Moral Responsibility for Actions and Omissions: The Asymmetry Thesis Rejected

Abstract

There is an important contemporary debate in moral responsibility about whether the following asymmetry thesis is true: moral responsibility for actions does not require alternative possibilities but moral responsibility for omissions does. In this paper, we do two things. First, we consider and reject a recent argument against the asymmetry thesis, contending that the argument fails because it rests on a false view about the metaphysics of omissions. Second, we develop and defend a new argument against the asymmetry thesis, one that avoids the problem with the first argument by not resting on any assumptions about what omissions are metaphysically.

There is an interesting contemporary debate about whether there is an asymmetry between the conditions for moral responsibility for actions and those for omissions. Specifically, the debate concerns whether the following asymmetry thesis is true: moral responsibility for actions does not require alternative possibilities but moral responsibility for omissions does. In the first part of the paper, we consider a recent argument against the asymmetry by Cyr (2017). We criticize his argument on the grounds that the metaphysical assumption on which it rests—that the agent’s omission is identical to his action—is false. Furthermore, we also show that Cyr’s criticism of the asymmetry thesis would not work even if he were to adopt one of the other main views about the metaphysics of omissions instead. But despite rejecting his argument, we agree with Cyr that the asymmetry thesis is false. So, as an alternative, in the second part of the paper, we draw on work by Widerker (2000) to develop and defend a new argument against the asymmetry thesis, one that entails that moral responsibility for actions and omissions both require alternative possibilities. Moreover, unlike Cyr’s argument, our argument against the alleged asymmetry does not rely on any metaphysical assumptions about what omissions are.

1 The Asymmetry Thesis

There are arguably reasons to think that there is the following asymmetry between the conditions that render us morally responsible for actions and those for omissions: Moral responsibility for actions does not require that an agent could have done otherwise, but moral responsibility for omissions does. This asymmetry thesis, first defended by Fischer (1986), is motivated primarily by reflecting upon some important cases.¹ On the one hand, Frankfurt-style cases (originally due to Frankfurt 1969) have been used to undermine the Principle of Alternative Possibilities (PAP), which requires that an agent could have done otherwise in order for her to be morally responsible for her action. Here is a representative case:

Belle intends to poison her cousin Finn who stands between her and a considerable inheritance. Unbeknownst to Belle, a malicious neuroscientist, Black, also wants Finn dead, but would prefer not to kill Finn himself. Aware of Belle’s plan, the neuroscientist covertly monitors Belle’s mental states via a neural implant in her brain. If Belle were to show any sign of giving up her plan and not murdering Finn, the neuroscientist (via the implant) would intervene and make Belle kill Finn. As it happens, Belle kills Finn “on her own,” without the neuroscientist having to intervene.

According to the standard interpretation of this case, Belle cannot do otherwise than murder her cousin due to the existence of the malicious neuroscientist. Yet it seems that Belle can be responsible for the murder nonetheless. In this way, Frankfurt-style cases such as this one appear to undermine the principle that an agent is morally responsible for an action only if she could have done otherwise.²

There are, however, cases in which a person omits to do something while lacking alternative possibilities that do not seem to support the equivalent view that a person can be morally responsible for her omission even if she could not have done otherwise. In fact, such cases seem to support the opposing view that a person can be morally responsible for her omission only if she could have done otherwise. To illustrate (using a case similar to one described by Fischer and Ravizza 1998):

Eda walks by the beach and notices that there is a child drowning and calling for help. She decides not to do anything about it and continues walking, thereby omitting to save the child. Unbeknownst to Eda, however, there is a school of sharks in the water. Had Eda waded into the water to try to save the child, she would have been eaten by the sharks and, so, could not have saved the child from drowning.

The standard judgment in this case is that Eda is not morally responsible for omitting to save the child. Moreover, this is because, due to the sharks’ presence, she could not have saved the child (that is, she could not have done the thing that she omitted to do). This suggests that a person can be morally responsible for an omission only if she could have done otherwise.³

Pulling together the results from these two cases—the Frankfurt-style case and the Shark case—suggests that a person can be morally responsible for her action even if she could not have done otherwise, but she cannot be morally responsible for her omission if she could not have done otherwise. Thus, these types of cases seem to support the Asymmetry Thesis for moral responsibility:

(AT) Moral responsibility for actions does not require alternative possibilities, but moral responsibility for omissions does require alternative possibilities.

2 Cyr’s Criticism

Is this asymmetry thesis true? In a recent paper, Cyr (2017) argues not. He presents a Frankfurt-style omission case—Ben*—in which the person’s omission is arguably identical to his action:

Black wishes Ben to cover up his friend’s murder. In order to ensure that Ben does this, Black implants a chip in Ben’s brain which allows him to control Ben’s behavior during questioning. (Ben has no idea about any of this.) Black prefers that Ben cover up the murder on his own. But if Ben starts to become inclined to raise his hand (to indicate that his friend committed the murder), Black will immediately use his chip to cause Ben to remain perfectly still instead. As it turns out, though, Ben holds perfectly still on his own, omitting to raise his hand, and Black never exerts any causal influence on Ben’s behavior (Cyr 2017, p. 3155).⁴

In analyzing this case, Cyr concludes that if Ben’s omitting to raise his hand is identical to his keeping it by his side (as, Cyr says, is plausible), then the asymmetry thesis yields a contradiction. According to the asymmetry thesis, Ben can be morally responsible for his action (because moral responsibility for actions does not require alternative possibilities) but not responsible for his omission (since moral responsibility for omissions does require alternative possibilities). But since the omission and action are identical, then it follows from the asymmetry thesis, that Ben both can and cannot be morally responsible for the very same thing—a contradiction. Therefore, Cyr argues, the asymmetry thesis is false.⁵

Is Cyr’s argument convincing? We argue not. Specifically, we argue that the metaphysical view about omissions on which it rests—that Ben’s omission is identical to his action and, more generally, that at least some omissions are identical to actions—is false. We base our criticism on the claim that Ben’s omission is multiply realizable and, if it is, then his omission is not identical to his action after all.⁶ Before continuing, we note that our argument has broader consequences than simply undermining Cyr’s argument against the asymmetry thesis. For instance, according to one prominent view of omissions—the omission-action identity view (defended by Davidson 1985)—all omissions are identical to actions. By contrast, Clarke (2014) and Fischer (2017) develop a more nuanced hybrid view, proposing that while some omissions are identical to actions, others are absences—they are metaphysically nothing at all. In addition to refuting Cyr’s argument against AT, our argument that it’s false that at least some omissions are identical to actions also undermines both the omission-action identity view and the Clarke/Fischer hybrid view.

In order to develop the multiple realizability point, let us first consider the following claim about the logic of identity statements:

(I) If event-X = event-Y, then had Y not have happened, X would not have happened either.

This claim looks unassailable. But the Ben* case does not meet it. In that case, Ben’s omitting to raise his arm is said to be identical to his holding his arm still. Applied to the claim about identity, we get:

(I*) If Ben’s omitting to raise his arm = Ben’s holding his arm still, then if it had not been the case that Ben had held his arm still, then it would not be the case that he omitted to raise his arm.

But (I*) is false. Even if it had not been the case that Ben had held his arm still, he might still have omitted to raise his arm. This is because there is more than one way in which Ben could omit to raise his arm. One way he could do this is by holding his arm still, as Cyr suggests. But another way he could omit to raise his arm is by being paralyzed by fear. Suppose when the police officer asked him to raise his hand if he knew that his friend committed the murder, Ben became so paralyzed by fear that the police officer would find out what he knew that he could not move. Thus, by being paralyzed by fear (and not by holding his arm still), Ben omitted to raise his arm. So, even if it had not been the case that Ben had held his arm still, he might have been paralyzed by fear, in which case he would still have omitted to raise his arm, contrary to (I*).

It might be objected, however, that even if the multiple realizability claim works in this paralysis case, it will not work in a case when Ben’s omission to raise his arm is intentional. That is, there seems to be only one way in which Ben can intentionally omit to raise his arm, namely, by holding his arm still. So, it might still be argued that Ben’s intentionally omitting to raise his arm is identical to his holding his arm still.⁷ But Ben’s intentionally omitting to raise his arm is just as multiply realizable as his omitting to raise his arm. Instead of holding his arm still, Ben could intentionally omit to raise his arm simply by relaxing it. (‘Holding’ one’s arm still suggests that one is tensing one’s muscles as a way of being sure that one’s arm does not move. By contrast, relaxing one’s arm involves the opposite—relaxing one’s muscles—as a way of keeping one’s arm still.) Therefore, the multiple realizability claim—and our argument that Ben’s omission is not identical to his action—would not be refuted by the case of Ben intentionally omitting to raise his arm.⁸

So far, we have argued that Ben’s omission is not identical to his action and, therefore, that the metaphysical assumption about Ben’s omission that Cyr makes is false. But we now make a further point. Cyr—as well as Clarke (2014) and Fischer (2017)—argue that if there is any successful example of an omission being identical to an action, then Ben’s omitting to raise his arm by holding it still will be one such case.⁹ But since we’re arguing that the Ben* case is not a successful example of an omission-action identity, then—according to the thought shared by Cyr and others—we’re therefore justified in concluding that no omissions are identical to actions. In other words:

• (1) If there is any successful example of an omission being identical to an action, then (according to Cyr and others) the Ben* case will be one such example.

• (2) The Ben* case is not such an example (since Ben’s omitting to raise his arm is multiply realizable).

• (3) Therefore, there is no successful example of an omission being identical to an action.

• (4) Therefore, it is not the case—as Cyr and others contend—that at least some omissions are identical to actions.¹⁰

3 Further Discussion

However, even if the view that at least some omissions are identical to actions is successfully refuted by way of omissions being multiply realizable, there is a further concern to address. That is, besides the omission-action identity view, there are three other main views about what omissions are: the constitution view, the negative entity view, and the absence view. Although Cyr argues that Ben’s omission is identical to action, if (as we argue) this view of omissions is incorrect, could Cyr appeal to one of the other three main views of omissions to make his argument against AT work? In other words, will AT still yield the kind of contradiction that is present in cases such as Ben* if one of these other main views of omissions is assumed? We now argue not and therefore that cases like Ben* would not undermine the asymmetry thesis even if one of these three other views of omissions were true.

According to the constitution view—defended, for instance, by Fischer and Ravizza (1998) and Palmer (2018)—omissions are not identical to actions; instead, they are constituted by them.¹¹ If omissions are identical to positive actions, then since there is only one event, anything that is true of the one will be true of the other, including the person’s moral responsibility. By contrast, if omissions are only constituted by, but not identical to, actions, then there will be two events rather than one and it need not be true that the omission and the action will share all the same properties. In particular, it will not be true that if a person is morally responsible for the one, then she must be morally responsible for the other. (By analogy, if the statue is constituted by the clay, but not identical to it, then it is widely accepted that since the two are different things, the statue and the clay will not share all the same properties. The statue, for instance, might be beautiful; but it would not follow from this that the clay is beautiful too.)

Now let us apply this to the Ben* case. As we said above, it follows from the asymmetry thesis that Ben cannot be morally responsible for omitting to raise his hand, since he could not have done otherwise than omit to do what he did. But he can be morally responsible for holding his body still because according to AT, a person can be morally responsible for an action even if he could not have done otherwise. But if the constitution view is right, we do not get the contradiction. Specifically, if Ben’s omission is constituted by but not identical to his action, then his omission and action are two different things and need not share all the same properties. In this way, it will not be true that Ben both is and is not morally responsible for the same thing. And since there is no contradiction, the Ben* case would not refute AT while holding the constitution view.

If the previous argument is sound, then it follows that if we hold one of the other two main views of omissions—the negative entity view of omissions as discussed by Bernstein (2015) and Clarke (2014) and the absence view that is defended by Clarke (2014) and Fischer (2017)—then Cyr’s argument would not refute AT either. Specifically, if omissions are some kind of negative entity (like a negative fact or negative states of affairs), or are literally metaphysically nothing (as the absence view says), then even if—as AT holds—Ben can be morally responsible for his action but not for his omission—then we do not get the contradiction. For on neither of these views is Ben’s omission identical to his action. It is either literally nothing or a negative entity. And if his omission is not identical to his action, it will not follow, from AT, that Ben both is and is not morally responsible for the same thing. Thus, Cyr’s criticism does not refute the asymmetry thesis if any of the other three main views of omissions are assumed.

4 A New Argument Against the Asymmetry Thesis

In the previous sections, we argued that Cyr’s line of argument does not undermine the asymmetry thesis on the grounds that it assumes a false view about the metaphysical nature of omissions. Moreover, we demonstrated that his argument does not refute the asymmetry thesis if any of the other central views of the metaphysics of omissions are assumed instead. In this section, we switch gears. We develop a new argument against the asymmetry thesis, but one that does not involve making any assumptions about the metaphysical nature of omissions. In particular, we develop an argument from Widerker (2000) that moral responsibility for actions does require alternative possibilities. We then show how this same argument can be applied to omissions, entailing that moral responsibility for omissions also requires alternative possibilities. Since the argument entails that both moral responsibility for actions and omissions requires alternative possibilities, we conclude that the asymmetry thesis is false.

To begin, why think that moral responsibility for actions requires alternative possibilities? Widerker (2000, 2005) develops an argument for this claim. In the standard Frankfurt action case, a person, Jones, could not have done otherwise than, say, decide to break his promise (because of Black’s presence and plan). Yet it seems Jones can be blameworthy for his decision anyway. In response, Widerker poses a question to Frankfurt and defenders of Frankfurt’s position:

[S]ince you, Frankfurt, wish to hold him blameworthy for his decision to break his promise, tell me what, in your opinion, should he have done instead? Now, you cannot claim that he should not have decided to break his promise, since this was something that was not in Jones’s power to do. Hence, I do not see how you can hold Jones blameworthy for his decision to break the promise (p. 191).

Widerker calls this the “What-should-he-have-done defense” of PAP or the “W-defense” for short.

It is important to note that, by asking about what Jones “should” have done instead, Widerker does not mean to be asking about what Jones morally ought to have done instead. Rather, he means to ask about what it would be morally reasonable to expect Jones to have done instead. Furthermore, by asking what it would be reasonable to expect Jones to have done instead, Widerker isn’t asking about what it would be reasonable to “anticipate” or “predict” what Jones would have done instead. Rather, he is asking about what it would be reasonable to “demand” that Jones have done instead. The sense of demand here is the sense in which, to use an example from Capes (2010), we often expect—that is, demand—that our children behave in public. Moreover, we have these expectations of our children even if, as a matter of fact, we don’t think that it’s very likely that they will do what we are demanding that they do. It is, then, this expectation-as-demand, rather than expectation-as-prediction, that is at issue in the W-defense.¹²

With these clarifications, we present the W-defense for PAP as follows (Widerker 2000, p. 192):

• (1) An agent S is morally blameworthy for doing A only if, in the circumstances, it would be morally reasonable to expect S not to have done A.¹³

• (2) If S could not have avoided doing A, then on pain of expecting him to have done the impossible (and do something that he could not do), it would be morally unreasonable to expect him not to have done A.

• (3) Hence, if S could not have avoided doing A, then S is not morally blameworthy for doing A.

This, of course, is an argument for the claim that moral responsibility for actions requires alternative possibilities. But as we now explain, the same argument can also be used to show that moral responsibility for omissions requires alternative possibilities as well.

Suppose, using Cyr’s case, that Ben wants to cover up the murder his friend did. Ben is asked by the police to raise his hand if his friend committed the murder, and since he wants to cover it up, Ben omits to raise his hand. What conditions must Ben meet in order to be morally responsible for omitting to raise his hand (assuming that he can be responsible for it)? Applying the W-defense for PAP to omissions, we can reason as follows. Ben can be morally blameworthy for omitting to raise his hand only if it would be morally reasonable to expect him to have raised his hand—i.e., morally reasonable to expect him to have done what he omitted to do. But if he could not have avoided omitting to raise his hand (that is, if he was not free to do anything other than omit to raise his hand), then on pain of expecting him to have done the impossible (and do something—raise his hand—that he was not free to do), it would not be morally reasonable to expect him to have raised his hand. Therefore, if he could not have avoided omitting to raise his hand, he cannot be morally responsible for his omission.

Putting this in premise-conclusion form and applying it to any agent (and not just Ben), we get:

• (1*) An agent S is morally blameworthy for omitting to A only if, in the circumstances, it would be morally reasonable to expect S to have done A.

• (2*) If S could not have avoided omitting to A, then on pain of expecting him to have done the impossible (and do something that he could not do), it would be morally unreasonable to expect him to have done A.

• (3*) Hence, if S could not have avoided omitting to A, then S is not morally blameworthy for omitting to A.

If sound, this argument shows that moral responsibility for omissions requires alternative possibilities. Moreover, the earlier version of the W-defense shows that moral responsibility for actions requires alternative possibilities. So, we have an argument—the W-defense—that can be applied to both actions and omissions, showing that moral responsibility for actions and omissions both requires alternative possibilities. If this argument is sound, then the asymmetry thesis—that only responsibility for omissions, but not actions, requires alternative possibilities—is false.

We end this section by responding to an important criticism. It might be objected that the W-defense simply begs the question against the Frankfurt case defender. After all, anyone who accepts that the Frankfurt “action” cases undermine PAP will surely also think that they undermine the W-defense as well.¹⁴ In particular, they’ll think that the cases refute premise (1), showing that, contrary to what (1) says, a person can be blameworthy for her action even if it would not be morally reasonable to expect her not to have acted as she did (cf. Haji 2016, p. 185). Can we allay this concern?

There are two issues here: first, whether the W-defense (specifically premise (1)) begs the question against the Frankfurt case defenders and, second, whether there’s any good reason to think that the truth of (1) should trump the blameworthiness judgment from the Frankfurt cases or whether that blameworthiness judgment should undermine the truth of (1). Beginning with the question-begging charge, we concede that (1) would beg the question against the Frankfurt case defenders if there were no good reason aside from the truth of PAP to think that (1) is true. We believe, however, that there is such a reason. Widerker (2000) and Palmer (2013), for instance, suggest that a reason to think that (1) is true that’s independent of—i.e., that doesn’t rest on—the truth of PAP is that (1) explains why people are not blameworthy in a range of cases in which they could have done otherwise—in particular, in cases in which (even though they could have done otherwise) people’s non-culpable lack of knowledge explains why they aren’t blameworthy.¹⁵

Now, this justification for (1) isn’t beyond reproach (Haji 2016, for instance, offers important criticism). But if this is a good reason to think that (1) is true that’s independent of the claim that PAP is true, then, we argue, the charge that (1) simply begs the question against Frankfurt defenders is too quick. Moreover, this justification for (1) also provides a response to the second issue raised by the question-begging concern. For on this proposal, (1) has a broad “explanatory power” to explain why people aren’t blameworthy in a wide range of cases, including those in which they could have done otherwise. Furthermore, this fact about (1) is itself a reason to think that the truth of (1) should trump or call into question the blameworthiness judgment from the Frankfurt cases rather than the Frankfurt case judgment trumping (1). We conclude, therefore, that neither issue raised by the question-begging concern sticks.

5 Further Defense

We now strengthen our criticism of the asymmetry thesis by responding to two broad concerns about our argument. First, despite the fact that we have shown that Cyr’s argument does not work, critics might suggest that the way Cyr argues nevertheless enjoys an important dialectical advantage over the way we do. Specifically, Cyr argues against the asymmetry thesis without taking a stand on whether moral responsibility for actions and omissions requires or does not require alternative possibilities. He just argues that the asymmetry thesis is false because it entails a contradiction in certain cases. By contrast, our argument does take a stand on this, claiming that the asymmetry thesis is false on the grounds that there is a convincing argument that both responsibility for actions and omissions requires alternative possibilities. It might, therefore, be considered a dialectical advantage if an argument against the asymmetry thesis does not involve taking a stand on whether responsibility for actions and omissions does (or doesn’t) require alternative possibilities than if an argument does take such a stand.

But even if this were a weakness, a counterweight (and thus a strength of our way of arguing and a weakness of Cyr’s) is that it is dialectically preferable to argue against the asymmetry thesis in a way that does not rely on any metaphysical assumptions about what omissions are. As we explained in Sect. 3, there are a variety of views about what omissions are metaphysically, and so any such assumption will be controversial. Therefore, all else being equal, it’s preferable to argue against the asymmetry thesis in a way that does not involve any such assumptions, and this is a key feature of our argument.

Second, because our argument does not depend on any intuitions about Frankfurt-style action cases or Shark-type omission cases, there is a sense in which our rejection of AT is seemingly independent of these commonly accepted reasons for AT. This independence might be exploited by critics in two ways. First, proponents of AT might counter our argument by appealing simply to intuitions about Frankfurt-style action cases and Shark-type omission cases as reasons to think that AT is true after all. Others, however, might agree about the lesson from the Frankfurt-style action cases but draw a different conclusion about AT. They might appeal to intuitions from Frankfurt-style omission cases to suggest that AT is in fact false but in a different way than we say. According to this proposal, intuitions about Frankfurt-style action cases as well as those from Frankfurt-style omission cases suggest that neither moral responsibility for actions nor moral responsibility for omissions requires alternative possibilities (Fischer and Ravizza 1998; Pereboom 2015).

In reply, we argue that the W-defense does double-duty. On the one hand, it comprises an argument (when extended to omissions) for the symmetrical view of responsibility that we favor—both responsibility for actions and omissions requires alternative possibilities. But on the other hand, the W-defense can itself be used to challenge the intuitions about both the Frankfurt-style action cases and Frankfurt-style omission cases. First, it can undercut the intuition (that both types of critics are appealing to) that the agent in a Frankfurt-style action case is responsible. By this line of reasoning, we should judge, on balance, that she is not morally responsible for her action because (1) she can be blameworthy only if it would be reasonable to expect her not to act as she did, but (2) since she could not do otherwise, this expectation would not be reasonable.¹⁶ In addition, the W-defense can also be used to challenge the judgment that an agent in a Frankfurt-style omission case is responsible (as proponents of the other kind of symmetry thesis contend). After all, a person can be responsible for her omission only if it would be reasonable to expect her to have done what she omitted to do, but—again—such an expectation would not be reasonable if she could not do otherwise.

If, as we have suggested, the W-defense undercuts the judgments that agents are responsible in both Frankfurt-style action and Frankfurt-style omission cases, then the landscape changes. Not only does the W-defense comprise an argument for the symmetrical view of moral responsibility, a key component of the argument in favor of the alternative asymmetrical view (and both components of the argument for the alternative symmetrical view from ours) will be undermined. In contrast, then, with what some philosophers believe, not only are the requirements for responsibility for actions and omissions importantly symmetrical, they are symmetrical in a crucial way: both require alternative possibilities.

6 Conclusion

In this paper, we have done two things. First, we have shown that Cyr’s argument against the asymmetry thesis rests on a false metaphysical assumption about omissions and that his criticism will not work if we assume one of the other main views of omissions instead. Second, we have shown that an important argument for the view that moral responsibility for actions requires alternative possibilities—the W-defense—can be extended to show that moral responsibility for omissions requires alternative possibilities as well. We thus conclude that the asymmetry thesis is false and we do so without making any metaphysical assumptions about what omissions are.¹⁷