PRELIMINARIES

Some philosophers contend that if concepts C-1, C-2, and C-3 are all “concepts of omission,” then there must be some common conceptual element among them, and the common element, rather than C-1, C-2, or C-3, will be the logically fundamental concept of omission. Setting aside familiar family-resemblance responses, I do not reject the claim that there is some common conceptual element to the multiple-omission concepts I will explicate in Section III. The common element, however, is very abstract and lacks any significant value in the context in which philosophers employ the concept of omission. The common element is not an “empty concept,” whatever such a concept would be, but it does have minimal content and is relatively useless practically and theoretically.¹

The element in question is the mere idea of some agent not performing some act, that is, the idea captured by propositions of the form “It is not the case that A did s,” where “A” stands for some agent and “s” stands for some act type. Following Smith, I will call this the idea of “nonaction.”

Feinberg and Smith are insistent that nonaction is different from omission,²

and they are surely right if we are looking for a concept of omission that has some philosophical or scientific payoff in understanding human behavior and how it is to be regulated. Suppose that someone is drowning off a beach in New Jersey. The lifeguard on duty watches as the event unfolds and does nothing to help. I am in Minnesota at the time. It is true of both the lifeguard and me that we did not help the (now) drowned person. Nonaction is the concept that captures our behavior. However, by failing to distinguish the lifeguard on duty from me and all the billions of others who did not help the drowning individual, the concept of nonaction is not useful in most contexts.³ Recognition of the sparse content of the concept is precisely what led Feinberg and then Smith to construct a concept of omission by packing in additional content related to the conditions in which someone can exercise her agency and the reasonableness of her doing so.

I have no objection to treating nonaction as one of the logically primitive concepts in the domain of omission. My point here is that, as Feinberg supposed, a philosophically useful omission concept must be constructed by bringing in additional content. Later I will aim to show that once one begins to build up from the idea of nonaction, the idea that there is a unique, logically fundamental concept of omission is difficult to sustain.

It might be objected that my approach to understanding omission disregards the distinction between pragmatics and semantics.⁴

The use or value of a concept is different from its meaning. I do not wish to reject this distinction. My claim is that once one follows Feinberg in building up from the concept of mere nonaction, then pragmatic considerations will play a role in deciding how to construct the concept of omission and whether there are multiple paths that can be taken.

SMITH AND FEINBERG: COMPARED AND CRITICIZED

Smith endorses Feinberg's idea that not all of the true statements about what an agent has not done refer to omissions by the agent. Her central aim is to identify and explain what the extra element is that turns a nondoing into an omission. Focusing on one of the elements enumerated by Feinberg, she claims that the “primary distinguishing feature”⁵

of those nonactions that count as omissions is a normative element that concerns what it was reasonable to expect the agent to do. For example, my not handing back my students' (graded) exams within two weeks of the test was an omission of mine because it was reasonable to expect me to hand them back. Not going to India last week, in contrast, was a nonaction but not an omission because it was not reasonable to expect me to make such a trip.

For there to be a reasonable expectation, it must be possible for the agent to perform the act in question, and there must be some norm of conduct involving such acts and applying to the agent (at least under ordinary conditions). On Smith's analysis, if we assume that an act, s, is possible for an agent, A, then there are two sorts of norms, each individually sufficient to make it reasonable to expect agent A to do s:

The first type of norm is a “pattern of human conduct,”⁷

and its inclusion in Smith's analysis reflects Feinberg's idea that it can be reasonable to expect that A do X if “A or people in A's position ordinarily do X.”⁸ Norms of type 2 are not so much patterns of conduct as they are standards that deem conduct of a certain kind to be conduct that the agent ought to perform. The inclusion of this type of norm in Smith's account serves to bring together an array of other conditions cited by Feinberg as helping to make a nonaction into an omission.⁹ Smith then unifies Feinberg's discussion by explaining that omissions are essentially forms of “deviant behavior,” that is, they are deviations from some applicable norm. Hence she writes that “if the norm is gone, then so is the omission.”¹⁰

The guiding idea of Smith's analysis is that the norms that turn nonaction into omission are standards that give agents reasons for action. In a case of mere nonaction, there is no norm that gives the agent a reason to perform the action in question. Norms for rescuing a swimmer drowning off the New Jersey coast do not give a person canoeing in Minnesota a reason for action with respect to the swimmer. The norms do not apply to the person in the canoe, and hence we have nonaction rather than omission. In contrast, the norms do apply to a lifeguard who is on duty at the beach where the swimmer is drowning, and so we have an omission when the lifeguard violates the norm.¹¹

Smith utilizes the elements of her account to explain the ideas of unintentional, excusable, and justifiable omissions. Unintentional omissions are ones in which the act undone is one that the agent did not think of performing at the time in question. Excusable omissions are ones in which there is a norm that provides the agent with a reason to perform the act she has left undone, but there is another norm that, under the circumstances, exculpates her from not performing the act. Justifiable omissions are ones in which there is a norm that would normally provide the agent with a reason to perform the act left undone, but there is another norm that, under the circumstances, provides a reason that defeats the first reason and is itself undefeated by any other reason.¹²

Smith's discussion of unintentional, excusable, and justifiable omissions is a clear advance over Feinberg's account, which is tailored to deal with intentional omissions such as those of bad Samaritans who deliberately choose not to rescue a stranger from grave peril even when they know that the rescue will involve little cost to themselves. I have no objections to her accounts of those specific forms of omission. However, her general account of omission, despite simplifying and unifying Feinberg's analysis, carries over some of the flaws in his approach.

Consider the term “reasonable expectation,” central to how both Smith and Feinberg explain omission. The term is ambiguous in a way that plagues their discussions of the normative element of omissions. A “reasonable expectation” can refer to what one has good grounds to think will be done or to what one has good grounds to think ought to be done. In the one case, we have an epistemic sense of “reasonable expectation”; in the other, we have a moral or practical-reason sense. Perhaps our folk concept of omission conflates these two senses. Empirical studies might reveal as much, and then Smith and Feinberg would be faithfully reflecting the folk concept. But a philosophical account should surely do more than simply reflect conflations embedded in folk psychology. Rather, any conflation of the epistemic and the moral ought to be flagged and then eliminated in a philosophically rehabilitated omission concept.

Before suggesting in the next section how such a rehabilitation might proceed, there are serious problems to address concerning the two types of norms invoked by Smith to explain how nonaction is turned into omission. Consider norms of type 1, which are essentially patterns of behavior. It seems to me that such patterns are not by themselves norms. Even as “norm” is used in the social sciences, a behavioral regularity is insufficient for the existence of a norm among a certain population. H.L.A. Hart gives perhaps the most illuminating account of the insufficiency of a purely behavioral account of norms. He argues that mere convergent behavior in a society does not amount to a rule, as deviation might not bring forth censure or criticism or any other normative response. In contrast, when there is a social rule, or norm, that calls for the behavior in question, “not only is…criticism in fact made but deviation from the standard is generally accepted as good reason for making it.”¹³

If Hart's account of this “internal aspect” of social rules is roughly correct—and I think that it is—then it is wrong to think that a norm can exist simply in virtue of behavioral regularities.

Moreover, related elements of Hart's theory pose problems for norms of type 2. Hart claims that the duties imposed by social rules are conceptually distinct from those imposed by morality. Social rules impose duties “[w]hen the general demand for conformity is insistent, and the social pressure brought to bear upon those who deviate or threaten to deviate is great.”¹⁴

But morality, that is, critical rather than conventional morality, imposes duties quite apart from social pressure or insistent demands for conformity. Thus, on Hart's view, the idea of a social-rule obligation is one thing and the idea of a moral obligation is quite another.

If Hart's view is more or less accurate, then the problem for Smith is that she describes type-2 norms in a way that glosses over the distinction between social-rule obligations and moral obligations.¹⁵

In doing so, she leaves it unclear whether she thinks that any social rule that imposes a requirement or responsibility on an agent can turn a nonaction into an omission. In the next section, I argue that distinguishing the two sorts of obligations leads to two equally fundamental concepts of omission.

TWO CONCEPTS OF OMISSION

If we aim to construct a philosophically or scientifically useful concept of omission from the more primitive idea of nonaction, then we should proceed by (1) untangling social rules from moral ones; (2) bracketing the question of whether morality is a matter of practical reason; and (3) presuming that there are norms of practical reason whose validity is not simply a matter of what people believe or want, whatever the exact content of those norms may be. This way of proceeding makes sense in order to avoid the problems that Hart analyzes when the social and the moral are not distinguished, to sidestep the long-standing debate over whether morality is a matter of reason or emotion, and to operate with only minimal presuppositions about practical reason. Approaching the matter of omission in this way leads to two concepts of omission, neither one logically more fundamental than the other.

The first is the social concept of omission, [omission-sc], according to which a deviation from an applicable social rule constitutes an omission by the agent. If there is sufficient convergent behavior in doing s, and enough people in the relevant population think that there is good reason to criticize people in A's situation who do not do s, then A's not doing s is an omission. The question of whether the people in that population are right in thinking that there is reason to criticize nondoers is beside the point if one is employing the social concept. The actual reasonableness or unreasonableness does not dictate the application of [omission-sc]. The abstraction from the norms of practical reason is useful in describing, understanding, explaining, and predicting human behavior. This is not to say that such norms are useless or wholly dispensable for those purposes. Rather, it is to register the fact that people usually do what they think reasonable and they are often wrong in what they think. So it is sometimes sensible to set aside the question of whether they are wrong and to focus on what they think.

The second concept of omission is the practical-reason concept, [omission-pr], according to which a deviation from an applicable norm of practical reason is necessary and (given the agent's ability and opportunity to satisfy the norm) sufficient to constitute a course of conduct as an omission. Deviation from a social rule would be insufficient, as the rule might itself be unreasonable. For purposes of evaluating an agent's behavior and the social rules under which she is operating, [omission-pr] is indispensable.

It might be argued that [omission-pr] is logically more fundamental than [omission-sc]. People cannot be wrong in thinking that something is reasonable unless there are norms of practical reason whose validity is not determined by what people think. Accordingly, the idea of actual practical reasonableness is more fundamental than the idea of perceived practical reasonableness in the same way that the idea of triangularity is more fundamental than the idea of perceived triangularity.

This argument misses the mark, however. My claim is not that actual reasonableness and perceived reasonableness are logically on a par. I agree that the former concept is logically prior to the latter. Rather, the claim is that a concept of omission that abstracts from the idea of actual practical reasonableness is no less logically fundamental than a concept of omission that incorporates the idea.

Some thinkers will question my understanding of the distinction between a descriptive/explanatory perspective, on the one side, and an evaluative perspective, on the other. They might argue that when it comes to human conduct, the evaluative is conceptually more fundamental than the descriptive/explanatory.¹⁶

Accordingly, they will conclude that [omission-pr] is more fundamental than [omission-sc].

I do not have anything substantive to say here about the relative priority of the evaluative and the descriptive/explanatory standpoints. If the question of whether there are two equally fundamental concepts of omission comes down to that issue, then I do not think that I am playing a weak hand. In any event, I hope to have offered a plausible start in making a case that there are distinct concepts of omission, each useful in its own way and neither logically more basic than the other.