The American Journal of Bioethics, 16(7): 3–14, 2016 Copyright © Taylor & Francis Group, LLC ISSN: 1526-5161 print / 1536-0075 online DOI: 10.1080/15265161.2016.1180442 Target Article Withdrawing Versus Withholding Freedoms: Nudging and the Case of Tobacco Control Andreas T. Schmidt, Princeton University Is it a stronger interference with people’s freedom to withdraw options they currently have than to withhold similar options they do not have? Drawing on recent theorizing about sociopolitical freedom, this article identifies considerations that often make this the case for public policy. However, when applied to tobacco control, these considerations are shown to give us at best only very weak freedom-based reason to prioritize the status quo. This supports a popular argument for so-called “endgame” tobacco control measures: If we believe that cigarettes would and should be withheld from entering markets in hypothetical scenarios in which they do not yet exist, then we also have reason to seek their abolition in situations, such as ours, in which cigarettes do exist—if necessary by banning their sale. The same considerations are then used to disarm objections that have recently been raised to using nudges in public policy. Keywords: health policy, philosophy, public health Imagine cigarettes did not yet exist as a consumer good in our societies and a company sought approval from the relevant authority (such as the Food and Drug Administration [FDA] in the United States) to introduce cigarettes as a new product. Assume the authority had evidence similar to ours about the addictive nature of smoking and its detrimental health effects. Cigarettes, it is reasonable to assume, would not be admitted as a new product (Ashcroft 2011, 88; Conly 2013, 169; Goodin 1989, 611; Khoo, Chiam, Ng, Berrick, and Koong 2010). Moreover, we might believe that restricting the access of new products into the market in this way is compatible with a system that strikes a good balance between individual freedom and population health. But if we think that withholding the option to smoke cigarettes would be all-things-considered justified in such a hypothetical scenario, does that mean that it is justified to withdraw the option to smoke cigarettes now? Khoo and colleagues believe that “tobacco is such a public health hazard that it is only an historical accident that makes its use lawful” (Khoo et al. 2010, 356). Sarah Conly and Robert Proctor each argue for a wholesale ban on the sale of cigarettes (Conly 2013; Proctor 2013). Others, including the British Medical Association, have proposed a partial ban limited to those born after the year 2000 (Berrick 2013; Daynard 2009; Khoo et al. 2010).1 One of the greatest worries about such drastic policies, however, is that banning the sale of cigarette is somehow too strong an interference with personal freedom. But can such a judgment be squared with the idea that it would be permissible to block the introduction of cigarettes in our hypothetical scenario? To do this, we need to hold that there is an important difference—from the perspective of personal freedom—between withdrawing an option that a person already has and withholding an option that a person could have. In this article, I argue that there often are indeed freedom-based reasons for why withdrawing options requires a stronger justification than withholding them. However, whether these theoretical reasons speak for “conservative” health policies depends on whether these considerations actually apply to individual situations. I show that in the case of tobacco control such reasons do not provide strong freedom-based objections to a ban. I then show how the argument provided here can enrich the current debate about nudging. Specifically, I defuse the oft-expressed worry that nudging conflicts with a concern for people’s freedom. Though I focus here on tobacco control and nudging, there are of course further areas of public policy to which the withdrawing/withholding problem applies. For example, is withholding the option to consume currently illegal drugs—such as marijuana and ecstasy in most countries— a lesser interference with people’s freedom than 1. I focus on a ban on the sale of cigarettes, but these considerations will apply, mutatis mutandis, to other proposals for achieving a tobacco-free society. See Malone (2010), Thomson, Edwards, Wilson, and Blakely (2012), and Warner (2013) for an overview. Address correspondence to Andreas T. Schmidt, Princeton University, University Center for Human Values, 315 Wallace Hall, Princeton, NJ 08544, USA. E-mail: andreas.schmidt@princeton.edu ajob 3 The American Journal of Bioethics withdrawing the freedom to consume alcohol would be? Do we require a stronger moral justification to regulate alcohol simply because it already is widely available? Or should we approach such problems with a “fresh starts view” and just compare the benefits and harms of the different drugs without taking into the status quo? Or consider gun control: If it is justified to withhold the option to own a gun in Europe, does that mean we are equally justified to withdraw such an option in Utah where gun ownership is currently an available option? I proceed as follows. In the second section, I specify the problem. In the third, fourth, fifth, and sixth sections, I draw on theories of sociopolitical freedom to show that there sometimes can indeed be reasons why withdrawing options is a stronger interference with people’s freedom than withholding them. In the seventh section, I apply these theoretical findings to the tobacco case. In the eighth section, I show how these considerations can be used in support of nudging. NONEQUIVALENCE OR STATUS QUO BIAS? Are there good arguments why the following should be true for public health policies? Nonequivalence: withdrawing an option x from a status quo choice-set X is a stronger interference with people’s freedom than withholding x to be added to a status quo choice-set Y that does not yet contain x.2 We might think that if we have good reasons for withholding, then surely the same reasons should apply to withdrawing. The fact that something exists seems in itself morally arbitrary. So, is ascribing moral significance to the withholding/withdrawing distinction the result of an irrational status quo bias rather than founded on good reasons? Status quo bias is a widespread and well-researched phenomenon: Repeated exposure to something will increase the probability that we like it (Bornstein 1989); we often rationalize preferences for existing options, particularly when we chose these options in the past (Brehm 1956); typically, people give less weight to gains than to equal losses (Kahneman, Knetsch, and Thaler 1991) and are often motivated by regret avoidance (Anderson 2003). Sometimes we even judge things more favorably simply because they exist (Eidelman, Crandall, and Pattershall 2009). Moreover, such a bias not only affects consumer choices and aesthetic judgments and so on, but even political preferences—or so it is often argued—are partly the 2. There has been a debate in medical ethics about whether there is a moral difference between withholding and withdrawing treatment in critical care (Gedge, Giacomini, and Cook 2007; Sulmasy and Sugarman 1994; Wilkinson and Savulescu 2014). Though related, my focus is different: Rather than with individual medical decisions, I am here concerned with population-level public health policies that affect which options are available (rather than which treatments should be administered or stopped). 4 ajob result of a status quo bias (Jost, Banaji, and Nosek 2004; Jost, Glaser, Kruglanski, and Sulloway 2003). Bostrom and Ord argue that this status quo bias leads to many overly conservative ethical judgments about public policy and applied ethics (Bostrom and Ord 2006). So, why not believe that nonequivalence in public health policy is yet another instance of such a bias? I answer this question in the following section. Before doing so, let me briefly set aside the following reasons for thinking the status quo morally special. First, my discussion is about whether a concern with people’s freedom implies giving the status quo special importance. I exclude other possible types of moral reasons—such as purely “consequentialist” or justice-based considerations—for thinking the status quo is special. Second, I mainly focus on the justification of public health interventions with respect to the freedom of consumers and patients. I am less concerned with the changes such interventions might have for producers and other stakeholders. An outright ban of tobacco, for example, would impact tobacco companies. I exclude these “supply side” considerations here (for a discussion of the economics of tobacco control see Max 2001; Warner 2000). Third, when discussing how public health interventions affect the freedom of individuals, I focus on how such interventions reduce freedom. To keep things simple, I will not discuss in any detail how far such interventions might also increase freedom (e.g., in the sense of “autonomy-preserving paternalism”; see Conly 2013; Scoccia 1990; VanDeVeer 1980).3 QUANTITATIVE CONSIDERATIONS Let us now see how far existing work on sociopolitical freedom might provide us with considerations that can often make nonequivalence the case. Since Isaiah Berlin’s famous lecture on freedom, it has become standard to distinguish between “negative” and “positive freedom.”4 Negative freedom is about the absence of interference: I am normally said to be free to the degree to which no man or body of men interferes with my activity. Political liberty in this sense is simply the area within which a man can act unobstructed by others. (Berlin 1969, 34) This is contrasted with positive freedom. Roughly, positive freedom is concerned not so much with a person’s 3. For the record, I believe a good case can be made for the claim that tobacco regulations can increase people’s freedom—by increasing their life expectancy, for example—across time by preventing people from engaging in behavior that reduces their future freedom. 4. Berlin mainly uses the term “liberty.” I use the term “freedom” and take them to be interchangeable. Wolff (1997) suggests a distinction between the two, which, however, does not seem to have caught on in the literature. July, Volume 16, Number 7, 2016 Nudging and Tobacco Control external options but with whether a person lives according to her “real” or “authentic” self (though there might be many other ways to describe it) (Taylor 1979). In this article, I am not concerned with “positive freedom” in this sense.5 Let us start by considering liberal, negative theories of freedom. I first consider a version of the negative view I call the Quantitative View. According to Ian Carter—a proponent of the Quantitative View—to assess public policies in terms of freedom requires gauging how such policies will affect people’s level of overall freedom and not merely their specific freedoms (Carter 1999). Carter furthermore argues that gauging the extent of people’s overall freedom is a purely “empirical” undertaking and can be done without invoking evaluative considerations. A person’s overall freedom is a function of how many acts (and their respective combinations) she is free and unfree to do. These specific freedoms and unfreedoms are individuated in purely spatiotemporal terms. To take a simple case: If a prisoner is free to occupy all areas of his 10-m2 cell, then he is less free than another prisoner who is free to occupy all areas of his 20-m2 cell. To gauge a person’s available sets of compossible freedoms, we thus divide space–time into equally sized regions and then ascertain “the sets of spatial regions that can be occupied by his body and by concomitant objects during some specified span of time” (Carter 1999, 380). At first sight, the Quantitative View does not give us good reason for nonequivalence. For it would seem surprising if existing options should matter more for a person’s overall freedom than nonexistent options, given that options are simply individuated by their spatiotemporal properties. However, one reason might be found in the so-called fecundity of specific options. “Fecundity” is a term introduced by Joel Feinberg and denotes the extent to which specific options can lead to further future options (Feinberg 1980, Chap. 1). For example, Alan’s freedom to step into his car and turn on the ignition is quite fecund, because it gives him the further freedoms to drive around the Norfolk countryside, to an owl sanctuary, and so on. Now it might turn out that in some situations existing options are more fecund than nonexistent options, because we have better knowledge about these options or because existing options are more likely part of coordinated social activities. Such coordination requires expectation about the behavior of other persons, which is more easily had for existing options. For example, it is easier for me to play football with the local amateur football team than to coordinate a match of a new sports game I have just created. 5. “Positive freedom” is sometimes also used to refer to the capability view of freedom (or “effective freedom”) (Miller 2006). My discussion of negative freedom is compatible both with seeing freedom as the absence of socially imposed constraints and with seeing it as “effective freedom.” July, Volume 16, Number 7, 2016 Of course, whether there is a difference in terms of fecundity for any particular public policy is empirically contingent. When discussing questions of tobacco control, we can see an example where purely quantitative considerations do not play an important role. PREFERENCES Most theorists of freedom believe that the Quantitative View is inadequate by itself. A person’s freedom is not only determined by how many options she has but also by how good these options are (Kramer 2003; Pattanaik and Xu 1998, 2000; Sen 1988; 1991; Taylor 1979). If we hold that the value of options contributes to how free a person is, then we might seek to support nonequivalence by holding that existing options are more valuable than (similar or equivalent) nonexistent options. But what account of value should we use? Let us first consider theories of freedom that use a preference-based account of value. Such accounts can presumably support nonequivalence by holding that people actually want existing options, whereas they typically have weaker or no preferences with regard to nonexisting options. The mechanisms mentioned in the second section do indeed imply that people typically have stronger preferences with regard to existing than with regard to nonexisting options. When coupled with a Hobbesian account of freedom, we get straightforward support for nonequivalence. According to the Hobbesian view, one is free to the extent that one is not interfered with in terms of the options one actually wants to pursue. When options are removed that one does not desire to do anyway, then this does not affect one’s freedom at all. Or if one’s preferences are less strong with regard to nonexistent options, then withholding options is a lesser interference than withdrawing options. However, the Hobbesian view is open to a well-known objection by Isaiah Berlin: Consider the contented slave whose preferences are closely aligned with the very few options available to him. Such a slave is not unfree at all according to the Hobbesian view. But it seems implausible that one can make oneself free in the social sense of the term—just like that—by changing one’s preferences to only those options available (Berlin 1969, xxxviii).6 Preference accounts risk confusing the psychological with the social sense of being free, particularly so in cases of adaptive preferences. Besides the more general theoretical problem identified by Berlin, there is a more specific problem when we try to 6. Instead of the strict Hobbesian version of preferencedependence, we could also say that options that one desires matter more for one’s overall freedom than those one does not desire (though both matter). Sen (1991) could be interpreted along these lines. However, such a view is also open to Berlin’s objection, albeit in a somewhat attenuated form. For it still holds that the slave can make himself more free—though not perfectly free—by adapting his preferences to the few options available. ajob 5 The American Journal of Bioethics The theories of freedom discussed so far conceptualized overall freedom (or rankings of opportunity-sets) by invoking preferences. We have seen that this is problematic. Alternatively, we could endorse an objectivist theory of value, according to which the evaluative weighting factors that determine a person’s freedom are not based on subjective preferences. While such theories are nonsubjectivist, they still leave room for the idea that different people need different things to lead good or flourishing lives. (For example, while a happiness theory of well-being is objectivist, it would still hold that many people require very different lifestyles to be happy.) According to such theories, how much freedom a particular choice-set provides a person is a function of the objective quality of options in that choice-set. Such theories can also be hybrid theories and hold that it is both the quality and the size of a choice-set that determines how much freedom a person has (Arneson 1985, 443–444; Berlin 1969, xxxix–xl; Kramer 2003, Chap. 5.2).7 Invoking objective value in a hybrid theory could support nonequivalence, if it is the case that existing options are objectively more valuable than nonexistent ones. Indeed there are some considerations for why existing options will often (or at least sometimes) be better. First, existing options can be valuable in a path-dependent way, when they have become part of people’s identities. For example, in a very Catholic society, the freedom to eat a wafer as part of communion will be more valuable than the freedom to eat a wafer in a purely Atheist society. We can make this claim without invoking subjective preferences, if we assume that what is objectively good for a person still depends on that person’s very specific and path-dependent situation. It might be objectively valuable for a person to have a sense of self and identity and pursue relevant activities accordingly. Another, related reason for assigning higher value to existing options is that certain practices often become part of valuable communal activities. Without the tradition of Halloween, for example, the option to walk around in scary costumes and knock on people’s doors would clearly be less valuable. So, certain activities can become more valuable, if they are repeated regularly—become a tradition—are done communally and form part of people’s identities. We might think that this also applies to some health-related lifestyle choices, such as people’s eating and drinking behavior, their use of recreational drugs, what types of sports they play, and so on. Second, there is also a more structural reason why existing options might be more valuable than new options. Being free also implies being able to determine the shape of one’s life according to one’s conception of the good. We care about our ability to pursue plans that range into the future. We would not be considered very free if we could only ever plan the next 10 minutes (even if we had very many choices for these 10 minutes). To be able to make longer term plans, however, requires some level of crosstemporal stability in one’s options. If I do not know which options I will have in a week’s time or if these options constantly change, then this makes it very difficult to plan my life. To be able to control the shape of one’s life also requires foreseeability. Existing options will typically be more important for one’s ability to foreseeably plan one’s future life on a long-term basis than currently nonexisting options.8 Moreover, for many options that have existed for a while it is more likely that they already form part of people’s plans and that people have adjusted their expectations accordingly. Therefore, the continued existence of current options is typically more valuable than the addition of new options when the former contribute to people’s ability to realize their life plans. Third, related to the preceding reason, a further consideration is that withdrawing options can create transition costs that do not exist in the case of withholding options. For example, as is well known from theorizing about path dependence, setting out on certain paths can make it difficult to switch to a better path, because such switching is often costly. We can imagine something similar for individuals. Finally, we might also look for an individualistic and epistemic reason for giving some priority to existing options: We might think—inspired by Mill—that often individuals themselves know best what is good for them. The reason certain options “still exist” might be because unlike other, less popular options, these options are still in demand. These options still being in demand might be an indication that these options are more valuable than others, given our uncertainty about individuals and their respective psychological makeup and so on.9 Now, 7. Though the authors listed in the text endorse a hybrid theory, some of them rely on a desire-based or preference-based theory of value. Note also that I am somewhat flexible as to what counts as an “objectivist” theory. I want to allow idealized preferences views which abstract from a person’s actual preferences. The arguments presented in this section are compatible both with most traditional objectivist theories of the good and with most idealized preference views. 8. Of course, we could previously notify people before changing options. However, even then it would often be difficult to pursue longer-term plans across time. Say you could only ever have one job for 1 year before being transferred into a very different job. This would make skill development and longer term projects difficult. 9. Though we should be careful not to overstretch this indicative function (Conly 2013, Chap. 2). use a preference-based measure to support nonequivalence. Many people have incoherent or even inconsistent desires. For example, while smokers desire to smoke, the majority of U.S. American smokers also wish they could quit (74%), and most of them have tried to do so in the past (85%)

szegi 2004; Newport (Agaku et al. 2014; Gruber and Ko 2013). Smoking is highly addictive, which is why people have incoherent desires regarding smoking. OBJECTIVE VALUE 6 ajob July, Volume 16, Number 7, 2016 Nudging and Tobacco Control compare withdrawing an existing popular option with withholding a nonexistent option. Here, we have less information about the potential popularity of the nonexistent option than we do about the existing option. This might give us (defeasible) epistemic reason to often or at least sometimes judge existing popular options to be more valuable than nonexistent options. By invoking a (partly) evaluative view of freedom, we have so far found the following support for nonequivalence: First, existing options can sometimes be more valuable than nonexisting ones, if they contribute to identity, community, and the ability to plan one’s life on a longterm basis, and there might sometimes be epistemic reason why existing options have more value than nonexistent options. Second, because a (partly) evaluative view of freedom gives more weight to valuable options, existing options weigh in more strongly when determining a person’s level of overall freedom than nonexistent similar options. Therefore—and this is nonequivalence—there often is a difference in terms of freedom between withdrawing and withholding options. STATUS FREEDOM Thus far, I have looked for support for nonequivalence by invoking different negative theories of freedom. Some people take issue, however, with the idea that a person’s freedom is a function of her options (and maybe the number and severity of restrictions). According to civic republicanism—not to be confused with the political views of the Republican party in the United States—for example, freedom is not only about having options. Instead, freedom is a property of persons; it is about having a certain type of social status. Philip Pettit, for example, argues that being unconstrained in one’s options is not enough, because if another person has the capacity to arbitrarily interfere with my life, then this makes me unfree even if that person decides not to exercise this capacity (Pettit 1997; 2003; 2014). On this view, freedom is about being free from domination, about not being subject to the whim of another person. To bring out more clearly the difference between civic republicanism and negative, liberal views of freedom, consider a slave whose benevolent master lets him do whatever he wishes to do. On a liberal, negative view, the slave is considered free inasmuch as he is free from interference. According to civic republicanism, on the other hand, the status of being a slave—of being subject to another person’s arbitrary power—is in itself sufficient to render the slave unfree. The slave is unfree even if he is in fact never interfered with by his master. Of course, people will always have some form of capacity to interfere with other people’s lives. But if such power is suitably constrained, say, through proper democratic and legal procedure, then this power is nonarbitrary. Someone interferes on a nonarbitrary basis if the procedure (by which to determine whether to interfere or not) is designed to force those taking the decision to track the July, Volume 16, Number 7, 2016 relevant interests of the individual whose options are being constrained (Pettit 2002, 290). Accordingly, state interference does not undermine a citizen’s status as a free person, if such interference is the result of nonarbitrary power. A master’s title over a slave, on the other hand, grants him arbitrary power, which is why the slave lacks the status of a free person. Does the idea of status freedom mean that in the realm of public policy withdrawing options is a stronger interference than withholding options? In principle, there is no principled difference between withdrawing and withholding, as long as such interferences are nonarbitrary (and thus happen against the backdrop of a nondominating distribution of power). However, questions of domination will come up more dramatically if the decision to interfere is about something people have strong preferences about. Because people typically (but not necessarily) have stronger preferences regarding existing options than nonexistent ones, this might make the issue of whether decisions to withdraw options are imposed on an arbitrary or nonarbitrary basis more pressing.10 In cases of withdrawing, the justificatory stakes are thus often higher than in cases of withholding (even though both can be instances of domination or nondomination). BANNING THE SALE OF CIGARETTES The considerations proffered for nonequivalence show that we often require a stronger justification for withdrawing than withholding options. What does this theoretical debate imply for public health policy? On the one hand, it implies that prioritizing the status quo need not always be the result of a status quo bias. On the other hand, whether there is an important freedom-based difference between withdrawing and withholding depends on how far these considerations apply to the specific policy at hand. Do these considerations imply that a concern for personal freedom rules out a ban on the sale of cigarettes? To answer this question, let us go through the aforementioned considerations for nonequivalence and see how far they apply to smoking. Here is the list of considerations again: 10. Some preferences will be self-regarding—such as a preference to smoke. However, people might also have “social” or otherregarding preferences. For example, one might have a preference to live in a society in which smoking is legal even when one does not desire to smoke oneself (or, conversely, a preference for smoking to be illegal while also desiring to smoke). If people tend to have stronger (social) preferences to keep the status quo, then this might further raise the justificatory stakes for cases of withdrawing. One challenge for civic republicanism hereby is to delineate which of these preferences should be respected in the process. For some preferences—such as other-regarding discriminatory preference—should not enter the procedure, whereas other otherregarding preferences might. Pettit favors a solution in which it is determined procedurally in a nondominating way which interests count as a relevant. To keep things simple, I do not pursue these issues further here. ajob 7 The American Journal of Bioethics Quantitative view Objective Value Status freedom Fecundity Community and identity Popularity as indicative of value Ability to plan one’s life Transition costs Withdrawing can raise justificatory stakes because of stronger preferences Quantitative considerations Existing options might sometimes be more fecund—that is, lead to comparatively more additional options—than possible new ones. However, does this apply to tobacco control? Some considerations might support the idea that banning cigarettes, for example, effectively bans further options in its wake. For example, it is common to observe smokers stepping outside together for a cigarette and a chinwag. Such further options are not available, one might argue, for as-yet nonexistent products. However, fecundity alone does not support nonequivalence in the case of smoking. The reasons why we might find the freedom to smoke more important than the freedom to try out new, similar drugs is not explained by merely quantitative considerations. Remember that a purely Quantitative View such as Carter’s individuates freedoms by their spatiotemporal properties and does not take into consideration how good individual options are. Now, consider the physical movements that can be performed with a cigarette in hand, such as holding it in one’s hand, moving it toward one’s mouth, and so on. The freedom to perform such movements could, in principle, be substituted by the freedom to use carrots or other similarly shaped objects. In purely physical terms, holding a carrot would allow one to perform very similar movements as holding a cigarette. But it seems that smokers care more about what cigarettes do to their bodies rather than what they can do with cigarettes as objects in the external world. Substituting cigarettes with carrots—or other similarly shaped objects— does not seem preserve the same freedom. The Quantitative View with its focus on spatiotemporal extensions and its associated notion of fecundity does not cohere with our pretheoretical judgment in this case. Objective value We saw that stronger reasons for nonequivalence are provided by a (partly) evaluative theory of personal freedom. Sometimes withdrawing is a stronger interference than withholding, because existing options can become more valuable through their existence. I mentioned four reasons for this. Let us now see how far they apply to a ban on the sale of cigarettes. First, existing options can be more valuable than possible new ones on account of the path-dependent value they get through their relation to people’s identities, valuable communal activities, and so on. However, is smoking 8 ajob valuable in this way? There are some reasons why cigarette smoking might be part of communal life. Sharing cigarettes might strengthen friendship ties, for example, or form part of youth culture (Nichter 2003; Unger et al. 2001). Smoking cigarettes might also form part of people’s identities; at least it is often presented this way in films and advertising—though somewhat less so these days—to conjure up images of rebellious libertines, pensive intellectuals, artists, and so on. However, I think we should resist inferring from such individual examples that smoking is generally or typically an important part of people’s identities and conceptions of the good life. First of all, many aspects of such presumed identities are the result of clever marketing, and this should give us prima facie reason not to believe that every positive narrative around smoking reflects a real and valuable identity. Moreover, as Robert Goodin writes, maybe the reality of (most instances of) smoking is very different from idealized images: [T]he “smoking ritual” might play an important role in the way of life of several broad classes of people. Insofar as we find those ways of life valuable, and insofar as they truly cannot be sustained without the smoking ritual, perhaps we ought not discourage smoking (among those groups, at least). But both those provisos are likely not to be satisfied. Central though smoking may be, few whole ways of life would disappear without it. And since it is principally those with very boring or very stressful life-styles who get hooked into these patterns of repetitive behavior (Leventhal and Cleary 1980; Spielberger 1986), these are unlikely to be ways of life we would care to see perpetuated. (Goodin 1989, 612) How much value would be lost (or gained) if people smoked less—or stopped completely—is hard to judge from the philosophical armchair. It is clear that there might genuinely be valuable instances of smoking. But one should keep in mind, first, that the potential value had by smoking practices strongly depends on the different cultural and social contexts within which the smoking ritual takes place and, second, that the reality of most cases of smoking in the majority of smoking instances—usually a form of addictive behavior—is very different from a few isolated positive examples. Second, maybe we have prima facie reason to believe that smoking is valuable simply because it is so popular. Around 18% of U.S. American adults smoke despite widespread knowledge of its bad health effects and despite high taxes on cigarettes. Is this not an indication that smoking must in some sense be valuable? Smoking does indeed have a positive stimulating effect by triggering the release of neurotransmitters in the brain. It is often also believed that smoking helps with weight control and relieves stress. However, these perceived benefits might be overestimated, particularly for those addicted to nicotine. Persons with a nicotine addiction usually smoke to modulate mood and/or to relieve withdrawal symptoms (Benowitz 2010). Conly therefore argues that “the physical pleasure eventually felt by practiced smokers appears to be largely negative, in that it is the elimination of something like a July, Volume 16, Number 7, 2016 Nudging and Tobacco Control pain, rather than purely enjoyable in itself” (Conly 2013, 171). Moreover, smokers seem to experience more stress overall than nonsmokers and those who have quit smoking successfully (Parrott 1998; West and Hajek 1997). Thus, the continued popularity of cigarette smoking is maybe more indicative of its addictive qualities than of its great additional value. As noted earlier, most smokers wish they did not smoke and most have tried to quit in the past. While these considerations do not rule out that smoking cigarettes (particularly when done occasionally) can add value to people’s lives, it does weaken the idea that the continued popularity of cigarette smoking is indicative of its value, particularly for those addicted to nicotine. Third, existing options can be more important, if these form part of one’s plan of life. Withholding new options can thus sometimes affect people’s abilities to plan their lives across time to a lesser extent than withdrawing existing options. However, the preceding considerations speak against the idea that banning cigarettes would undermine people’s ability to pursue long-term plans. Most smokers want to quit and have tried to do so in the past. We can of course imagine people whose long-term life plans include smoking cigarettes, but these will be a small minority of actual smokers. For the most part, the ability to plan one’s life across time does not seem to depend very strongly on the availability of cigarettes. Finally, would withdrawing the option to smoke create transition costs that would be absent for withholding? There clearly are such considerations for smokers who would have withdrawal symptoms (assuming a ban would be effective). Thus, a good policy would require tools to help smokers with the transition (through medical help, the availability of other nicotine products, and so on). Note also that unlike a wholesale ban, a partial ban would get round some of these problems. Because such a ban would only prohibit the sale of cigarettes to those who do not yet smoke (those born after the year 2000 and those younger than the legal age of smoking), these people would not be subjected to any transition costs (assuming, again for the sake of argument, the ban would be effective). all the more reason to go through nondominating procedures and to enact such a ban in a democratic, transparent, and accountable manner.11 So, if a cigarette ban is put in place in a nonarbitrary way against the backdrop of nondominating social relations, a concern for status freedom does not by itself rule out its justifiability.12 Here I have not tried to argue that it is, all things considered, right (or best) to ban cigarettes. To do so would require assessing other relevant empirical and moral considerations, such as consequentialist and distributional concerns. Instead, the aim of the tobacco control discussion has been twofold. First, the cigarette case was given as an example of how to apply the kinds of more general considerations for nonequivalence to an individual case of public policy. Second, it was shown that, in this case, these considerations do not give us an overriding argument against banning cigarettes on the basis of a concern for personal freedom. If we think we should not allow products similar to cigarettes, we have good reason—all of this depending on empirical considerations, of course—to also enact tobacco control measures aimed at eradicating cigarettes completely. Note that this also gives us more leeway for an integrated strategy to regulate conventional cigarettes together with other nicotine products. While proposing any such regulation in detail is beyond the scope of this article, let me nonetheless add some comments on one such alternative nicotine product, the electronic cigarette. E-cigarettes are the most important subgroup of so-called electronic nicotine delivery systems, which vaporize liquids that contain nicotine but no tobacco. Recently, ahead of a meeting of the Framework Convention for Tobacco Control of the World Health Organization (WHO), a rather heated debate between tobacco control specialists has emerged regarding the proper regulation of e-cigarettes (Fairchild and Bayer 2015; “Letters to WHO on nicotine science and public policy,” 2014). Proponents of a more liberal strategy hold that e-cigarettes might be a less harmful substitute for conventional cigarettes and can function as a cessation tool, citing some (preliminary) studies speaking to their effectiveness (Brown, Beard, Kotz, Michie, and West 2014; McRobbie, Bullen, Hartmann-Boyce, and Hajek 2014). Opponents to e-cigarettes respond that e-cigarettes might serve as Status freedom Proponents of civic republicanism hold that freedom is about having a particular type of status, namely, being free from domination. The question of nondomination cuts across the withholding/withdrawing distinction, as both can be compatible with a person’s status freedom if they are imposed on a nonarbitrary basis. However, it might turn out that the justificatory stakes in the case of a cigarette ban will be higher than in the case of withholding cigarettes in our hypothetical scenario. This is so because people in the actual world have stronger preferences regarding smoking than those in our hypothetical scenario. However, as mentioned in my earlier discussion, that the justificatory stakes are higher simply means that we have July, Volume 16, Number 7, 2016 11. Smoking has a social gradient such that those of lower socioeconomic status (SES) tend to smoke more than those of higher SES. Tobacco control should thus not be done in a way that it increases any domination to which those of lower SES might already be subject. See Voigt (2010) on related issues. 12. The popularity of partial and complete bans of cigarettes seems to vary from country to country. While support for such measures seems widespread in Australia, New Zealand, and Hong Kong (ranging from 50 to 70%), it is comparatively low in many European countries (Daynard 2009; Edwards et al. 2013; Gallus et al. 2014; Hayes, Wakefield, and Scollo 2014; Khoo, Chiam, Ng, Berrick, and Koong 2010; Wang, Wang, Lam, Viswanath, and Chan 2015). Given that these studies are designed differently, however, we should not be too confident when comparing their results. ajob 9 The American Journal of Bioethics gateway drugs, that they can lead to dual use (i.e., smokers smoking both conventional and e-cigarettes), that their purported comparative harmlessness remains empirically unsubstantiated, and that a liberal strategy risks the renormalization of smoking and plays into the hands of tobacco companies. Rather than contributing to—or taking a stand on— this largely empirical and strategic debate, let me briefly explain how the withdrawing versus withholding distinction is relevant in this context. On the one hand, I have argued that it is sometimes easier to justify withholding the introduction of a new product than to justify the withdrawal of a similar established one. This can give us reason to be cautious about allowing new products. On the other hand, in the case of tobacco control, I argued that the considerations speaking for nonequivalence do not give us strong reason against more draconian measures to regulate conventional cigarettes. Now, if e-cigarettes are indeed a better alternative to conventional cigarettes or are effective as a cessation tool (a “big if”), this would suggest being more draconian about existing conventional cigarettes while at the same time being open to utilizing (in a regulated form) e-cigarettes as part of a harm reduction strategy (Fairchild, Bayer, and Colgrove 2014). NUDGING Let me now show how to apply these aforementioned considerations to debates about much less invasive health interventions. Recently, nudging has received a lot of attention (Sunstein 2013, 2014; Thaler and Sunstein 2008). The main idea is that we can nudge people into taking better decisions simply by changing the way options are presented—the choice architecture—without removing any options from the choice-set or, at least, without drastically changing the incentive structure.13 A growing body of literature in behavioral psychology shows that people’s decision making is far from perfect. For example, quite often we tend to go with the default option. By changing the way options are presented—by making the healthy option the default option, for example—we can exploit these biases to nudge people into better decisions.14 Though nudging is less invasive than completely withholding or withdrawing options, it has still come under attack by some authors who argue that deliberately changing the choice architecture to exploit our faulty decisionmaking procedures is problematic as far as people’s 13. I stick to this more narrow understanding of “nudging.” In recent formulations, Sunstein classifies a broader range of interventions—even those of the traditional carrot and stick type—as “nudges.” I also avoid the label “libertarian paternalism,” because not all the policies that proponents of nudging endorse are paternalist in a stricter sense of the term. 14. See Blumenthal-Barby and Burroughs (2012) and Saghai (2013) for more examples of nudging in health care and public health. 10 ajob freedom and autonomy are concerned (Bovens 2009; Goodwin 2012; Gr€ une-Yanoff 2012; Hausman and Welch 2010). Proponents of nudging often respond that nudging is inescapable. Even if we decide not to nudge people into better behavior, they are still being nudged by the way the choice situation is structured. Now, if we have to choose between nudges that have positive effects and those that lead us to bad decisions, we should clearly go with the former. Consider one of Thaler and Sunstein’s examples (Thaler and Sunstein 2008, 1–2). They note that the way food is arranged in a cafeteria influences what we tend to buy. For example, we are more likely to choose food that comes first in a buffet, is at eye level, or is easily within reach. What should a specific cafeteria do in light of these effects? Let us distinguish four different cases.15 Case 1: The cafeteria intentionally offers dishes in an order that maximizes its profit, even though this has consistently negative effects on people’s health. Case 2: The cafeteria intentionally offers dishes in an order that makes people choose the unhealthiest dishes. The aim hereby is to make people worse off. Case 3: The cafeteria intentionally offers dishes in an order that makes people choose the healthiest dishes. Case 4 (The “status quo option”): The cafeteria offers dishes in an order they have always offered the dishes without giving any thought to the effects this order will have on people’s choices. Let us assume that the cafeteria would achieve its intended aims in all cases. Proponents of nudging would now argue that Case 3 is preferable over Case 2 over Case 1. But what about Case 3 and Case 4? We might think that trying to influence people’s choices is an interference with their freedom in a way that leaving things as they are is not. Now, do the considerations offered in favor of upholding the withholding/withdrawing distinction in public health policy apply, mutatis mutandis, to nudging too? Is nudging problematic, simply because it changes the status quo? Because nudging leaves the range of options (more or less) intact, the first four considerations that spoke for nonequivalence do not apply here. First, there is no relevant difference in terms of fecundity of options here, as the range of available options stays the same. Second, for the same reason, the change in the presentation of these options alone does not change people’s ability to pursue plans across time. If anything, nudging is intended to help people overcome weakness of will and improve their ability to make choices that reflect their long-term interests 15. The list of cases is somewhat different than Thaler and Sunstein’s original list, to fit in with the focus of the current discussion. Also, I focus on the freedom of consumers in this case. A separate question, surprisingly absent in discussions of nudging, is how far regulations to enforce nudges interfere with the freedom of companies (or other “supply-side institutions”). July, Volume 16, Number 7, 2016 Nudging and Tobacco Control and plans. Third, for the same reason, communal and identity-related values are not endangered, because the options to pursue those remain. Fourth, the epistemic reason does not apply either, because no option—however valuable prima facie—is removed. However, there might, fifth, be a legitimate worry about nudging, if we understand freedom in the status sense. Nudging might make us subject to other people’s arbitrary power and thus undermine our status as free persons (Gr€ une-Yanoff 2012, 638).16 This might be the case even if the range of options remains the same. (I leave aside the worry that such trivial questions about how food is arranged in a cafeteria might be irrelevant for a person’s status freedom.) Some might think that respecting people’s status as free persons requires refraining from instrumentalizing their often-faulty decision-making habits. If we want to respect people as free, self-determining persons, imposing one’s will on another person seems more problematic than leaving such things to chance (Gr€ une-Yanoff 2012, 639; Hausman and Welch 2010, 130).17 However, I do not think that this provides us with a strong objection to nudging. Theorists of status freedom— such as Pettit, for example—make it clear that even proper interferences can leave people’s status as free persons intact. If they are chosen on a nonarbitrary basis, they do not undermine people’s status as free persons. A fortiori, the same is true for nudging. As long as nudging is done on a nonarbitrary basis, it does not undermine people’s status as free persons. In our simplified cases, this would mean that such decisions would have to be done in a transparent and, ideally, suitably democratic way. For public policy on a bigger level, this implies due democratic process and transparency. On a smaller scale, for example, in a school cafeteria, this might imply that decisions to nudge people should be taken transparently, in a way that is sensitive to people’s avowed interests and, if necessary, that gives them a voice in the process. Thus, when an organization like a school intends to implement nudges, for example, it makes sense to integrate proper debate about such a change, rather than implementing it top-down and behind closed doors.18 Such nudging is perfectly compatible with status freedom, even though someone intentionally changes the choice architecture. 16. Pettit himself thinks that as long as nudges are not deceptive, they do not qualify as interferences and thus do not affect people’s status freedom (Pettit 2014, 35). 17. Thaler and Sunstein notice a difference between intended and unintended behavior change but do not really discuss whether this difference is morally relevant (Thaler and Sunstein 2008, 10). 18. It has been argued that transparency might undermine the effectiveness of nudging (Bovens 2009, 217; Gr€ une-Yanoff 2012, 638). However, neither Bovens nor Gr€ une-Yanoff provides empirical evidence for this claim. In an unpublished paper, Loewenstein and colleagues argue that informing people about the use of default-nudges does not affect their effectiveness (Loewenstein, Bryce, Hagmann, and Rajpal 2014). July, Volume 16, Number 7, 2016 We might even arrive at a stronger conclusion. Case 1 might be considered a greater affront to people’s freedom than Case 3. In Case 1 the canteen takes advantage of people’s faulty decision-making habits to increase their profit. Often a purely random choice-architecture will not be available, as marketers try to nudge people into decisions that are best for their company but not the consumer. As proponents of nudging point out, such nudges are ubiquitous in consumer societies. This suggests that status freedom actually favors more rather than less “benevolent” nudging. Unwittingly being instrumentalized by other people can quite plausibly be considered a greater affront to one’s status freedom than being influenced by people who have one’s best interest at heart, who do so transparently, and who have to be responsive to one’s avowed interests in the process. Thus, if positive influence is exerted in a transparent and accountable fashion, and if the alternative to well-intentioned nudges is “negative nudges,” nudging might foster rather than undermine status freedom. In conclusion, none of the reasons that gave (modest) support for nonequivalence speak against using nudging to improve people’s health-related decisions. Instead, a concern for a person’s status freedom sometimes even supports the use of nudges. CONCLUSIONS I argued that there often is a difference in terms of personal freedom between withdrawing and withholding options, such that the former requires a stronger justification than the latter. On the one hand, this shows that prioritizing the status quo in this sense is not always an irrational bias. On the other hand, it also shows that rather than being overly conservative about public policy in general and public health policies in particular, we should check whether and how far such reasons actually apply to specific policies. Policies to regulate food and alcohol consumption, cigarette smoking, tobacco control, gun control, and so on will all differ in the extent to which “freedom-based” reasons to favor the status quo apply. I argued that the relevant considerations that support a difference between withholding and withdrawing in the case of tobacco control are comparatively weak. Therefore, one of the standard arguments for so-called endgame measures of tobacco control should be allowed to stand: If we believe that cigarettes would not and should not be allowed to enter a market in hypothetical scenarios in which they do not yet exist, then concerns about personal freedom do not give us strong reason against also seeking their abolition in situations, such as ours, in which cigarettes do exist. I also tried to show how these aspects can be used to defuse freedombased objections to nudging. Reasons that speak for nonequivalence in other contexts do not even provide pro tanto reasons against using nudges. But they do support the idea that decisions to nudge should be taken in a transparent and democratic fashion. If done right—and if nudging ajob 11 The American Journal of Bioethics is indeed effective—it might in some situations even strengthen people’s status as free persons. Given limited space, I have here only analyzed how far freedom-based reasons for prioritizing the status quo might apply to nudging and tobacco control. But by providing a list of empirically contingent considerations for nonequivalence, I hope to have provided a more general public policy tool—a checklist—that can be used to assess for specific policies whether freedom-based considerations justify prioritizing the status quo or whether doing so is the result of a mere bias. ACKNOWLEDGMENTS I would like to thank Nir Eyal, Mark Budolfson, Alecia McGregor, two reviewers and the editors of AJOB for generous comments on earlier versions of this paper. For helpful discussions, I also thank the audience at a talk at the Center for Health and Wellbeing in Princeton. & REFERENCES Agaku, I. T., B. A. King, and S. R. Dube. 2014. Current cigarette smoking among adults—United States, 2005–2012. MMWR Morbidity and Mortality Weekly Report 63(2): 29–34. Anderson, C. J. 2003. The psychology of doing nothing: Forms of decision avoidance result from reason and emotion. Psychological Bulletin 129(1): 139–67. Arneson, R. J. 1985. Freedom and desire. Canadian Journal of Philosophy 15(3): 425–48. Ashcroft, R. 2011. Smoking, health and ethics. In Public health ethics: Key concepts and issues in policy and practice, ed. A. Dawson, 85–99. New York, NY: Cambridge University Press. Benowitz, N. L. 2010. Nicotine addiction. New England Journal of Medicine 362(24): 2295–303. http://doi.org/10.1056/NEJMra0809890 Berlin, I. 1969. Four essays on liberty. Oxford, UK: Oxford University Press. Berrick, A. J. 2013. The tobacco-free generation proposal. Tobacco Control 22(suppl. 1): i22–26. http://doi.org/10.1136/tobaccocontrol-2012-050865 Blumenthal-Barby, J. S., and H. Burroughs. 2012. Seeking better health care outcomes: The ethics of using the “nudge.” American Journal of Bioethics 12(2): 1–10. http://doi.org/10.1080/ 15265161.2011.634481 Bornstein, R. 1989. Exposure and affect: Overview and meta-analysis of research, 1968–1987. Psychological Bulletin 106: 265–89. Bostrom, N., and T. Ord. 2006. The reversal test: Eliminating status quo bias in applied ethics. Ethics 116(4): 656–79. http://doi.org/ 10.1086/505233 Bovens, L. 2009. The ethics of nudge. In Preference change: Approaches from philosophy, economics and psychology, ed. T. Gr€ uneYanoff and S. O. Hansson, 207–19. Berlin, Germany and New York, NY: Springer Science & Business Media. 12 ajob Brehm, J. W. 1956. Postdecision changes in the desirability of alternatives. Journal of Abnormal and Social Psychology 52(3): 384–89. http://doi.org/10.1037/h0041006 Brown, J., E. Beard, D. Kotz, S. Michie, and R. West. 2014. Realworld effectiveness of e-cigarettes when used to aid smoking cessation: A cross-sectional population study. Addiction 109(9): 1531– 40. http://doi.org/10.1111/add.12623 Carter, I. 1999. A measure of freedom. Oxford, UK: Oxford University Press. Conly, S. 2013. Against autonomy: Justifying coercive paternalism. Cambridge, UK: Cambridge University Press. Daynard, R. A. 2009. Doing the unthinkable (and saving millions of lives). Tobacco Control 18(1): 2–3. http://doi.org/10.1136/ tc.2008.028308 Edwards, R., N. Wilson, J. Peace, D. Weerasekera, G. W. Thomson, and H. Gifford. 2013. Support for a tobacco endgame and increased regulation of the tobacco industry among New Zealand smokers: Results from a National Survey. Tobacco Control 22(e1): e86–93. http://doi.org/10.1136/tobaccocontrol-2011-050324 Eidelman, S., C. S. Crandall, and J. Pattershall. 2009. The existence bias. Journal of Personality and Social Psychology 97(5): 765–75. http://doi.org/10.1037/a0017058 Fairchild, A. L., and R. Bayer. 2015. Smoke and fire over e-cigarettes. Science 347(6220): 375–76. http://doi.org/10.1126/ science.1260761 Fairchild, A. L., R. Bayer, and J. Colgrove. 2014. The renormalization of smoking? E-Cigarettes and the tobacco “Endgame.” New England Journal of Medicine 370(4): 293–95. http://doi.org/ 10.1056/NEJMp1313940 Feinberg, J. 1980. Rights, justice, and the bounds of liberty: Essays in social philosophy. Princeton, NJ: Princeton University Press. Gallus, S., A. Lugo, E. Fernandez, A. B. Gilmore, M. E. Leon, L. Clancy, and C. La Vecchia. 2014. Support for a tobacco endgame strategy in 18 European countries. Preventive Medicine 67: 255–58. http://doi.org/10.1016/j.ypmed.2014.08.001 Gedge, E., M. Giacomini, and D. Cook. 2007. Withholding and withdrawing life support in critical care settings: Ethical issues concerning consent. Journal of Medical Ethics 33(4): 215–18. http:// doi.org/10.1136/jme.2006.017038 Goodin, R. E. 1989. The ethics of smoking. Ethics 99(3): 574–624. Goodwin, T. 2012. Why we should reject “nudge.” Politics 32(2): 85–92. http://doi.org/10.1111/j.1467-9256.2012.01430.x

szegi. 2004. Tax incidence when individuals Gruber, J., and B. Ko are time-inconsistent: The case of cigarette excise taxes. Journal of Public Economics 88(9–10): 1959–87. http://doi.org/10.1016/j. jpubeco.2003.06.001 Gr€ une-Yanoff, T. 2012. Old wine in new casks: Libertarian paternalism still violates liberal principles. Social Choice and Welfare 38 (4): 635–45. http://doi.org/10.1007/s00355-011-0636-0 Hausman, D. M., and B. Welch. 2010. Debate: To nudge or not to nudge*. Journal of Political Philosophy 18(1): 123–36. http://doi. org/10.1111/j.1467-9760.2009.00351.x July, Volume 16, Number 7, 2016 Nudging and Tobacco Control Hayes, L., M. A. Wakefield, and M. M. Scollo. 2014. Public opinion about ending the sale of tobacco in Australia. Tobacco Control 23(2): 183–84. http://doi.org/10.1136/tobaccocontrol2012-050777 Pattanaik, P. K., and Y. Xu. 1998. On preference and freedom. Theory and Decision 44(2): 173–98. Jost, J. T., M. R. Banaji, and B. A. Nosek. 2004. A decade of system justification theory: Accumulated evidence of conscious and unconscious bolstering of the status quo. Political Psychology 25(6): 881–919. http://doi.org/10.1111/j.1467-9221.2004.00402.x Pettit, P. 1997. Republicanism: A theory of freedom and government: A theory of freedom and government. Oxford, UK: Oxford University Press. Jost, J. T., J. Glaser, A. W. Kruglanski, and F. J. Sulloway. 2003. Political conservatism as motivated social cognition. Psychological Bulletin 129(3): 339–75. Kahneman, D., J. L. Knetsch, and R. H. Thaler. 1991. Anomalies: The endowment effect, loss aversion, and status quo bias. The Journal of Economic Perspectives 5(1): 193–206. Khoo, D., Y. Chiam, P. Ng, A. J. Berrick, and H. N. Koong. 2010. Phasing-out tobacco: Proposal to deny access to tobacco for those born from 2000. Tobacco Control 19(5): 355–60. http://doi.org/ 10.1136/tc.2009.031153 Kramer, M. H. 2003. The quality of freedom. Oxford, UK: Oxford University Press. Letters to WHO on nicotine science and public policy. 2014. Retrieved from http://nicotinepolicy.net/n-s-p/2003-glantz-let ter-towho-the-importance-of-dispassionate-presentation-andinter pretation-of-evidence Leventhal, H., and P. D. Cleary. 1980. The smoking problem: A review of the research and theory in behavioral risk modification. Psychological Bulletin 88(2): 370–405. http://doi.org/10.1037/00332909.88.2.370 Loewenstein, G., C. Bryce, D. Hagmann, and S. Rajpal. 2014. Warning: You are about to be nudged (SSRN Scholarly Paper No. ID 2417383). Rochester, NY: Social Science Research Network. Retrieved from http://papers.ssrn.com/abstractD2417383 Malone, R. E. 2010. Imagining things otherwise: New endgame ideas for tobacco control. Tobacco Control 19(5): 349–50. http://doi. org/10.1136/tc.2010.039727 Max, W. 2001. The financial impact of smoking on health-related costs: A review of the literature. American Journal of Health Promotion 15(5): 321–31. McRobbie, H., C. Bullen, J. Hartmann-Boyce, and P. Hajek. 2014. Electronic cigarettes for smoking cessation and reduction. Cochrane Database of Systematic Reviews. John Wiley & Sons, Ltd. Retrieved from http://onlinelibrary.wiley.com/doi/10.1002/ 14651858.CD010216.pub2/abstract Miller, D. 2006. Introduction. In The liberty reader, ed. D. Miller, 1–20. Edinburgh, UK: Edinburgh University Press. Newport, F. 2013. Most U.S. smokers want to quit, have tried multiple times. Gallup Polls. Retrieved from http://www.gallup.com/ poll/163763/smokers-quit-tried-multiple-times.aspx Nichter, M. 2003. Smoking: What does culture have to do with it? Addiction 98: 139–45. http://doi.org/10.1046/j.1360-0443.98. s1.9.x Parrott, A. C. 1998. Nesbitt’s paradox resolved? Stress and arousal modulation during cigarette smoking. Addiction (Abingdon, England) 93(1): 27–39. July, Volume 16, Number 7, 2016 Pattanaik, P. K., and Y. Xu. 2000. On diversity and freedom of choice. Mathematical Social Sciences 40(2): 123–30. Pettit, P. 2002. Keeping Republican freedom simple: On a difference with Quentin Skinner. Political Theory 30(3): 339–56. Pettit, P. 2003. Agency-freedom and option-freedom. Journal of Theoretical Politics 15(4): 387–403. http://doi.org/10.1177/ 0951692803154003 Pettit, P. 2014. Just freedom: A moral compass for a complex world. New York, NY: W. W. Norton & Company. Proctor, R. N. 2013. Why ban the sale of cigarettes? The case for abolition. Tobacco Control 22(suppl. 1): i27–30. http://doi.org/ 10.1136/tobaccocontrol-2012-050811 Saghai, Y. 2013. Salvaging the concept of nudge. Journal of Medical Ethics 39(8): 487–93. http://doi.org/10.1136/medethics2012-100727 Scoccia, D. 1990. Paternalism and respect for autonomy. Ethics 100(2): 318–34. Sen, A. 1988. Freedom of choice: Concept and content. European Economic Review 32(2): 269–94. Sen, A. 1991. Welfare, preference and freedom. Journal of Econometrics 50(1): 15–29. Spielberger, C. D. 1986. Psychological determinants of smoking behavior. In Smoking and society: Toward a more balanced assessment, ed. R. D. Tollison, 89–134. Lanham, MD: Lexington Books. Sulmasy, D. P., and J. Sugarman. 1994. Are withholding and withdrawing therapy always morally equivalent? Journal of Medical Ethics 20(4): 218–24. Sunstein, C. R. 2013. Simpler: The future of government. New York, NY: Simon and Schuster. Sunstein, C. R. 2014. Why nudge?: The politics of libertarian paternalism. New Haven, CT: Yale University Press. Taylor, C. 1979. What’s wrong with negative liberty. In The idea of freedom, ed. A. Ryan, 175–193. Oxford, UK: Oxford University Press. Thaler, R. H., and C. Sunstein. 2008. Nudge. New Haven, CT: Yale University Press. Thomson, G., R. Edwards, N. Wilson, and T. Blakely. 2012. What are the elements of the tobacco endgame? Tobacco Control 21(2): 293–95. http://doi.org/10.1136/tc.2010.040881 Unger, J. B., L. A. Rohrbach, T. B. Cruz, L. Baezconde-Garbanati, K. A. Howard, P. H. Palmer, and C. A. Johnson. 2001. Ethnic variation in peer influences on adolescent smoking. Nicotine & Tobacco Research 3(2): 167–76. http://doi.org/10.1080/ 14622200110043086 VanDeVeer, D. 1980. Autonomy respecting paternalism. Social Theory and Practice 6(2): 187–207. Voigt, K. 2010. Smoking and social justice. Public Health Ethics 3(2): 91–106. http://doi.org/10.1093/phe/phq006 ajob 13 The American Journal of Bioethics Wang, M. P., X. Wang, T. H. Lam, K. Viswanath, and S. S. Chan. 2015. The tobacco endgame in Hong Kong: Public support for a total ban on tobacco sales. Tobacco Control 24(2): 162–67. http:// doi.org/10.1136/tobaccocontrol-2013-051092 West, R., and P. Hajek. 1997. What happens to anxiety levels on giving up smoking? The American Journal of Psychiatry 154(11): 1589–92. Warner, K. E. 2000. The economics of tobacco: Myths and realities. Tobacco Control 9(1): 78–89. http://doi.org/10.1136/tc.9.1.78 Wilkinson, D., and J. Savulescu. 2014. A costly separation between withdrawing and withholding treatment in intensive care. Bioethics 28(3): 127–37. Warner, K. E. 2013. An endgame for tobacco? Tobacco Control 22 (suppl. 1): i3–5. http://doi.org/10.1136/tobaccocontrol-2013-050989 Wolff, J. 1997. Freedom, liberty, and property. Critical Review 11(3): 345–57. 14 ajob July, Volume 16, Number 7, 2016