Corollary to deny that there is a difference), we have an apparent tension between the Control Principle and intuitions about praise and blame.

Let’s examine this tension more closely. Epilepsy and No Epilepsy are instances of constitutive luck since the differences in internal factors about the agents’ constitutions renders Ephrem unable to save the child—and therefore ineligible to be responsible for his failure to do so—while rendering Nina able to save the child—and therefore eligible to be responsible for her omission. There would not be any moral luck here if we could also say that Nina was not morally responsible for her failure (perhaps for other reasons not related to abilities), but again, she plausibly is morally responsible in a way that Ephrem is not. Thus, we have an instance of an apparent difference in moral responsibility that derives from luck in agents’ constitutions. In other words, we have an instance of constitutive luck.

Relatedly, Child and No Child are instances of circumstantial luck. The differences in the agents’ opportunities—in other words, their external circumstances—render it such that Heathcliff faces a moral test that Catherine does not, and it is not under either agent’s control whether there is a nearby drowning child. Still, Heathcliff’s failure seems morally worse than Catherine’s precisely because Heathcliff had an opportunity to save the child while Catherine lacked that opportunity. Hence, we also have an instance of an apparent difference in moral responsibility that derives from luck in agents’ circumstances—or, in other words, circumstantial luck.

These cases of constitutive and circumstantial luck easily generalize. For cases of constitutive luck, simply take an asymmetric-abilities case with two features: the asymmetry derives from luck in the agents’ relevant general abilities, and the case is morally relevant (which ensures that the agent with the relevant ability is responsible in a way that the agent without that ability is not). For cases of circumstantial luck, do the same thing, except have the asymmetry derive from luck in the agents’ opportunities. Thus, constitutive and circumstantial luck are rampant for asymmetric-abilities cases.

Recall that asymmetric-abilities cases can only arise for omissions, not actions. For, asymmetric-abilities cases involve a difference in responsibility because of a difference in unexercised abilities, and responsibility for actions does not require consideration of unexercised abilities. This entails that constitutive and circumstantial luck are rampant for omissions in ways that they are not for actions. This rampant—and thus far unnoticed—constitutive and circumstantial luck for omissions, in turn, puts pressure on the various proposed solutions to the problem of moral luck, for this is a result that these solutions need to be able to accommodate.

4.c Challenges for existing solutions

The threat for the current solutions to the problem of moral luck is not just that there is more constitutive and circumstantial luck than previously realized. Rather, asymmetric-abilities cases show that the threat raised by constitutive and circumstantial luck is different in kind for omissions than for actions since there is a source of constitutive and circumstantial luck for omissions that does not exist for actions—cases in which there is corresponding luck in agents’ relevant unexercised abilities and opportunities. Thus, the proposed solutions to the problem of moral luck need to be able to accommodate this additional way that constitutive and circumstantial luck can arise. The various solutions to the problem of moral luck can be grouped into several broad families of views, and they each face pressure from asymmetric-abilities cases.

The first family of views holds that resultant, circumstantial, and constitutive luck do not exist—in that they cannot affect agents’ praiseworthiness or blameworthiness—because luck already systematically undermines the control necessary for moral responsibility. ²⁰ These philosophers already need to provide some sort of error theory for our ordinary judgments that agents can be morally responsible for their behaviors. This error theory now also needs to be able to handle the apparent differences in responsibility in asymmetric-abilities cases like Epilepsy and No Epilepsy, having to deny, for instance, that Nina is blameworthy for her failure to save the child in a way that Ephrem is not.

Other theorists argue that resultant, constitutive, and circumstantial luck do not exist and do not make a difference to moral responsibility because luck is irrelevant to moral responsibility and morality is luck free. ²¹ These views already need to explain away our intuitions in cases involving luck—such as the intuition that the drunk driver who actually hits someone on the sidewalk does something morally worse and is more blameworthy than the drunk driver who does not hit anyone. Asymmetric-abilities cases raise a further challenge for these views: these views must also explain away the intuitively plausible and theoretically backed difference in responsibility in these cases—like the differences in responsibility between Nina and Ephrem and between Heathcliff and Catherine.

The third group of solutions to the problem of moral luck holds that circumstantial and constitutive luck can (unproblematically) make a difference to agents’ responsibility, but resultant luck cannot. ²² This group is the most hard-pressed by asymmetric-abilities cases since these cases show that constitutive and circumstantial luck are particularly rampant for omissions. Hence, the ensuing tensions between the Control Principle and constitutive and circumstantial luck are more widespread than previously appreciated.

Finally, some hold that resultant, circumstantial, and constitutive luck can each affect moral responsibility and that morality is not luck free. ²³ Since asymmetric-abilities cases highlight the widespread extent to which constitutive and circumstantial luck can affect moral responsibility for our failures to act, the challenge to the legitimacy of moral responsibility raised by the problem of moral luck is more pressing than previously thought. These philosophers thereby face more pressure than previously realized to square luck in our constitutions and circumstances with the legitimacy of us being responsible for our behavior.

Moral luck is a threat to the legitimacy of being morally responsible for our behavior. That is, if the Control Principle is right and if it turns out that we lack sufficient control over our behavior, then it seems that few or no agents are ever, in fact, morally responsible for it. A good theory of moral responsibility for our failures to act therefore needs to be able to account for the extra threat to the legitimacy of being responsible that is raised by rampant constitutive and circumstantial luck in our omissions. Theories of moral responsibility that only focus on actions fail to account for the importance of this threat. Furthermore, we have seen that only focusing on constitutive and circumstantial luck for actions leaves out an important part of what these types of luck are and why they threaten the legitimacy of our being morally responsible. We must also consider how these types of luck arise for cases of omissions.

5. Minding the (epistemic) gap

In addition to expanding the threat of constitutive and circumstantial luck, asymmetric-abilities cases illuminate a second threat for the legitimacy of moral responsibility: a threat to the legitimacy of (at least some of) our practices of moral responsibility. In particular, asymmetric-abilities cases contain, and thereby highlight, an apparent epistemic gap in our assessments of praise and blame for others’ omissions—a gap that threatens to undermine our moral assessments of omissions.

To get an idea of this epistemic threat, consider how other people would judge Ephrem and Nina’s failures to save the drowning children in Epilepsy and No Epilepsy. It is likely that Ephrem and Nina’s failures would be judged equally blameworthy for failing to rescue the child. Ephrem is not currently having a seizure and his epilepsy isn’t otherwise obviously visible (recall, he only would have had a seizure if he had started to try to rescue the child), so, to an external observer, Ephrem and Nina each apparently notice a nearby drowning child yet continue sunbathing despite each appearing to be able to save the child. (Again, there is an important difference between blaming Ephrem and Nina for having bad moral character and blaming them for their particular failure to save a nearby child. The focus here is on accountability responsibility for actions and omissions, not attributability responsibility or responsibility for overall character.) According to the ability requirement for moral responsibility specified in Asymmetry, however, only Nina is actually responsible for her failure to save the child near her while Ephrem is, in fact, unable and therefore ineligible to be responsible for his failure to save the child near him. Hence, it is likely that external observers would make the wrong moral judgments and incorrectly hold Ephrem morally responsible for his failure to do something that he was unable to do.

More generally, given the earlier ability requirement for moral responsibility for omissions, accurately blaming or praising agents and holding them responsible for their omissions requires that we are sufficiently aware of their relevant unexercised general abilities and opportunities. However, we cannot infer what an agent’s relevant unexercised abilities are simply by looking at what abilities she in fact exercises. Thus, there is an epistemic gap in our practices of holding people morally responsible for their omissions. Note that this epistemic gap does not concern the epistemic condition for being morally responsible, which, recall, requires roughly that agents are sufficiently aware of important features of their behavior in order to be morally responsible for it. Rather, it centers on the epistemic condition for holding morally responsible: appropriately holding others accountable requires that blamers are sufficiently aware of blamees’ abilities.

In contrast, recall that (as far as abilities are concerned) holding agents responsible for their actions does not require assessing their unexercised abilities and only requires considering their relevant exercised ones. And, it is relatively straightforward to assess what an agent’s exercised abilities are—just look at what she does. Hence, this epistemic gap does not arise when holding agents responsible for their actions.

This epistemic threat to our practices of holding others accountable might not be particularly problematic if it were just restricted to a limited subset of asymmetric-abilities cases, but this epistemic worry expands beyond these cases. First, asymmetric-abilities cases, like constitutive and circumstantial luck, seem rampant for non-doings, particularly when groups of people are involved. When a group of people is blamed for failing to help someone or failing to prevent harm when any one individual was apparently able to do so, it is not unusual if some of the group members turn out to be unable to perform the act in question. ²⁴ For example, if a crowd watches a child drown, each bystander might be held responsible, yet perhaps only some were actually strong enough swimmers to save the child.

More broadly, the second and central reason why the epistemic gap in holding responsible is rampant for non-doings is that it applies to all non-doings: whenever an agent is (justifiably) held responsible for an omission, we need to accurately assess her relevant general abilities and opportunities. Additionally, the epistemic challenges posed by asymmetric-abilities cases generalize to all omission cases because we can turn any omission case into an asymmetric-abilities case by adding another agent with different general abilities or opportunities. For instance, suppose a doctor is blamed for failing to complete a difficult surgery and that part of the justification for blaming that doctor involves assessing that she was able to perform that surgery. To make this an asymmetric-abilities case, just add in another hypothetical doctor who was unable to complete that surgery. Both doctors (if actual) would likely be blamed, even though only one would have been eligible to be responsible—a likely epistemic error. It is also just a matter of luck whether the second doctor is real or merely hypothetical.

Further complicating this epistemic gap in holding responsible is that there is some suggestive psychological evidence that we are not particularly good at assessing the abilities of ourselves and others—particularly with respect to assessing our abilities relative to others’. Kruger and Dunning (1999) have famously studied and documented the psychological effect that people who are unskilled at particular tasks often greatly overestimate their abilities relative to their peers because the skills needed for the relevant task are also the same skills needed to assess how good they are at those skills—inept people are too inept to realize their own ineptitude. A reverse trend was also found with people who were skilled at a task underestimating their abilities because they assumed that tasks that were easy for them were easy for others. Additionally, there are many studies suggesting that most people think they have above-average abilities: they think they are above-average drivers, have better cognitive abilities, are more popular (and therefore are better at socializing), are harder workers, and so on. ²⁵ It is, of course, mathematically impossible for the majority of people to be above-average. These errors in the assessments of comparative abilities highlight potential errors in self-blame. Someone might inappropriately blame himself for failing to avoid a car accident after skidding on ice because (in part) he believes he was able to do so, when it turns out that avoiding the accident required far more control over his car than he could reasonably be expected to have in such bad conditions.

Clearly, we sometimes are able to assess the abilities of ourselves and others: it is not hard to tell that someone who holds still is unable to jump twenty feet straight up, nor is it hard to tell that world sprinting champion Usain Bolt is able to jog. However, assessment of abilities becomes trickier than is often appreciated, especially when considering whether someone was specifically able to do something at some particular moment. I might know that Usain Bolt typically has the ability to jog, but it is less clear that he is specifically able to do so at the particular moment when he is lying on the ground while catching his breath right after winning a race. We are often pretty sure of others’ abilities in particular situations, but this lack of certainty is important. In principle, it is very likely that at least some of our attributions of responsibility will be incorrect because we misevaluate the specific abilities of others at particular moments and in particular situations and again because there is suggestive psychological evidence that these mistaken evaluations are much more frequent than we expect.

These findings do not by themselves show that it is impossible for us to assess the abilities and opportunities of ourselves and others, nor do they show that all of our attributions of responsibility are mistaken for our omissions. Still, they spotlight a special epistemic threat to the legitimacy of our practices of holding agents responsible for their omissions.

6. Conclusion

To conclude, we have explored the connection between abilities and responsibility, as well as an important asymmetry between our actions and omissions. This then led to the identification of asymmetric-abilities cases. Such cases only arise for failures to act, and they raise two challenges for theories of moral responsibility. First, they show that constitutive and circumstantial luck are far more widespread for non-doings than has been fully appreciated thus far, and they highlight that constitutive and circumstantial luck can arise for non-doings in a different kind of way than they do for actions. This thereby puts pressure on the various proposed solutions to the problem of moral luck to be able to accommodate this prevalent luck with additional ways that it can arise. Second, asymmetric-abilities cases highlight an epistemic gap in our assessments of others’ abilities and therefore a gap in our practices of holding responsible. This paper is a lighthouse marking two dangers for theories of moral responsibility for non-doings and thereby highlighting two important features that a good theory of moral responsibility for our non-doings needs to be able to accommodate. Since these two challenges only arise for our non-doings and not for our actions, they also suggest that the project of developing a good theory of responsibility for omissions should not be treated as a mere afterthought or as a project that is derivative of the more fundamental project of giving a theory of responsibility for actions. Instead, developing a good theory of moral responsibility for omissions is a worthy and important project of its own.