The W-Defense Defended

1. The W-Defense

The W-defense begins as a response to Frankfurt’s attempt (and those modeled on it) to identify counterexamples to PAP. In a typical case of the sort Frankfurt and others envision, Jones decides on his own to break a promise. Unbeknownst to Jones, though, if he hadn’t decided on his own to break the promise, an evil neuroscientist would have compelled him to decide to break it, and there is nothing Jones could have done to stop the neuroscientist from doing so. Jones therefore couldn’t have avoided deciding to break his promise. But because he decided on his own to break it, without being aware of or compelled by the neuroscientist, Frankfurt and others contend that, given certain uncontroversial background assumptions about the case (e.g., that Jones is sane, knows right from wrong, was aware at the time that he was making a morally bad decision, etc.), Jones is blameworthy for deciding to break his promise, nonetheless.

Widerker responds to Frankfurt as follows:

since you, Frankfurt, wish to hold [Jones] 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 the 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. (2000: 191)

Widerker calls this the What-should-he-have-done defense or the W-defense, for short, and says that it “points to an important reason why it would be unreasonable to judge an agent morally blameworthy” if the agent couldn’t have avoided doing what he did (2000: 191).

The reason, in a nutshell, is this. If it wouldn’t have been reasonable to expect an agent not to do what he did, then the agent isn’t blameworthy for doing it. But it wouldn’t have been reasonable to expect an agent not to do what he did if he couldn’t have avoided doing it. Hence, an agent isn’t blameworthy for what he did if he couldn’t have avoided doing it.

This argument can be stated somewhat more precisely as follows:

1. A person is blameworthy for what he did only if it would have been reasonable to expect him not to do it (reasonable, that is, for someone who is morally competent and knows all the facts needed to make a sensible judgment about the matter).

2. If a person couldn’t have avoided doing what he did, it wouldn’t have been reasonable (for someone who is morally competent and knows all the facts needed to make a sensible judgment about the matter) to expect him not to do it.

3. Hence, PAP: if a person couldn’t have avoided doing what he did, then he isn’t blameworthy for doing it (from 1 and 2).²

The argument is valid. The question, then, is whether its premises are true. I’ll now argue that they are and that the main objections that have been leveled against them are unsuccessful.

2. Blameworthiness and Reasonable Expectations

Start with 1. As David Palmer points out, “a reason to think that (1) is true” is that “there are a wide variety of cases in which, although people could have done otherwise, they are intuitively not blameworthy for their wrong action,” and 1 “can explain, in a very natural way, why people are not blameworthy in these instances” (2013: 561).³ Palmer cites as examples certain cases in which people act on the basis of blameless ignorance and cases involving small children. However, the variety of cases in which 1 “can explain, in a very natural way, why people are not blameworthy” for their behavior is even broader than Palmer indicates.

Many of our ordinary and seemingly uncontroversial moral practices appear to presuppose 1. Just think about the sorts of considerations we take into account when attempting to determine whether we have been too hard on someone for something the person did or failed to do. One such consideration is whether we could reasonably have expected the person to behave any differently. If, having reflected on the matter, we arrive at the conclusion that we couldn’t reasonably have expected the person to behave any differently, that would suffice, it seems, to establish that we have indeed erred in blaming the person for her behavior.

To illustrate, suppose I blame my friend for something she did. Does she deserve it? If she has lived up to every reasonable expectation I might have regarding her behavior, it seems not. In that case, I seemingly have no grounds for blaming her for what she did, and any blame I direct at her in that regard would thus appear to be entirely undeserved.

The adjective ‘reasonable’ here is, of course, crucial. My friend might fail to live up to my normative expectations for her behavior and yet still not be an apt target of blame if my expectations are unreasonable. If they are, then the fact that my friend has failed to live up to those expectations clearly isn’t a legitimate basis for blaming or sanctioning her.

Suppose, to illustrate this last point, that I expect my friend to ignore all her other friends and to devote every moment of her life to me and to our friendship. In that case, my friend would clearly be correct to insist that she doesn’t deserve blame for not living up to my expectations regarding our friendship, for while she may have failed to live up to those expectations, that’s no basis for blaming or sanctioning her given that the expectations in question are entirely unreasonable. It’s only if my friend fails to live up to my reasonable expectations regarding her behavior that she might deserve blame from me for what she has done or failed to do.

Consider too the strategies we ourselves pursue when we want to persuade others that we aren’t blameworthy for something we did or failed to do. One of the most obvious strategies is to argue that, in behaving as we did, we didn’t fail to live up to any reasonable expectations. If we can establish this, by showing either that we haven’t failed to live up to others’ expectations regarding our behavior or that, if we have, the expectations in question are unreasonable, that would suffice, we take it, to demonstrate that we aren’t blameworthy for our behavior.

The preceding considerations strongly suggest that whether a person deserves blame for what he did depends at least in part on whether it would have been reasonable (for someone who is morally competent and knows all the facts needed to make a sensible judgment about the matter) to expect him not to do it. In short, they strongly suggest that 1 is true.

Frankfurt, however, isn’t convinced. He contends that cases like the one about Jones breaking his promise (the “Frankfurt cases” as they are commonly known) are counterexamples to both PAP and 1. Such cases, he claims, show that a person can be blameworthy “for having done something that he cannot reasonably be expected to have avoided doing” (2003: 344).

Is Frankfurt right about that? It depends, obviously, on whether Jones is in fact blameworthy for deciding to break his promise. Opponents of PAP like Frankfurt find it intuitively obvious that Jones is blameworthy for deciding to break his promise. W-defenders, however, contend that, because we couldn’t reasonably have expected Jones to do otherwise, Jones is not blameworthy for deciding to break the promise. In doing so, W-defenders either don’t share Frankfurt’s intuition about the case or, if they do, they think it should be rejected in favor of the principles underlying the W-defense. We have, then, a clash of judgment.

Which of these competing judgements we are initially inclined to endorse may depend in part on which aspects of the story we concentrate on. When we focus on the fact that, through no fault of his own, Jones couldn’t have avoided deciding to break his promise and, consequently, that we couldn’t reasonably have expected him not to decide to break the promise, the judgement of W-defenders can seem quite appealing.⁴ When we emphasize other features of the case, though, Frankfurt’s judgment may seem to be the intuitively more plausible one.

Consider in this connection Michael McKenna’s (2005; 2008) L-reply to the W-defense. (The “L” here stands for “look-at-what-he-has-done.”) Whereas the W-defense asks us to think about whether it would have been reasonable to expect an agent in Jones’s position to avoid doing what he did, the L-reply invites us to look at what the agent actually did and the reasons why he did it. Let’s accept the invitation. Jones, you’ll recall, decided to break his promise, without being coerced or compelled to break it, and he did this (we may suppose) for selfish reasons, despite knowing that he was making a morally bad decision. When we focus on these features of the case, it appears that Jones is blameworthy for what he did. It’s true that he couldn’t have avoided deciding to break the promise and that it would therefore have been unreasonable to expect him not to decide to break it. But, as Frankfurt points out, this can easily seem irrelevant to whether Jones is blameworthy for what he did when we bear in mind that “the circumstances that make it unreasonable to expect him to have done otherwise [viz., the evil neuroscientist] had nothing to do with leading him to do what he did” (2003: 344).

We seem, then, to find ourselves in the following situation. If we focus on the fact that we couldn’t have reasonably expected Jones not to decide to break his promise, it may appear as if Jones isn’t blameworthy for deciding to break it. But if we focus instead on the fact that Jones decided on his own to break his promise despite knowing that he was making a morally bad decision, it may appear as if Jones is blameworthy for what he did. We thus seem to be pulled in opposite directions, depending on which features of the case we attend most closely to.

Is there a way to resolve this tension in our intuitions? I believe there is. I’ll argue that we can accept the intuitions elicited by the W-defense as well as those elicited by the L-reply without having to jettison either 1 or PAP. This will involve invoking the well-known flicker of freedom strategy. There are different versions of the strategy, all of which rely in one way or another on the observation that there is something an agent like Jones could have done differently, even though he couldn’t have avoided deciding to break his promise. The version of the strategy I’ll focus on turns on the observation that Jones could have avoided deciding on his own to break the promise (where the locution “on his own” indicates that the decision resulted from Jones’s own, unaided action producing mechanisms, not from outside coercion or force).

Note that, although Jones couldn’t have avoided deciding to break his promise, he didn’t have to decide on his own to break it. He could have avoided deciding on his own to break the promise, in which case he would have been forced by the evil neuroscientist to decide to break it. Seizing on this feature of the case, some defenders of PAP have argued that what Jones is really blameworthy for isn’t deciding to break his promise (since he couldn’t have avoided doing so). What he is really blameworthy for, they say, is deciding on his own to break the promise. But since Jones could have avoided deciding on his own to break the promise, the fact that he is blameworthy for deciding on his own to break it poses no difficulties for PAP.⁵

Swenson and I refer to this version of the flicker strategy as the fine-grained response to Frankfurt cases “because it insists that a correct assessment of [such] cases . . . requires being very precise about what agents in those examples are blameworthy for” (Capes & Swenson 2017: 969). If the fine-grained response is correct, then we can consistently accommodate the intuitions elicited by both the W-defense and the L-reply. Here’s how. By focusing on the fact that Jones couldn’t have avoided deciding to break his promise and, consequently, that we couldn’t have reasonably expected him not to decide to break it, it becomes clear that Jones isn’t blameworthy for deciding to break the promise. But when we focus instead on what Jones actually did and why he did it, as the L-reply enjoins us to do, we see that there is something Jones did that we could reasonably have expected him not to do and for which he is arguably blameworthy, namely, deciding on his own to break his promise.⁶ Proponents of the L-reply are thus correct to insist that when we look carefully at what Jones did and why he did it, we’ll see that Jones is blameworthy. They are simply mistaken, says the fine-grained response, about what exactly Jones is blameworthy for.

The fine-grained response also enables us to address Frankfurt’s (2003) claim, echoed by McKenna (2005; 2008), that the W-defense yields counterintuitive results insofar as it requires us to absolve seemingly blameworthy agents like Jones of any moral responsibility for what they have done. By appealing to the fine-grained response, the W-defender can acknowledge that Jones is blameworthy for something and that he therefore isn’t off the moral hook, while also insisting, in keeping with the principles underlying the W-defense, that Jones isn’t blameworthy for anything he couldn’t reasonably have been expected not to do.⁷ In this way, we can accept the premises of the W-defense without absolving seemingly blameworthy agents like Jones.

I’ve just argued that the fine-grained response to Frankfurt cases can be used to reinforce the W-defense. Before proceeding, it’s worth noting that the W-defense can return the favor. A common criticism of the fine-grained response is that it’s too contrived, that it’s simply an ad hoc attempt to evade the intuitive force of Frankfurt cases. Robert Kane, for example, claims that the fine-grained response “artificially separates” blameworthiness for performing an action from blameworthiness for performing that action your own (1996: 41). And Michael Otsuka regards the response as “controversial, since it is arguable that one needs to draw too fine a distinction in order to maintain that Jones is blameworthy for [deciding on his own to break his promise] while at the same time denying that he is blameworthy for [deciding to break it]” (1998: 690).

The W-defense can be used to show that this objection to the fine-grained response is unfounded. For given the premises on which the W-defense turns, there is a principled reason why Jones is blameworthy for deciding on his own to break his promise but not blameworthy for deciding to break it, namely, that whereas we could reasonably have expected Jones not to decide on his own to break his promise, we couldn’t reasonably have expected him not to decide to break it. Proponents of the fine-grained response therefore aren’t artificially separating blameworthiness for acting from blameworthiness for acting on one’s own, as the separation is motivated by a pair of independently plausible principle about blameworthiness.

3. Reasonable Expectations, Avoidability, and the Maxim

Having motivated and defended premise 1 of the W-defense, I’ll now do the same for the argument’s second premise, which you’ll recall says that if a person couldn’t have avoided doing what he did, then it wouldn’t have been reasonable (for someone who is morally competent and knows all the facts needed to make a sensible judgment about the matter) to expect him not to do it. This follows from the more general principle that it’s unreasonable (. . .) to expect something of someone that the person can’t deliver, and this more general principle is quite plausible.

Example. Sandy and George are sitting on the beach when they notice a child drowning in the surf. There’s no one else around, no way to call for help, and so Sandy and George quickly surmise that the child will drown unless one of them jumps in and rescues her. Sandy is perfectly capable of doing that, whereas George, who is recovering from open heart surgery, isn’t.

Other things being equal, it would be reasonable to expect Sandy to jump in the water and rescue the child, whereas it wouldn’t be reasonable to expect George to do so. But why not? Why is it reasonable to expect Sandy to rescue the child but not George? The answer, it seems, is glaringly obvious; it’s because Sandy can save the child, whereas George can’t save her.

The plausibility of this answer can perhaps best be appreciated by imagining what would happen if George were to miraculously regain the ability to rescue the child (with no risk to his health). In that case, I should think it would be no less reasonable to expect George to rescue the child than it would be to expect Sandy to do so (assuming, of course, that George realizes that he has regained the ability to rescue the child), which suggests that the reason it wouldn’t be reasonable to expect him to save the child in the actual situation is because he can’t rescue her.

Another example. The next day, Sandy encounters twenty drowning children. She can’t rescue them all; there’s not enough time and no one else around to help. So, what should she do? That is, what would it be reasonable to expect her to do? Save as many of the drowning children as she can, obviously. But why isn’t it reasonable to expect her to save all twenty? Again, the answer seems obvious; it’s because she can’t save all twenty no matter how hard she tries.

A basic idea underlying 2, I take it, is that expectations are properly made only of those who can understand and comply with them. Another is that the most we can reasonably expect of people, whether morally or otherwise, is that they do their best, even if their best falls short of some ideal. To further illustrate this second idea, consider two more examples. Imagine you assign a speech to your public speaking class. You have a student with a diagnosed fluency disorder, who can’t help sometimes stuttering or repeating certain utterances. Knowing this about the student, would it not be unreasonable to expect the student to completely avoid stuttering or repeating himself during his speech? Surely the most you could reasonably expect of him is that he communicate as clearly as he can. Or suppose a friend promises to meet you for lunch but through no fault of her own gets stuck in traffic and thus can’t avoid being late. Knowing this, it seems entirely unreasonable of you to continue to insist that your friend arrive on time. The most you could reasonably expect of her, it seems, is that she get there as fast as she (safely) can.

The ideas underlying 2 are intuitively quite plausible, as the preceding examples illustrate. However, they may also seem to presuppose the popular yet controversial deontic maxim that ‘ought’ implies ‘can’ (the Maxim, for short). This, in turn, might seem to vindicate John Fischer’s claim that rejecting the Maxim “completely disarms the W-defense” (2006: 210).

In response to this worry, Widerker (2005) challenges Fischer’s assumption that the W-defense relies on the Maxim. He does so by appealing to a case that he claims is a counterexample to the Maxim but not to the premises of the W-defense. In it, “Jones borrows from Smith a rare copy of Principia Mathematica and promises to return it by October 10^th.” Unfortunately, on October 9^th, the book is stolen from Jones (through no fault of his own), making it impossible for him to return the book as promised. Widerker claims that, in this case, Jones “failed to fulfill his obligation to give the book back to Smith,” and, consequently, that “‘ought’ does not (always) imply ‘can’.” However, Widerker contends that these judgments are consistent with the premises of the W-defense, for although it was still true even after the book had been stolen that Jones ought to return the book to Smith, it would have been unreasonable to continue to expect Jones to return the book once it had been stolen (2005: 303–4).

It would be nice if Widerker’s response to Fischer were successful, for then the W-defense could be defended without having to bother about the Maxim. Unfortunately, though, there are at least two reasons to doubt that the response is successful. First, the example on which it’s based isn’t a clear counterexample to the Maxim, at least not if we understand the ‘ought’ at issue to be the ‘ought’ of all-things-considered moral requirement, for one might reasonably insist that, after the robbery, Jones is no longer morally required to return the book (even if there is some ideal sense of ‘ought’ in which Jones still ought to return it).⁸ Second, David Copp (2003: 271–72) has argued that we can’t plausibly accept 2 and the ideas that motivate it while also rejecting the Maxim. If he is right about that—and I’m inclined to think he is—then it must be admitted that the W-defense probably does presuppose (in the sense that it entails) the Maxim.

Fischer rejects the Maxim and claims that doing so “completely disarms the W-defense.” He therefore takes the modus tollens. But why not take the modus ponens instead? As we have seen, the ideas at the heart of the W-defense (such as the idea that we can’t reasonably expect a person to do something we know she can’t do) are plausible. So, if, as Fischer assumes, and as seems to be the case, those ideas entail the Maxim, why not conclude that the Maxim is true?

Fischer (2006: 204–6), responding to a similar argument of Copp’s, seems to reject the claim that it would be unreasonable to expect a person to do something we know she can’t do. In support of his position, he appeals to this example:

Imagine that a person—call him “Stanley”—deliberately keeps himself very still. He refrains, for some reason, from moving his body at all. . . . [S]uppose that here [sic] is someone with a powerful interest in having Stanley refrain from making any deliberate movements, who arranges things in such a way that Stanley will be stricken with general paralysis if he shows any inclination to move. Nonetheless, Stanley may keep himself still quite on his own altogether independently of this person’s schemes. Why should Stanley not be morally responsible for keeping still, in that case, just as much as if there had been nothing to prevent him from moving had he chosen to do so? (Fischer 2006: 205, quoted from Frankfurt 1994: 620–21)

Fischer contends that, although Stanley couldn’t have moved his body, he nevertheless “could be . . . blameworthy [for not moving it] should something morally important hang on his moving his body” (2006: 206). Given that Stanley is blameworthy for not moving his body only if it would have been reasonable to expect him to move it (a claim W-defenders are in no position to deny), it follows that it would have been reasonable to expect Stanley to move his body even though he couldn’t have moved it, in which case both the W-defense and Copp’s argument for the Maxim (both of which rely on the premise that it’s unreasonable to expect or require or demand a person to do something one knows the person can’t do) are unsound.

Fischer’s objection turns on the judgment that Stanley may well be blameworthy for not moving his body even though he couldn’t have moved it. I’ve argued elsewhere that attempts to motivate this judgment (or similar judgments about similar cases) are unsuccessful and that there is a strong case to be made against it.⁹ I won’t repeat those arguments here though. Instead, I’ll argue that the fine-grained approach described earlier can again be used to deflect an objection to the W-defense, this time by enabling W-defenders to account, in a way that’s consistent with both the W-defense and the Maxim, for the sense that Stanley may indeed be blameworthy.

Stanley couldn’t have moved his body even if he had been inclined to move it, for if he had shown any inclination to move, he would have been “stricken with general paralysis.” Still, there are various things Stanley could have done but didn’t do. For example, he could have seriously considered moving his body, and he could have tried to move his body. Moreover, these are things we could have reasonably expected Stanley to do (assuming that something of moral significance depended on whether Stanley moved his body). This suggests that what Stanley is really blameworthy for in this case, if he is blameworthy for anything, isn’t his failure to move his body (since he couldn’t have moved it). What he is really blameworthy for, it seems, is his failure to seriously consider moving his body and his failure to try to move it.

In this way, W-defenders can account for the intuitive sense that there is something for which Stanley may be blameworthy in this case, without having to grant Fischer’s claim that Stanley is blameworthy for not moving his body (even though he couldn’t have moved it) and thus without having to abandon either the Maxim or (the plausible ideas underlying) premise 2 of the W-defense. Taking the fine-grained approach to cases like this thus enables us to strike a nice balance between our intuitive judgments about such cases and certain highly plausible moral principles. Fischer’s position, by contrast, seems significantly less appealing in this respect, for by insisting that Stanley can be blameworthy for not doing something he couldn’t have done, Fischer must reject principles about both moral obligation and moral blameworthiness that, as we have seen, are seemingly well-motivated. For this reason, the fine-grained approach to these cases strikes me as much more appealing than the approach Fischer advocates.

4. An Irresolvable Tension

I’ve shown how W-defenders can address some of the most prominent criticisms of their position. Contrary to what some critics of the W-defense have supposed, the argument’s central premises are well-motivated, they don’t force upon us highly counterintuitive judgments about particular cases, and while they may rely upon (in the sense that they may entail) the ‘ought’ implies ‘can’ principle, this doesn’t provide the basis for a compelling objection to them. Once we recognize all this, it becomes clear that the W-defense yields both a forceful response to Frankfurt’s classic objection to PAP and a strong reason to think the principle is true.

Along the way, I also explained how the W-defense can be usefully combined with the well-known flicker of freedom strategy. I argued that these two responses to Frankfurt cases are mutually reinforcing positions that, when combined, show that proponents of PAP can reach a happy reflective equilibrium between plausible general principles about blameworthiness and our moral intuitions about particular (Frankfurt) cases. It’s worth noting in closing, however, that those who reject PAP, whether on the basis of Frankfurt cases or for some other reason, aren’t similarly well-positioned, contrary to what McKenna (2008) claims. I made a similar point above in response to Fischer’s objection to both the Maxim and the W-defense. But I think it’s worth emphasizing that the point applies to proponents of Frankfurt’s position more generally.

McKenna acknowledges the intuitive force of the W-defense. Indeed, in his estimation, the “W-defense cannot be directly refuted,” precisely because it taps “into an important moral intuition favoring a conception of free agency and moral responsibility in terms of alternative possibilities.” He further acknowledges that “Frankfurt’s conclusions do lead to an ‘irresolvable tension’ in our thinking about blame and its implications (regarding what an agent should have done instead),” insofar as it forces us to reject either 1 or 2 (2008: 784–85). However, McKenna thinks that the L-reply points to a similarly irresolvable tension in the thinking of W-defenders; it shows that, despite the intuitive appeal of the W-defense and the principles underlying it, the defense forces us to deny the strong intuition that an agent like Jones is blameworthy for what he did (even though he couldn’t have avoided doing it).¹⁰ Both views, McKenna concludes, leave us with residual intuitions that can’t be accommodated, and, to that extent, the two views are on par.

As we have seen, however, the tension in W-defenders’ thinking highlighted by the L-reply isn’t irresolvable. With the help of the fine-grained response, we can resolve the tension by pointing out that our clearest and most uncontroversial moral judgments about Frankfurt cases (e.g., that there is something for which the agent in such cases is blameworthy) are consistent with the principles underlying the W-defense. Accepting those principles therefore doesn’t require that we reject any clear and uncontroversial judgments about particular cases.

Rejecting PAP, by contrast, does require rejecting some plausible principles about moral obligation and blameworthiness. To consistently maintain the claim that an agent can be blameworthy for what he did even though he couldn’t have avoided doing it, critics of PAP must reject either 1 or 2 (and, if they reject 2, it seems they may have to reject the Maxim as well). Insofar as these principles are intuitively quite plausible, rejecting PAP does indeed lead to the irresolvable tension McKenna mentions. Thus, if our goal is to reach a happy reflective equilibrium between general principles and our moral intuitions about particular cases, it seems to me that we can better achieve that aim by combining the W-defense with the fine-grained approach than we can by embracing Frankfurt’s claims about agents like Jones.