BURDEN OF PROOF

Shafer-Landau offers a number of challenges to Feinberg's antipaternalism. Before examining particular ones, however, it will be helpful to first address his general methodology for it contains one of the most problematic aspects of Shafer-Landau's critique of Feinberg's antipaternalism. Shafer-Landau sets up the debate as if the burden of proof rests on the antipaternalist, with the balancing strategy winning by default just in case Feinberg fails to provide “conclusive” arguments for his view. For example, Shafer-Landau writes of the liberal needing arguments to “secure” particular kinds of conduct from criminalization (as if criminalization is legitimate unless and until a conclusive case can be made for liberty) and more generally of the absence of a “decisive argument against hard paternalism.”⁶

Moreover, Shafer-Landau maintains that Feinberg himself accepts this burden of proof when, according to Shafer-Landau, Feinberg admits that “the burden is on the liberal to show that a balancing strategy must fail.”⁸

But Shafer-Landau's methodology can be challenged in at least three ways. First, it is misleading to view Feinberg as accepting a true burden of proof, that is, a burden by which the opposing side is automatically deemed true just in case Feinberg fails to provide “conclusive” arguments for his view or, to quote Shafer-Landau, fails to provide arguments that are not “vulnerable” in one way or another.¹⁰

Nor is Feinberg conceding that he alone needs arguments, even if arguments for the other side are easier to muster. To use an analogy, it is harder to establish that crows never have yellow legs than it is to establish that at least some crows do not have yellow legs. But the failure conclusively to prove that crows never have yellow legs does not thereby establish that some crows do have yellow legs. Instead, in the absence of conclusive arguments, the rational thing to do is to look at the evidence for both sides. Thus proponents of the view that some crows have yellow legs (and, by analogy, the view that paternalistic reasons sometimes have weight on the scales) need to give us reasons for their view as well. Until they do, it would be hasty to presume that the balancing strategy is correct just because the arguments for Feinberg's view are “vulnerable” in some way. It may be that the arguments for the balancing strategy are more vulnerable still. I return to this comparison below. My main point here is that Feinberg has not accepted the burden of proof in the way Shafer-Landau implies.

Second, let us suppose (contrary to fact, in my opinion) that this is issue is one in which one side or the other must carry the burden of proof, with the other side getting the potential luxury of winning by default. Shafer-Landau might be read as claiming that even if Feinberg has not actually accepted the burden of proof, he is stuck with the burden nonetheless, because the balancing strategy is intuitively much more plausible as a default position. But how could that be (at least for anyone who admits to having liberal tendencies, as does Shafer-Landau)? Consider the presupposition of each view.

1. Only other-regarding conduct is subject to the coercive power of the state unless and until a convincing argument can be made to the contrary.
2. All conduct, regardless of whether it is primarily self-regarding or other-regarding, is subject to the coercive power of the state unless and until a convincing argument can be made to the contrary

The first alternative captures a basic claim of Feinberg's antipaternalism (with the “unless and until” clause added to cast the view as a default position). It should be attractive to anyone with liberal tendencies insofar as it is consistent with a variety of overlapping liberal concerns. For example, it is consistent with recognizing autonomy as a fundamental (as opposed to derivative) value and with the view that the state should leave people free to choose and pursue their own conception of the good. It is also consistent with general consent theories about the origin of the state and political obligation (though I am not claiming that such theories are problem-free). It allows us to explain the government's coercive authority over our conduct that risks harm to others in terms of our consent to that authority and to explain the lack of that authority over conduct that affects only ourselves (or others with their consent) in terms of our unwillingness to concede that power to the state. (After all, it would laughable to imagine our hypothetical original consenters saying “In order to gain the benefits of a legal society, I consent to the authority of the state to govern my acts that threaten others and to punish in my name those who have harmed me. And, what the heck, I'll consent to having the government use coercion to substitute its perception of what is good for me for my perception of what is good for me even when all agree that my choices are informed, voluntary, and self-regarding.”) Thus Feinberg's view is consistent with a true presumption in favor of liberty and with viewing the state as a creation of the people and the people as free to live their own lives until a convincing argument can be made to the contrary.

In contrast, the second alternative captures the presumptions of the balancing strategy. It asserts that all conduct is initially on the table for coercive interference by the state and that the state balances competing concerns in order to decide which conduct a person should be free to pursue without the threat of punishment and which conduct the state should coercively prohibit. Such a view may be attractive (at least initially) to anyone with consequentialist leanings insofar as it declines to draw a distinction between self-imposed and other-imposed harm unless and until an argument for such a distinction can be made. The view might also appear to give weight to the value of personal autonomy because the state seems to grant that people should be allowed to pursue some self-regarding activities without coercive micromanagement by the state, even when those activities threaten mild to moderate self-harm.

Indeed, the preceeding view is the one that Shafer-Landau seems to support when he says that a plausible balancing strategy would begin with a concession to the liberal entailing that the “protection of an agent's voluntary self-regarding choice is always a good and relevant reason against” coercive state intervention.¹¹

A proponent of the balancing strategy may object to this characterization and assert that a good balancing strategy really does value personal autonomy for its own sake (and not just as a means to advance overall welfare); but it also values the prevention of harm for its own sake and it tries to strike a balance between the two. Yet even this view is challenged (and I think fatally challenged insofar as we are asking which view is more plausible as a default position) by the fact that the view cannot explain where the state got its authority to use coercion to protect people from themselves. That is, we can understand why it wants to use the coercion (i.e., the value of preventing harm for its own sake) but not where it got the authority to do so. (Analogously, I can imagine my neighbor wanting to use force to stop me from windsurfing on an overly gusty day, but that wanting does not give my neighbor the authority to use the force.) The mere fact that the state concerns itself with the prevention of harm in other areas does not entail that it is authorized to do so here, especially if the authority to use coercion in other areas came from the consent of the governed.

To put it in another way, recognizing a conflict between two values is one thing. But imposing one's view as to how that conflict should be resolved is quite another. When the state tries to strike a balance between the two values and impose its perception of the proper resolution, it is assuming an authority over the individual that is, in the absence of the individual's consent to that authority, inconsistent with true respect for the individual's autonomy. The state is saying “I have legitimate power over you” not because you have given it to me but because that is the default position. To anyone with even mild liberal leanings, such a view should be counterintuitive at best, if not repugnant. It perceives the authority of the state as somehow superior to that of the people and the people as having no true presumption in favor of liberty but instead only zones of pseudo-liberty when and where the state wants to grant them.

To sum up the preceding point, if we had to view the issue as one in which one side—either Feinberg's antipaternalism or the balancing strategy—had the burden of proof and the other side had the potential luxury of winning by default, then it seems to me that the presumptively favored view should be Feinberg's antipaternalism. It is clearly more in keeping with our considered judgments and a plausible view about the relation between persons and the state than is the balancing strategy. And if that is correct, then it is the proponents of the balancing strategy that have the burden of proof. They need to prove that the state has the authority to use coercion to substitute its perception of a person's own good for that person's own perception in cases in which the state agrees that the person's perception is sufficiently voluntary and informed and the conduct affects no one but the agent (or others with their consent). I cannot imagine what such an argument would look like, at least not one that is not “vulnerable” in some way. But without such an argument, paternalistic considerations about preventing self-harm never get on the scales to begin with. And if they are not on the scales to begin with, then it is misleading to characterize Feinberg as claiming that “one kind of reason invariably takes normative priority over another.” Without an argument getting them on the scales, they are not relevant reasons at all.

However, there is no need to approach the issue as if one side or the other has the burden of producing conclusive arguments. (This is the third flaw in the methodology.) Instead, the issue is well suited to be one for which evidence for each side is given, the drawbacks and vulnerabilities of each side are articulated, and the evaluator is left to decide which view, all things considered, is more plausible. This methodology is more in keeping with Feinberg's writings than the presumption that one side or the other begins with the burden of proof. Indeed, as Shafer-Landau notes, Feinberg is not a first principles or “bottoms up” kind of reasoner. Instead, Feinberg examines the options, painstakingly working out the commitments and implications of each, showing which considered judgments they can account for and which they cannot and how they fit with our deepest commitments, and ultimately leaving it to the reader to decide whether she too favors the view that Feinberg does. He writes:

Moreover, Feinberg is aware that his favored view is not perfect. Sometimes he elects to abandon an initially plausible intuition, in a reflective equilibrium sort of way, and on occasion he is willing to accept a sort of ad hoc adjustment in order to account for a deep-seated judgment. (Consider his views about criminalizing failures to provide easy rescues or limiting free speech in Skokie-type cases, for example.) Thus, not only is it not necessary but it is also unfair to cast this issue as if a mere weakness or vulnerability in Feinberg's arguments loses the case for him. The real issue is whether his view is more plausible than the alternative—whether, in Feinberg's words, it “resonates” more.

PRINCIPLED LINES VERSUS BRIGHT LINES

Two of the reasons why Feinberg thinks that his view is more plausible than the balancing strategy are: (1) it is better able to capture the vast majority of our intuitive judgments (and even more of our judgments after careful reflection); and (2) unlike the balancing strategy, it permits a plausible principled line to be drawn between conduct that is and conduct that is not legitimately subject to the coercive power of the state. Let us begin with the second point. Shafer-Landau tries to capture this point in the following argument, which he refers to as Feinberg's central or “Master” argument against hard legal paternalism:

1. There must be a sharp, principled boundary that distinguishes the self-regarding choices that are immune from legal interference from those that are not.
2. There are only two plausible candidate boundaries: (i) the state may not intervene to prevent voluntary self-regarding choices; or (ii) the state may intervene to prevent such choices, provided that it thereby protects a citizen's true interests and thus her good.
3. (ii) is not a viable option.
4. Therefore self-regarding choices are immune from legal interference if they are voluntary.

It is worthwhile to remember that Shafer-Landau constructed this argument on Feinberg's behalf. Hence we might legitimately question whether the particular wording provides the best characterization of Feinberg's view. For example, the current wording of premise 1 focuses our attention on the concept of voluntariness as a way to distinguish self-regarding choices that are subject to coercive interference from self-regarding choices that are not. In so doing, it seems to presuppose that self-regarding conduct is subject to coercive interference until we can, on principle, carve out a portion that is not. An alternative wording might have captured the same need for a principled boundary but instead focused our attention on autonomy, consent, and the justification for coercive interference by the state. For example, it might have asked for a principled line that distinguishes conduct that is legitimately the business of the criminal law from conduct that is not. Discussions of voluntariness would later come up, but they would not be the initial focus.

The wording of clause (ii) of the second premise might also be questioned. It seems to paint paternalism with an excessively rosy glow by suggesting that paternalistic interferences would only be allowed when the state knows a citizen's “true interests” better than the citizen does and just wants to help her achieve those interests, even though the state grants that the choice that is being coercively blocked is the voluntary, informed choice of a competent person regarding her own life. A less flattering wording of clause (ii) might have referred to the state's right to use the threat of punishment to substitute its perception of a citizen's good for that citizen's own perception of her good when the state believes that a sufficient amount of harm can be prevented by doing so (or believes that doing so is in some other way “really important”). Such a wording would not mask the fallibility of the state in thinking that it knows a person's true good better than the citizen herself does, nor mask the paternalist's general willingness to substitute his perception of the good life for the citizen's own perception. However, there is no space here to evaluate all the nuanced versions of the argument that Shafer-Landau has constructed on Feinberg's behalf. So, having noted that the given version may not be the ideal reflection of Feinberg's concerns, I will focus on the argument as stated and on Shafer-Landau's assessment of it.

The details of Shafer-Landau's evaluation of the master argument cannot be recounted here without quoting pages and pages of text. But they can be summarized as follows. Shafer-Landau concedes that the paternalistic viewpoint embedded in clause (ii) of premise 2 cannot draw a principled line that distinguishes self-regarding conduct that is immune from coercive interference by the state from self-regarding conduct that is not immune—in principle it is all on the table. He then questions whether the absence of a principled line shows that clause (ii) of premise 2 (i.e., the balancing strategy) is untenable. If it does, then premise 3 is correct and the balancing strategy is not a viable option. He does this by challenging the truth of premise (1) in 2 ways:

In challenging the first supporting claim, Shafer-Landau first assumes that personal sovereignty includes property rights (he says that it is implausible to think that it does not) and he notes that Feinberg allows that property rights may sometimes be justifiably infringed, as in the case of the stranded hiker who must break into an unoccupied house in order to save his own life. Thus, according to Shafer-Landau, claim (i) is in doubt because the mere fact that something falls within the bounds of sovereignty does not show that it is immune from all kinds of interference.

But this argument is far from convincing. Its three underlying assumptions—that sovereignty is best defined as a constellation of rights, that sovereign rights would include the sort of trivial property rights covered in the hiker case, and that recognizing that some kinds of rights can be permissibly infringed commits one to holding that any kind of right can be permissibly infringed (see premise 4 of Shafer-Landau's reductio argument)—are clearly in need of defense. Further, even with those defenses, the analogy between paternalism and the hiker case is too weak to do any real work. In the hiker case, the property owner was permissibly wronged, all things considered: Her rights were violated, and compensation/restitution is due (e.g., for the food that was consumed and the lock that was broken) to minimize the extent of the wrong. But surely paternalists are not claiming that paternalistic interferences are wrongs or violations that are permissible, all things considered, nor that the victim deserves compensation or restitution. Instead, paternalism views the interferences as justified in themselves, not merely as permissible, all things considered.

More importantly, our focus is not on interference per se but on coercive interference aimed at protecting a person from her own, voluntarily assumed risks. There is a huge difference between, on the one hand, violating a nation's sovereignty by flying though its airspace in order to get to another nation in time to render emergency aid that was requested by that other nation, and, on the other hand, interfering with a nation's sovereignty by flying (or threatening to fly) into its airspace in order to drop a bomb on that nation's television transmission tower because the interfering nation thinks that the receiving nation is harming itself by allowing its citizens to watch too much TV. The first case is analogous to the kind of noncoercive violation of personal sovereignty that is found in the case of the hiker (and even then, we have to make the assumption that personal sovereignty includes trivial property rights). The second case involves the kind of coercive thwarting of sovereignty that is part of paternalism. Were Feinberg pressed on the analogy between sovereignty and the hiker case, he might respond that recognizing that sovereignty might, in emergency situations, be permissibly violated, all things considered (which is a view that grants that sovereignty existed and that a wrong occurred), tells us nothing about whether coercive interferences with sovereignty can be justified (which, if they can, implies that no wrong or violation occurred because the sovereignty was limited, suspended, or nonexistent in that particular case). So again, unless paternalists are willing to view their coercive interferences as true violations or justified wrongs, Shafer-Landau's first critique has hit a dead end.

Let us turn to Shafer-Landau's second critique of the master argument. Shafer-Landau challenges Feinberg's claim that we need a sharp, principled line that distinguishes conduct that is subject to coercive interference by the state from conduct that is not, by noting that Feinberg accepts imprecision in other areas of legal reasoning and does not view that imprecision as fatally arbitrary. For example, he notes that Feinberg is well aware that the distinctions between sufficiently voluntary and not sufficiently voluntary, and between reasonable and unreasonable, are imprecise and that they require us to balance a number of incommensurable factors to determine their scope. And since we do not view this imprecision as fatally arbitrary, we “may well wonder whether the absence of such precision is really the threat to the balancing strategy that Feinberg takes it to be.”¹⁴

The problem with this criticism is that it fails to appreciate the difference between a sharp or principled boundary in a theory and a sharp or bright line in practice. Feinberg is calling for the former, but Shafer-Landau's criticism focuses on the latter. To use an analogy, consider the claim that insane people cannot truly be punished. Such people might have an evil imposed on them, but it would not be legitimate punishment if they were insane and thus not capable of being responsible for their actions. In this case, the difference between sanity and insanity provides a principled boundary between who is a candidate for legal punishment and who is not a candidate. It is not just a boundary, as it might have been if we had said that only people over six feet tall are candidates for punishment. It is a principled boundary because it is based on an appropriate reason or rationale for drawing a line there and not somewhere else.

But that principled boundary between sanity and insanity, which draws a sharp line within our theory about legitimate punishment, is not the same thing as having a sharp or bright line in practice that tells us who counts as insane and who as sane. That line might be fuzzy (though there will be a wealth of clear cases on each side of the gray or fuzzy area). Perhaps the line is fuzzy because we disagree about some of the proper conditions of insanity or about the proper tests for determining whether those conditions are met. Or we might even agree about the conditions and tests but find that the conditions themselves are incommensurable in a way that precludes a bright line in practice. But the absence of a bright line in practice (which we can address in the normal ways in which we address line-drawing problems in public policy and law)¹⁵

Thus Feinberg is not being inconsistent when, on the one hand, he calls for a principled line within our theory about the moral legitimacy of coercive state interference that tells us what is and what is not legitimately the state's business and, on the other hand, accepts some imprecision, or the absence of a bright line, in the practical, real-world application of that theory. His antipaternalism initially draws the theoretical line between self and other-regarding conduct and grounds that line in our views about autonomy, consent, and our considered judgments. It then further divides the realm of self-regarding conduct into (substantially) voluntary and nonvoluntary conduct. This latter distinction does not undercut or replace the former one but is instead a narrowing of the realm self-determination that is necessitated by the values, concepts, and/or commitments that ground the first distinction. In other words, only conduct that is substantially voluntary (where the concept of voluntariness is fleshed out to include, among other things, knowledge or information about the risks and benefits of the various options and the capacity to balance those risks and benefits in light of one's desires, goals, values, and commitments) could reasonably count as the self-governing or sovereign choice of a competent person.

Moreover, it is not unreasonable to imagine people consenting, at least hypothetically, to interference by the state that is aimed at protecting them from risks they have not voluntarily chosen to run while still objecting to interference within the realm of choices that are agreed by all to be voluntary. And the fact that there may be some imprecision in the practical determination of which actions count as sufficiently voluntary and which do not does not undercut the usefulness of, nor the rationale for, the distinction itself. As Feinberg notes, many useful distinctions lack bright lines in practice, such as the distinction between day and night.

In contrast, Feinberg disfavors the balancing strategy because (among other things) it is unable to draw a plausible principled line that distinguishes between conduct that is and conduct that is not subject to the coercive power of the state.¹⁶

In summary, Feinberg's call for a principled line that distinguishes conduct that is from conduct that is not subject to the coercive power of the state is not just a worry about avoiding a fuzzy or, in that sense, arbitrary line. It is a call for a theory that embeds a message about the status of persons as autonomous agents that is missing in a view in which all one's acts are presumptively on the table and the state gets to decide which decisions a person will be free to make on her own and which ones need to conform to the state's view of what is best for her.

CONSIDERED JUDGMENTS AND THE APPARENT NEED FOR PATERNALISM

Part of the plausibility of the principled line that Feinberg favors lies in its ability to capture our considered moral judgments about particular cases. Shafer-Landau explains this general point in the following argument which he constructs on Feinberg's behalf and calls the “considered opinion argument”:

1. If we carefully reflect on the prospect of legal interference with our own voluntary self-regarding choices, then we will find ourselves opposed to such interference.
2. If we find ourselves opposed to such interference, then we should oppose it not only for ourselves but also for others.
3. Therefore if we carefully reflect on the prospect of legal interference with our own voluntary self-regarding choices, then we should oppose it not only for ourselves but also for others.
4. If a view about personal legal sovereignty is indicative of or an implication of the considered opinions of the citizenry and is such that it would violate no rights if widely implemented, then it should be enshrined into law.
5. Antipaternalism meets these conditions.
6. Therefore antipaternalism should be enshrined into law.¹⁸

   Shafer-Landau, supra note 3, at 182.

Shafer-Landau criticizes this argument in two ways. First, he notes that it is not likely to convince those who do not share the pretheoretic intuitions that Feinberg supposes. Shafer-Landau is obviously right on this point, but I will not discuss it further because the methodology in question is not intended to convince everyone. Second, Shafer-Landau claims that the argument does not deliver the “verdict that Feinberg seeks” (even for those who share the intuitions) because we also have intuitions that seem to be “best explained” by an endorsement of hard paternalism.¹⁹

Of course, these cases really are tough cases for the antipaternalist only if both of the following conditions are met: (1) we want coercively to prohibit the conduct in question; and (2) we can justify such coercion only on paternalistic grounds, that is, only on the need to protect the agent from her own foolish choices. When viewed this way, some of the purportedly tough cases are not tough at all.

Consider the issue of bigamy. If we really wanted to prohibit bigamy on paternalistic grounds, would we not also want to prohibit conduct that acts like bigamy but without the legal certification, for example, living with and/or having sex with and/or having children with one or more people who are already legally married? Would it not be even more self-harmful to act as if one has the benefits of being a second spouse without actually having them? I think so. But we do not prohibit that conduct, which indicates that our objection to bigamy is not paternalistic (we are not protecting that poor second spouse from his or her foolish choices) but something else. Perhaps it is moralistic concern or perhaps it is an efficiency concern. That is, for reasons of efficiency—record-keeping and the like—we say that one can claim the legal benefits of marriage with only one person at a time. We then also say that it is criminal to try to deceive the government or another person into believing that X gets the legal benefits of marriage to Y when Z already has those benefits. But that coercive interference is not paternalistic, because deception undermines voluntariness.

Restrictions on dueling are also open to a nonpaternalistic justification. Feinberg cites and then modifies the following argument from Richard Arneson:

Feinberg adds that this argument works even if some members of the society do not share the preferences of the majority. “When most of the people subject to a coercive rule approve of the rule, and it is legislated (interpreted, applied by courts, defended in argument, understood to function) for their sakes, and not for the purpose of imposing safety or prudence on the unwilling minority (“against their will”), then the rationale of the rule is not paternalistic.”²¹

Even the case of voluntary slavery contracts, to which Shafer-Landau devotes much attention, need not require paternalistic coercion in order to capture the bulk of our intuitions.²²

Further, we do not even make it a crime to try to sign a slavery contract. Instead, we only refuse to recognize the agreement as a binding, legal contract and we will not expend our resources to enforce it if one party changes her mind. (Consider, too, that many states permit paid surrogate-motherhood arrangements but refuse to resolve contested cases within contract law.) But that does not require an endorsement of hard paternalism. A liberal society can refuse to aid, support, or promote activities that it regards as unseemly and self-harmful without criminalizing them and, as I discuss in the next section, it can prohibit other people's involvement in them when the other people are not in a position to determine whether the choice was truly sufficiently voluntary and a mistake cannot be corrected. (This follows in part from the widely recognized view that the greater the risk of harm in an activity, the greater the evidence must be that the harm was voluntarily accepted. But this is an evidentiary standard, not a theoretical one about “more” consent. Since people are presumed to not want to harm themselves, we want more evidence of the exceptions to make sure they really are exceptions.)

Of course, case-by-case discussions such as the ones above cannot show that every apparently tough case can be reasoned away and shown to be either nonpaternalistic or not in need of coercive interference. Sometimes we might be willing to discount the intuitions themselves if the overall cost of satisfying them is the acceptance of an unprincipled balancing strategy and its greater number of counterintuitive implications. For example, many antipaternalists seem willing to forgo criminal statutes requiring seat belt use by adults if they cannot find nonpaternalistic justifications for the laws. Still, it is possible that that there are other cases that are so compelling that we cannot simply discount our judgments, and any theory that cannot account for them is in that regard problematic. I think that this is what Shafer-Landau had in mind with respect to the issue of voluntary limb amputation. So let us turn to that in the final section.

PATERNALISM AND SELF-MUTILATION

In the beginning of his essay, Shafer-Landau discusses the case of a person who wants to amputate a healthy limb. We are assumed to want to prohibit this behavior but cannot clearly justify the prohibition on soft-paternalist grounds because, as Shafer-Landau convincingly argues, we cannot plausibly conclude that every possible case of such behavior will fail an assessment of voluntariness. Thus we appear to have a tough case for the antipaternalist.

But do we really want to criminalize this behavior, and if so, can we do so without advocating hard paternalism? To answer these questions, we need first to get clear on what we are talking about. The issue is not whether we want to encourage, promote, or subsidize significant alterations of the body, as in amputation of limbs, but whether we want to make it criminal to make such alterations. Second, as we assess our intuitions, it is important that we do not unknowingly slip into thinking about cases in which the person who wants a limb amputated is mentally disturbed or irrational, or the choice is otherwise not sufficiently voluntary. So let us not assume that the person has no reason for wanting a limb cut off (would that not be a sign that the choice was not an autonomous desire?) but rather that the reason she has is not good enough in our view. (Note that no one seriously objects to all amputations, which implies that some reasons are good enough.) For example, suppose a woman says she wants an amputation because looking at her hand, which she used to steal candy as a child, grossly repulses her to the point of making it difficult to get on with her life. It is also important that we not think of cases in which the person would become a ward of the state, because that burden on us (which differs from the consideration of mere psychic costs) may cloud our judgment.

So, imagine a wealthy, forty-year-old recluse with medical training who wants to cut off his thumb, his hand, or even his arm. Perhaps he wants to look like his dead father, who had lost a hand in a farming accident. The recluse has been through lots of counseling, and there are no grounds for thinking him incompetent or his choice less than sufficiently voluntary. With his medical training he can do the act himself, and with his money and hermit-like lifestyle, there is no worry that he will ever call on society for help. So he cuts his hand off and stitches himself up.

Do we really want to say that what he did was a crime and that he should be punished? My intuitions say no. And if I am right, then we do not have a compelling test case for Feinberg's antipaternalism. Further, if one is thinking “Well, cutting off a hand shouldn't be criminal, but cutting off a whole arm should be,” then we have to have a reason for the difference. But what could that be? We cannot say that the one should be criminalized because it reduces a person's options or capabilities more than the other. After all, we do not punish people who have had plastic surgery to look like a lion or a reptile, even though such changes affect a person's job prospects arguably as much or more than missing an arm. Nor do we make it a crime to have oneself sterilized, even though that removes the capacity to parent a child biologically.

It might be argued that we have to view these acts as punishable because that is the only way we can justifiably prohibit other people from developing and/or advertising a business in the practice. But this view is flawed because it forgets that there can be grounds for prohibiting commerce in an activity that do not necessarily imply that the activity itself should be prohibited. (And, contrary to Shafer-Landau, this shows that it is false to claim that whatever applies to the single-party case applies to the two-party case mutatis mutandis.) For example, a few years ago in Chicago, a bikini car wash (in which bikini-clad buxom women would lovingly soap up the car and themselves) was not allowed to open, but no one had to argue that washing a car in a bikini was a criminal act in order to do so. Similarly, there can be grounds for prohibiting physicians from participating in assisted suicide that do not commit one to holding that suicide itself should be criminal as well. For example, one might argue (though I am not endorsing the argument) that participation in assisted suicide can negatively affect a physician's general outlook on the aim of her practice. But, it might be asked, what if someone is willing to assist in an amputation for free? How can we prohibit the assistance if we do not prohibit the act itself? One of Feinberg's discussions about voluntary slavery contracts can be adjusted to apply here.

Let us grant that not every case in which a person wants a healthy limb removed would fail a test of voluntariness. Still, the consequences of such a choice are so serious, so permanent, and so contrary to the statistical norm of what people want (as evidence, consider the lengths to which people will go to reattach accidentally severed limbs and the efforts to which amputees will go to learn to use artificial limbs) that we may presume nonvoluntariness as a default position. And if that is the case, that is, if most of the people who want to amputate a digit, a hand, or a limb are mentally disturbed or otherwise incompetent, then we can prohibit other people from providing the requested service (even for free) without committing ourselves to the permissibility of hard paternalism. For the aim of the restriction would be to protect the incompetent persons. Part of the way we protect them is by not allowing other nonprofessionals to assume that they can accurately determine which requests for amputation help are sufficiently rational, voluntary, and informed and which are not. The risks are just too great when the mistaken judgment cannot be corrected.

Such a restriction would have the effect of limiting the opportunities of those who truly were competent and their choice voluntary, because those people could not get help from another. But these limits would be the by-products of the soft paternalism and not the aim of hard paternalism. That is, the justification for the restriction was not to keep competent people from altering their bodies. That was just a derivative effect on a minority, in much the same way that laws requiring employers to pay overtime rates limit some people's opportunities to work more than forty hours a week (because the employer will not pay overtime). But such laws are not instances of hard paternalism. Instead, such worker protection laws get the majority of workers what they want and cannot get on their own.²³

Finally, the preceding discussion exposes a problem in the formulation of Shafer-Landau's “liberal argument for criminalizing self-mutilation” (which is an argument that he proposes and then rejects on the liberal's behalf). For if the state is justified in assuming that most cases in which a person wants a limb removed are cases which fail the test of voluntariness for whatever reason—the person is mentally ill, not rationally appreciative of the consequences (e.g., she thinks she can still windsurf after both feet are amputated)—then Shafer-Landau's Comparative Claim is wrongly worded. It is not that a hundred competent people would be wrongly denied the aid in order to protect one incompetent person. Rather, given that most requests would come from noncompetent people, and given that nonprofessionals are likely to make mistakes in determining who is and who is not competent (or whose desire is or is not sufficiently voluntary), it is better that we not allow any nonprofessionals to make and act on such judgments of competency or voluntariness (with the inevitable result that a few competent people would be wrongly denied assistance) than it would be to allow such judgments with the inevitable result that some noncompetent people will wrongly and irreparably have a limb removed. I view it as a virtue that this view does not necessarily preclude amputations within a medical setting. The crime lies in taking too great a risk with somebody else's welfare (given the odds of competency and the difficulty of accurately determining who is actually competent), not in taking a risk with your own.