she could have done otherwise. It makes the possibility of doing otherwise a necessary condition for responsibility for actions and omissions. This section explains and then defends a different principle, PAPB. Here is a rough statement of PAPB, which will need some explanation and refinement:
If a person is morally responsible for her behavior, then she could have behaved differently.
The key difference between PAP and PAPB is that PAPB does not require a counterfactual scenario in which the agent does anything at all. Instead, it only requires an alternate possible scenario in which the agent’s behavior is relevantly different. To bring out this difference between PAP and PAPB, let us define some terminology and then consider an illustration.
PAP relies on the possibility of doing otherwise. I understand the things that an agent does to include her voluntary actions and her voluntary omissions, but nothing else. Including omissions within the category of things a person does may seem strange or even like an oxymoron. For some purposes it would be a mistake not to distinguish between actions and omissions, but for my purposes here I see no serious problem with lumping them together.105 An omission may be thought of as a case where one allows something to happen or allows something to be the case though one has some opportunity to prevent or change it. Allowing is something that a person “does” in the sense that I am concerned with.
Actions and omissions are voluntary in the weak sense that I have in mind if they can be explained by an agent’s beliefs, desires and other
105 Note the significance of calling a voluntary omission something that the agent
does. For one thing, it means that I read PAP as stating a single condition on responsibility for both actions and omissions. Not everyone has supposed that this is so. Van Inwagen, for example, has suggested that PAP concerns only “performed acts,” not omissions. Van Inwagen (1983) pp.164-65. Incidentally, Van Inwagen is willing to concede PAP to the Frankfurt examples, but not his “Principle of Possible Action,” which states that a “person is morally responsible for failing to perform a given act only if he could have performed that act.” On my approach, the Principle of Possible Action is simply a more restricted version of PAP.
motivational states in the normal, familiar manner. This means that an act or omission may be voluntary although it is done only in response to a threat or compulsory legal order. A person who, fearing a death threat, does something that she finds loathsome still does that thing voluntarily on this account. Actions and omissions may also be voluntary even if they are brought about only through a mistaken belief. The person who shoots a friend, mistaking him for an enemy, does so voluntarily even though he may rightly point out that he did not mean to do it.106 It should also be noted that an omission may be voluntary not only if it is affirmatively motivated by the agent’s desires (e.g., the omission occurs because the agent is motivated to do something else), but also if it can be explained by the absence or weakness of her motivation to perform the relevant act (e.g., the omission occurs because, lacking concern about her duty, the agent lies around and does nothing at all).
Unlike PAP, PAPB draws our attention to behavior and to the possibility of behaving differently. I understand “behavior” to cover far more than voluntary actions and omissions. “Behavior” also includes any non-voluntary motions of an agent’s body, as well as the absence thereof. Twitches, snores, and movements made in sleep are a few familiar examples of behavior. Behavior also includes falling over as the result of a push, simply remaining motionless, and the rising and falling of one’s chest from breathing. The fact that PAPB is addressed not only to what an agent does, but also to the rest of
106 Thus, “voluntarily” is weaker than “intentionally.” I suspect that an action will be
voluntary under each of its descriptions if it is voluntary under any of them, whereas an action can be intentional under only those descriptions that the agent believes to apply.
her behavior does not mean that I suppose that a person is likely to be responsible for non-voluntary behavior. Indeed, PAPB itself indicates that typically, if not always, non-voluntary behavior is not the sort of behavior for which we could be blameworthy.
Now, to illustrate the difference between PAP and PAPB, imagine that Jane faces a choice. She must decide whether to submit her dissertation tomorrow or to delay, perhaps indefinitely (i.e., not submit tomorrow). (Submitting today is not an option because Jane is in the wilderness on a camping trip.) Normally, Jane will do one or the other: submit or delay. But it is also possible that before she does either of these things, Jane falls into a coma that lasts for weeks. If so, then submitting her dissertation will not be among the things she does tomorrow, but it also will not be the case that she
does otherwise. There is nothing that she does once she falls into the coma.
Of course, it would be correct to say that she does not submit her dissertation, but not if this is interpreted to mean that she does something other than submit her dissertation. There is an implicit “voluntarily” here, and we must be careful about where it goes. It is correct to say that Jane does not voluntarily submit her dissertation, but it would be wrong to say that Jane voluntarily does not submit her dissertation.
So, while Jane is on her camping trip, still healthy and conscious, at least three courses of events seem possible: she might voluntarily submit, she might voluntarily not submit, or there might be nothing that she does voluntarily. The first two confront her as choices. The third does not. It arises
in a case where her agency is suspended, so to speak. These three possibilities bring out the difference between PAP and PAPB. PAP focuses our attention on only the first two options. It suggests that if Jane is praiseworthy for submitting, then it must have been possible for her to delay (i.e., voluntarily not submit). PAPB, on the other hand, asks us to consider all three possibilities. According to PAPB, Jane may be praiseworthy as long as there is a possible scenario in which she fails to submit, whether this is because she voluntarily delays or because there is nothing she does at all.
The claim that there is a third possibility not treated as relevant by the normal principle of alternate possibilities might seem surprising. One might suggest that, in PAP, “could have done otherwise” is intended to cover the full range of possibilities, including the possibility of doing nothing voluntarily whatsoever. After all, there is a loose sense in which a person’s non-voluntary behavior can be included among the things she “does.” We might ask a person, “Do you realize what you did in your sleep last night?” (Of course, we can talk this way about inanimate objects as well: “Watch what this thing does.”) If one wishes to use “does” in this sense, I could concede that PAPB is merely a restatement of PAP that helps to bring out this third possibility. But I do not believe that this is how “could have done otherwise” has normally been understood or that it is faithful to the supporters of PAP. This is most clear on the traditional compatibilist interpretation of PAP. Traditional compatibilists claim that “could have done otherwise” means something like “would have done otherwise, if she had so chosen.” This interpretation makes
explicit the restriction of relevant alternatives to those in which the alternative is one that may be chosen, or in other words, to those in which the alternative involves voluntary behavior. This restriction is often made explicit by those who reject the traditional compatibilist interpretation as well. The customary rejoinder is that “could have done otherwise” means not only “would have done otherwise if she had so chosen” but in addition “. . . and nothing prevented her from so choosing.” Furthermore, as I will argue below, if people had understood PAP to be satisfied by the possibility of not doing anything voluntarily, they should not have supposed that Frankfurt’s case presents a counterexample. Therefore, I understand PAPB to be distinct from PAP.
Two further refinements to PAPB should be made. First, I should clarify the notion of behaving differently. PAPB requires behavior that is different in kind, rather than numerically different. So, PAPB is not necessarily satisfied by the possibility of a different token event. PAPB, like PAP, is motivated by the idea that a person cannot be considered blameworthy unless we could reasonably have expected better, nor praiseworthy unless we could have reasonably expected worse. Thus, PAPB requires not simply the possibility of a numerically different behavior event, an event that would be a wash, morally speaking, but behavior that is different in a morally significant way. If the agent is blameworthy for her behavior, PAPB requires the possibility of behavior for which she would not be blameworthy, and if the agent is praiseworthy, PAPB requires the possibility of behavior for which she would
not be praiseworthy.107 So, the requirement of different possible behavior may be stronger than is immediately apparent from the statement of the principle above.
Second, I will defend a compatibilist version of PAPB only. Compatibilist versions of PAP typically make use of some sort of volitional event. They require that the person would have done otherwise, if he had so
chosen or if he had so decided or if he had so willed. This is not the approach
I favor for PAPB. The very point of PAPB is to take account of a third possibility, on which the agent does not act voluntarily or as the result of a volitional event. Instead, PAPB should tie the possibility of behaving differently to certain relevant differences in motivation or moral concern. The conception of blameworthiness developed in Chapter Two shows that blameworthiness for behavior requires that the behavior reveals that the agent’s motivation to act in accordance with the relevant normative
107 It might be suggested that PAPB should only require the possibility of behavior for
which a person is less blameworthy (or for praiseworthiness, the possibility of less praiseworthy behavior). But consider the proposed counterfactual scenario, in which the agent behaves in a way for which she is less blameworthy. Applying PAPB to this counterfactual blameworthy person, there would have to be yet another alternative scenario in which she would be still less blameworthy, and so on ad infinitum or at least until we find an alternative in which she is not blameworthy at all. So, we should take PAPB to require, for blameworthiness, an alternative in which the person is not blameworthy for her behavior.
One might worry that this rules out the possibility of situations in which a person is blameworthy no matter what she tries to do. It seems as if there may be such situations, but we can make room for them if we suppose that what the agent is blameworthy for is getting into a situation in which she has only bad options. Her blameworthiness for creating or permitting that situation may transfer and make her blameworthy regardless of the fact that she chooses the best option once she is in it. If, on the other hand, she is not at fault for creating her dilemma, then I suspect that she will not be blameworthy for choosing what she reasonably concludes to be the least bad option.
considerations is weaker than we may reasonably expect of her.108 So, we should expect that if the blameworthy agent were more motivated to act in accordance with normative considerations, if she were as concerned and motivated as we might reasonably expect, then her behavior would be relevantly different. With some slight changes, the same is true of praiseworthiness. If a person is praiseworthy for behavior, it is because that behavior shows that her motivation to act in accordance with normative considerations is stronger than we might reasonably expect. If her motivation were not this strong, then her behavior would be relevantly different. Therefore, the compatibilist version of PAPB should tie the possibility of behaving differently to relevant changes in the agent’s motivation to act in accordance with reasons.
The compatibilist version of PAPB can be expressed most precisely as separate principles for blameworthiness and praiseworthiness:
108 The idea of an agent’s motivation to act in accordance with relevant normative
considerations is somewhat awkward and arguably over-intellectualized. I have been tempted to rephrase the idea in simpler and more familiar terms, such as an agent’s concern to do the right thing or the best thing. But I believe the phrase I have chosen allows some needed flexibility. I suspect that there are many circumstances in which there are relevant normative considerations that we should be concerned about and motivated by, but there is no right thing to do. I also suspect that there are situations in which we might praise, or withhold blame, based on a person’s concern to do some good, even if the person lacks a desire to do the right thing or the best thing. Perhaps I should give more to charity and perhaps if I were less selfish I would, but I might still deserve some credit for being motivated to give as much as I do. I do not know how much is the right amount to give, and in the end, that question does not drive my decisions about how much to give. I give the amount that I do because I want to do some good, but I am not so strongly motivated to do the best thing that I possibly can. I recognize that there are reasons to give, and I act on them, but I choose not to be concerned about doing the best thing I could do. I do not know whether I should be considered praiseworthy or blameworthy for my giving, but in evaluating my blameworthiness or praiseworthiness, it should be relevant that I am motivated by normative considerations, even if I am not always motivated by a concern to do the right thing or the best thing.
If a person is blameworthy for behavior, then if she were more motivated to act in accordance with normative considerations, she would have behaved in some other way for which she would not be blameworthy.
If a person is praiseworthy for behavior, then if she had been less motivated to act in accordance with normative considerations, she would have behaved in some other way for which she would not be praiseworthy.
We should now be in a position to see what can be said for this proposal. B. The Argument for PAPB:
Now let us consider the basis for believing PAPB states a necessary condition for responsibility. An argument for PAPB, as applied to blameworthiness, can be made out if we accept the following premise:
If A is blameworthy for his behavior, then, during the relevant period prior to his behavior, A’s motivation to act in accordance with normative considerations was weaker than we may reasonably expect of him.
I will have more to say about this premise below, but let me first say how it can be used to support PAPB. If we accept it, then a second premise naturally follows through contraposition:
If, during the relevant period prior to his behavior, A’s motivation to act in accordance with normative considerations had been as strong as we could reasonably expect, then A would not be blameworthy for his behavior.
From these premises, I infer the following:
If A is blameworthy for his behavior, then if, during the relevant period prior to his behavior, A’s motivation to act in accordance with normative considerations had been as strong as we could
reasonably expect, then A would not be blameworthy for his behavior.
This is enough to establish PAPB as applied to blameworthiness, because, as I shall argue below, if in one scenario A would be blameworthy and in the second scenario he would not be, then his behavior in the two scenarios is different in kind. In the two scenarios, I shall argue, A behaves in a different way. With the necessary changes, the same premises could be used to establish PAPB as applied to praiseworthiness.
The argument for PAPB only gets going if we accept my first premise. I take that premise to follow from the conception of blameworthiness and praiseworthiness defended in Chapter Two. There, I argued that blameworthiness for behavior requires that the behavior in question reveals an agent’s weak concern for normative considerations, and that praiseworthiness for behavior requires that the behavior in question reveals an agent’s strong concern for normative considerations. The strength of an agent’s motivation to act in accordance with normative considerations is revealed by her behavior only if it figures among the causes or explanations of that behavior. So, if a person is blameworthy for her behavior occurring at a particular time, then over some relevant period prior to that behavior, her concern or motivation to act on moral reasons was not as strong as we might reasonably expect.
The argument follows naturally from this idea. If we take an actually blameworthy agent, and then imagine her to be strongly and consistently motivated by normative reasons, then her behavior could no longer be of a kind for which she would be blameworthy. In short, altering the features of her
will that are required to make her behavior blameworthy is sufficient to ensure that her behavior would be relevantly different.109 Similarly, praiseworthiness for behavior depends on behavior that reveals the agent’s strong concern for reasons. If we take a praiseworthy agent and imagine her to be less concerned for reasons, her behavior could no longer be caused or explained by the strength of her concern for reasons. Thus her behavior would be different. It would no longer be the sort for which she could be praiseworthy. In either case, PAPB follows from the fact that moral responsibility for behavior depends upon the strength of the agent’s motivation to act in accordance with normative considerations.
Contrast the foregoing rationale for PAPB from one that might be offered in support of PAP. It is sometimes supposed that alternate possibilities requirements get their support from an idea that blame must be avoidable (i.e., that it would be unfair to blame a person unless she could have done something to avoid blame). That is not the basis of my argument for PAPB. I am not certain what to make of the idea that blame must be avoidable as a