A summary of Clarke’s ‘Ability and responsibility for omissions’

[ PDF version ]

1. Kinds of omission

In this paper Clarke takes a look an asymmetry thesis concerning our moral responsibility for actions as compared to omissions.  The thesis in question, advanced by (among others) Fischer (1985) and Fischer & Ravizza (1991), maintains that although the possession of alternative possibilities is not a condition on responsibility for actions, it is a condition on responsibility for omissions.  Clarke challenges this thesis by appealing to Frankfurt-style cases.

Clarke begins by sketching a way to characterise different kinds of omission.  First, omissions can be intentional or unintentional.  Intentional omissions are those where “someone intends not to A, … does not A, and [the] not A-ing results, in a nondeviant way, from her intending not to A” (1994, p. 195).  The non-deviancy clause is there for the standard reasons.  Clarke distinguishes between two kinds of intentional omission.  First is the refraining or act of omission.  To illustrate: Tina sees Tom is about to fall, she decides not to help, and as a result, doesn’t help.  This is to be distinguished from intentional omissions which are not acts (not refrainings).  Clarke’s example here is Ulysses, who, intending not to succumb to the Sirens, but knowing he will want to at a later time, has himself tied to the mast.  This does not count as a refraining because at the time of the omission Ulysses is actively trying to get to the Sirens.  Clarke does not elaborate on this distinction, but trusts we have an intuitive grasp of it.  In addition, he thinks that conclusions about the asymmetry case will be the same in both cases (1994, p. 197).

These two kinds of intentional omission are contrasted with unintentional omissions, such as forgetting to do something that one has an obligation to do.  We can be responsible for such omissions in virtue of being responsible for an appropriate previous intentional action (or perhaps for some aspect of our character).

Clarke thinks there are three things we might be referring to when we talk about responsibility for intentional omissions (1994, pp. 196–197):

(1) Responsibility for the decision (or other forming of the intention) not to perform the action.

(2) Responsibility for a certain fact resulting from their intention (namely, the fact that the agent doesn’t perform some action).

(3) Responsibility for the fact that the agent’s not performing the action results (in the intended way) from their intention.

Defenders of the asymmetry thesis cannot be concerned with (1) because decisions are positive actions (and so won’t have any alternative possibility requirement).  The difference between (2) and (3) is subtle.  If we re-state them in order to highlight the fact that the agent is responsible for, we get:

(2’) the fact that the agent didn’t perform action A of type T

(3’) the fact that the agent didn’t perform action A of type T as a result of their intention [1]

We can now see that case (3) includes responsibility for some kind of connection between the intention and the fact, whereas (2) limits responsibility to the fact about the non-action.  Clarke says that (3) is most properly associated with refrainings, and this is because refrainings, being acts, have an active component.  When we consider the responsibility associated with refrainings, then, we’re considering responsibility for “a complex state of affairs,” whereas the responsibility associated with other intentional omissions is only for some fact which describes the omission.

What all of this means is that, excepting refrainings, responsibility for omissions is indirect – that is, it depends on “an earlier intention or an earlier action or actions” (1994, p. 198).  This is true of both unintentional omissions and non-refraining intentional omissions.  Responsibility for refrainings will either be (wholly) indirect or conjoint.  The latter will be the case if the refrainings involve an active component.  The former is possible, however, because sometimes a refraining is such that “the agent is only indirectly responsible for forming the intention not to perform the action in question” (1994, pp. 207 n.8).  This leaves us with something like the following breakdown:

At this point it is worth commenting on the difference between refrainings and non-refraining intentional omissions that Clarke seems to have in mind.  Clarke explicitly says that as far as non-refraining intentional omissions go, he is concerned with “responsibility for the fact that results from the decision or intention, the fact, that is, that the agent does not perform an action of a certain type” (Clarke 1994, pp. 197 /yearonly, emphasis mine).  So the responsibility is for a fact; but this fact results from decisions or intentions just as it does in the case of refrainings.

This means that the difference between refrainings and non-refraining intentional omissions cannot be a matter of whether or not there was a prior decision or intention.  The difference, instead, seems to be that than in the case of refrainings, the decision or intention is present at the time of the omission (or immediately before the omission).  Such a difference appears to be one of degree rather than of kind (if Clarke had restricted refrainings to those intentional omissions with an active component, it would be more plausible to say the difference was one of kind).

 2. Doubly and singly indirect responsibility

So far we’ve seen that responsibility for an omission always has an indirect component (and will often be wholly indirect).  Clarke goes on to make a further distinction between types of indirect responsibility: singly indirect responsibility and doubly indirect responsibility.

Singly indirect responsibility

Cases of singly indirect responsibility are cases where the responsibility for the omission depends “only on responsibility for an earlier action or actions, or for the intention that is included in a refraining” (1994, p.201).  Clarke thinks that in some cases of singly indirect responsibility for not φ-ing, an inability to φ would undermine responsibility, whereas in other cases it wouldn’t.  This is the general form of his argument (which he deploys for doubly indirect responsibility too, considered in the following section): sometimes inability undermines responsibility, sometimes it doesn’t; given that, the possession of alternative possibilities can’t be a requirement on being responsible for omissions. Clarke’s begins with the kind of example used by defenders of an AP requirement on omissions:

(Case 1) Joe is strolling along the beach when he sees a child struggling in the water. Joe believes that he can rescue the child with minimal effort and inconvenience, but he decides not to go to the trouble, and he continues walking along the beach. The child drowns. Unbeknownst to Joe, there is a shark patrolling the water between the beach and the struggling child, and had Joe jumped into the water, the shark would have attacked and eaten him (1994, p.201).

Clarke agrees with the defenders of the asymmetry thesis that in this case, Joe is not responsible for failing to save the child.  But Clarke asks us to compare Case 1 with the following:

(Case 2) Joe is strolling along the beach when he sees a child struggling in the water. Joe believes that he can rescue the child with minimal effort and inconvenience, but he decides not to go to the trouble, and he continues walking along the beach. The child drowns. Unbeknownst to Joe, a mad scientist is monitoring his thoughts. Had Joe shown any inclination to decide to rescue the child, the scientist would have intervened and ensured that Joe not make such a decision but instead decide not to rescue the child (and consequently refrain from rescuing the child). The scientist would not have intervened in any other way. As it happened, the scientist did not intervene at all; there was no need to (1994, p.202).

In Case 2 Joe is responsible for failing to save the child.  The crucial difference for Clarke is that in Case 2 but not in Case 1 it is true that “had Joe decided otherwise, he would have saved the child” (1994, p.202).  This is true, Clarke says, because the mad scientist would not have interfered post-decision if Joe had succeeded in making the decision to save the child.  Joe is free to execute the decision he makes.  Clarke takes Joe in Case 2 to possess a conditional ability which means that his omission depends on his decision in a way which is not true of Case 1.  (Analogous cases for unintentional omissions can be produced to mirror this difference).

Clarke therefore proposes the following principle as an improvement on the principles offered by the proponents of the asymmetry thesis (1994, p.204):

(SICRO) When an agent

(i) omits to perform an action of a certain type at a certain time, and

(ii) is unable to perform such an action at that time,

then she bears singly indirect or conjoint responsibility for the omission only if:

had she intended to perform that action, and

had she tried to carry out that intention,

then she would have performed the omitted action.

Principle (SICRO) is a necessary condition for bearing singly indirect responsibility for an omission ifit is true that the agent was unable to perform the action omitted.

Doubly indirect responsibility

All responsibility for omissions (or the result thereof) has some indirect component.  But sometimes there is what Clarke calls doubly indirect responsibility.  Clarke introduces this notion by saying that responsibility will be doubly indirect “when that responsibility depends on responsibility for having an inability, and when responsibility for having that inability depends, in turn, on responsibility for an action or actions from which the inability results” (1994, p.199).  Later comments (where Clarke presents an example of doubly indirect responsibility for an intentional omission) require the above characterisation to be taken as a sufficient condition, rather than a necessary condition.  Some of Clarke’s remarks suggest that it is characteristic of doubly indirect responsibility for not φ-ing that one be unable to φ, but this cannot be right because some cases of singly indirect responsibility involve an agent who is unable to perform the action in question.

One can be doubly indirectly responsible for both an unintentional omission and an intentional omission.  An example of the former would be a drunk driver who causes injury by omitting to hit the brakes, and is responsible for causing that injury despite not having the ability to drive safely.  As an example of the latter, Clarke says the following:

Pam arranges to have herself given the post-hypnotic suggestion that she refuse to help her house-mates paint the porch. She is given the suggestion and intentionally omits to help out.  Suppose that the way the post-hypnotic suggestion works in this case is to render Pam unable to do otherwise than comply with it.  Pam is responsible for failing to help her housemates, even though she could not have helped them (1994, p.200).

The crucial point about this example, Clarke thinks, is that Pam’s responsibility for failing to help her housemates is not dependant on Pam’s responsibility for her inability.  It is instead dependent on (a) the fact that Pam forms the intention to refuse to help, and (b) the fact that (a) was the result of an earlier action for which Pam was responsible.  Neither of these things requires that Pam be responsible for her inability, so responsibility for the inability is not a condition on being responsible for such an omission.  But if it is possible to be responsible for an omission of φ-ing when one is not able to φ and one is not responsible for the inability to φ, then it follows that an inability is not always an undermining factor when assessing responsibility for omissions.  And this, Clarke contends, will falsify the alternative possibility requirement on responsibility for omissions.Another way of saying the same thing: the defender of the asymmetry thesis cannot appeal to a tracing condition in order to explain the need for alternatives in those cases where there is responsibility for not φ-ing despite an inability to φ.  Such a tracing condition, which would appeal to the agent’s prior ability to avoid the inability, is shown to be false by Clarke’s Pam example.  The ability to avoid an inability might be what grounds some cases of responsibility in the presence of inability, but it does not ground all such cases, and if it does not ground all such cases, then there is no alternative possibility condition on responsibility for omissions.

Clarke suggests the following principle which sums up these points about doubly indirect responsibility:

(DIRO) When an agent

(i) omits to perform an action of a certain type at a certain time, and

(ii) is unable to perform such an action at that time, and

(iii) is either

responsible for having that inability or,

in the case of an intentional omission, is responsible for forming the intention not to perform that action at that time,

then the agent’s inability does not excuse her from doubly indirect responsibility for the omission.

The two different ways of bearing doubly indirect responsibility are seen in the two disjuncts of clause (iii).  Principle (DIRO) provides a sufficient condition for when an agent’s inability will not undermine doubly indirect responsibility for an omission.

3. The asymmetry thesis falsified

Clarke’s two conclusions, one concerning singly indirect responsibility (SICRO), and the other concerning doubly indirect responsibility (DIRO), lead him to conclude that the asymmetry thesis is false.  Whether ability is required for responsibility for an omissions “turns out to be somewhat complex” (1994, p.205).  In the case of doubly indirect responsibility there is no ability condition, whilst in the case of singly indirect responsibility there is only need for a conditional ability.

So in the case of singly indirect responsibility, Clarke accepts that there are cases where an inability does absolve one of responsibility.  The proponents of the asymmetry thesis have focused on such cases (like Case 1, above) to make their point.  But Clarke contends that there are other kinds of case (like Case 2 above) which show that inabilities do not always undermine responsibility.  The move central to Clarke’s argument is the introduction of a Frankfurt-style case covering those times at which the agent might’ve decided to act.

This move is repeated in the treatment of cases of doubly indirect responsibility, such as drunk driver cases.  The obvious response from the defender of the asymmetry thesis in such cases is to appeal to tracing: yes, it is true that the drunk driver had no ability to hit the brakes in time, but he had a prior ability to avoid getting drunk.  By introducing a Frankfurt intervener to cover this prior period, Clarke apparently shows that inability (and hence, lacking alternatives) is not the crucial ingredient in assessing responsibility.


Clarke, Randolph (1994): Ability and responsibility for omissions. In Philosophical Studies 73 (2), pp. 195–208. Available online at http://www.springerlink.com/index/TWH03N26R61V6108.pdf.

Fischer, John Martin (1985): Responsibility and failure. In : Proceedings of the Aristotelian Society, vol. 86, pp. 251–270.

Fischer, John Martin; Ravizza, Mark (1991): Responsibility and inevitability. In Ethics 101 (2), pp. 258–278.

[1] From here onwards I’ll leave the non-deviance clause implicit.