Shaping people’s preferences: liberal neutrality, means paternalism and tobacco control

Abstract

Liberal neutrality compels governments to respect individual preferences. Yet health-promotion campaigns, such as modern tobacco control policies, often seek to cultivate a preference for a healthy lifestyle. Liberal theorists have attempted to justify these policies by appealing to the concept of ‘means paternalism’, whereby these policies align with existing preferences. In contrast, this article argues that shaping preferences can be not only permissible but also morally required. Governments can preserve neutrality while influencing preferences by promoting generic goods valued in diverse societies and considering the preference-formation of future generations. This argument provides a stronger rationale for tobacco control policies.

1. Introduction

Governments promote healthy lifestyles through various behaviour change policies, encouraging citizens to eat better, drink less or exercise more. When do these policies go too far? Some critics worry about the imposition of lifestyles on those who might reject them, denouncing a form of ‘healthism’ where the promotion of health ‘is no longer a personal yearning but part of state ideology’ (Skrabanek 1994: 15). At the heart of this concern lies an important liberal principle, namely that government should not dictate what people are supposed to want. Indeed, many have even located the ‘constitutive morality’ of liberalism in a neutral stance regarding different possible conceptions of what is valuable in life (Dworkin 1985: 203). From this vantage point, a life spent pursuing Methuselah’s longevity is not inherently superior to one that sacrifices health for other goals.

Of course, most liberals committed to this form of neutrality do not believe their view precludes all health-promoting policies – not even paternalistic interventions that interfere with a person’s choices for his or her good. However, such interventions are typically considered justifiable only when they align with that person’s own ideas of the good. In other words, governments can and should assist people in safeguarding their health if this is something they themselves want but struggle to achieve (for instance due to misinformation, irrationality or weakness of the will). This so-called means paternalism respects individual preferences, which I shall understand here broadly as the ordering of choices that reflect a person’s desires, ends or values. ¹ In contrast, ends paternalism disregards and attempts to change individual preferences. Much of the recent literature on the topic subscribes to means paternalism and rejects ends paternalism. This is exemplified by the non-coercive nudging advocated by Thaler and Sunstein, whose goal it is to ‘make choosers better off, as judged by themselves’ (Thaler and Sunstein 2008: 5). It is equally apparent in more liberty-restricting approaches, such as that pursued by Conly, who aims to ‘do what is good for people according to the way people themselves conceive of their good’ (Conly 2012: 119). ²

But the problem is this: if respecting people’s preferences entails refraining from attempting to change them, then it is not clear that many existing behaviour change policies meet the demands of liberal neutrality. Put simply, efforts to promote healthy lifestyles aim not only at helping individuals satisfy their pre-existing preferences, but they typically attempt to influence and change preferences on a population level. Examples range from ‘active travel’ policies that promote walking over driving, to portion-size regulations for fast food and soft drinks. But perhaps nowhere is the point better illustrated than in the field of tobacco control. In recent decades, public health advocates have pursued increasingly ambitious policies to discourage smoking tobacco. Among the most far-reaching are so-called denormalization and endgame strategies:

• Denormalization aims to change norms surrounding smoking by using public health messages that portray smoking as undesirable, dangerous or irresponsible. Because people form their preferences against the backdrop of social norms, changing these norms can change preferences. Denormalization can deploy traditional measures such as tobacco taxes, or restrictions on cigarette advertisements to reduce the social acceptability of smoking, and in extreme cases even actively encourage social stigma (see Kim and Shanahan 2003; Voigt 2013).

• Endgame proposals attempt to eliminate smoking (almost) entirely. This may be achieved in various ways, including denormalization, but I will use the term here to refer to proposals to ban cigarettes, gradually reduce their supply, or make them unavailable to future generations (see Warner 2013; McDaniel et al. 2016). A recent example is the UK government’s (2023) plan for a ‘smokefree generation.’ These proposals would ensure that children and future people do not form preferences for smoking in the first place.

These tobacco control policies seem to go beyond means paternalism, for their goal is not only to ensure that people truly satisfy their preferences (say, by providing information about the risks of smoking), but to discourage, eliminate and prevent the preference for smoking itself. ³ While they may incidentally serve other interests held by the individuals targeted, the fundamental goal remains the transformation of preferences related to smoking. This raises broader concerns within a framework of neutralist liberalism that requires governments to refrain from promoting any specific conception of the good. Hence defending these tobacco control policies requires an account of the permissibility of shaping people’s preferences, one that can address concerns of ‘healthism’ and identify the limits of health-promoting behaviour change policies. Surprisingly, this challenge has received limited attention in the literature.

This article aims to fill this gap. It argues that shaping people’s preferences can be permissible and even morally required from a liberal perspective. Individual preferences are inevitably influenced by a social environment that must be designed in one way or another, but governments can maintain a commitment to neutrality when they are guided by a notion of generic goods valued in pluralistic societies. A duty to shape preferences in a manner that protects health as a generic good is supported by intergenerational considerations of justice pertaining to the interests of future people. Provided a reasonable balance between competing interests is struck, such a duty may provide a justification for tobacco denormalization and endgame strategies.

This argument makes several contributions to the literature. First, by emphasizing the neglected role of socialization in the process of preference-formation, it problematizes the concept of individual preferences, on which much of the literature’s focus on means paternalism rests. The distinction between means and ends – central to contemporary discussions of paternalism – carries less weight for the justification of health-promotion policies than commonly thought. Second, it provides an alternative conceptual framework to theorize the permissibility of these interventions, including but not limited to the field of tobacco control. Finally, it seeks to revise certain common assumptions about neutralist liberalism itself, highlighting justice-based reasons to promote the good (albeit in strictly limited ways) and reclaiming ground that is all too often ceded to perfectionist views.

The article is structured as follows: First, I examine existing strategies in the literature that justify stringent tobacco control policies by either appealing to harm, or by claiming they align with existing preferences for health. I show that these strategies are not fully successful. Next, I present my own argument for the permissibility of preference-shaping, which is based on the inevitability of shaping the social environment and constructs a neutral case for the promotion of a ‘thin’ idea of the good. Finally, I return to the issue of tobacco control, applying the general argument to denormalization and endgame strategies.

2. Incomplete Arguments for Tobacco Control

Over 8 million people die from tobacco smoking each year – no other behavioural risk factor causes more disability, ill-health and death (WHO 2023: 34). Given the magnitude of these harms, it is not hard to see why public health advocates have pressed governments to take the strongest possible measures to combat tobacco use, including measures that seemingly involve overriding or changing people’s preferences. But can such measures be reconciled with a liberal commitment to respecting people’s own ideas of the good? This section surveys some of the main argumentative strategies in the literature on paternalism. Although they all contain important truths, I suggest that they do not provide a fully adequate answer to the question at hand.

Consider first the argument that smokers are not the only victims of smoking. Indeed, smoking can have significant externalities such as second-hand smoke, environmental pollution or increased public healthcare spending. This suggests that smoking is a preference that governments are permitted to discourage due to its undesirable social costs. Just as we think that it is legitimate for government to cultivate environmentally friendly habits to curb the harms of climate change, smoking may be discouraged on account of the costs it imposes on others. The popularity of this argument no doubt derives from its ability to place demanding, liberty-restricting public health interventions under the auspices of a broadly Millean form of antipaternalism, according to which ‘the only purpose for which power can be rightfully exercised over any member of a civilized community, against his will, is to prevent harm to others’ (Mill 2003 [1859]: 80). So, does the harm principle provide a justification for shaping preferences?

It is not clear that it does. For one thing, the third-party harms associated with a given activity can often be minimized without eliminating the preference for engaging in it. For example, one can ban smoking in public places without adopting more demanding tobacco control policies that discourage the preference for smoking altogether, much in the same way that drink driving laws need not aim at banning or denormalizing alcohol consumption in general. Moreover, unavoidable externalities are often less sizable than the argument assumes. It is a debated empirical question whether smoking in fact imposes costs, is neutral, or even has positive effects on public finances. ⁴ And even if costs were substantial, these could be internalized, for instance through higher insurance premiums paid by smokers. Granted, it is not always possible to minimize or internalize all costs. Yet we often accept some level of harm to protect individual spheres of agency: we tolerate a certain degree of noise pollution, for example, because discouraging all noisy activities would eliminate freedoms that we value. Strictly and consistently applied, the harm argument would therefore likely make many more preferences targets of government intervention than a liberal society would be ready to accept (see also Heath 2020: 319–321).

A second set of arguments deny the premise that behaviour change interventions in the field of tobacco control in fact attempt to change people’s preferences. Instead, denormalization and endgame approaches could be simply helping people realize their own goals, thus falling under the remit of means (and not ends) paternalism. Even Mill (2003 [1859]: 158) thought that we would be justified in forcibly preventing an unsuspecting passer-by from crossing a broken bridge, ‘for liberty consists in doing what one desires, and he does not desire to fall into the river’. One way to make this argument in the case of smoking is to invoke the idea of latent preferences: despite their revealed preferences – that is, the preferences that can be inferred from their behaviour – smokers may have a ‘higher-order’ desire to stop smoking (Goodin 1991). This argument can be bolstered by pointing to surveys and other empirical research that shows that most smokers wish they could quit, and regret having started in the first place (for discussion of this evidence, see Wilkinson 2021). If smokers truly prefer not to smoke, then tobacco control efforts would not be attempting to change individual preferences; rather, they would be assisting in their satisfaction.

It is hard to disagree with the proposition that revealed and latent preferences can sometimes conflict, but what follows from it is far from obvious. In fact, smokers’ preferences may be inconsistent in more ways than one. The wish to stop smoking may coexist with a preference that government should not discourage or ban smoking; by helping realize the former, government may be frustrating the latter (Grill and Voigt 2016: 298). Moreover, when a desire to stop smoking conflicts with a desire to continue, it is not clear that government should take the former to be more authoritative or ‘higher-order’. While it is tempting to treat survey answers as considered judgements, these answers may often simply reflect what respondents perceive to be socially acceptable, rather than their true or deeper preferences (Whitman and Rizzo 2015: 421). What is more, even when latent preferences to stop smoking are genuine, the formation of these preferences may be partly explained by public health efforts that reduce the appeal of smoking, such as taxation and stigmatization. Presumably some people would not express regret about smoking if smoking were still as cheap and publicly acceptable as it was only a few decades ago. Hence, governments might be justifying paternalistic policies by appealing to preferences they have helped shape in the first place, and it would seem to bend the meaning of the term to consider this a form of means paternalism.

A third argument goes further: because people typically form their preference for smoking under conditions that do not guarantee autonomy or voluntariness, government is justified in overriding these preferences. Consider the fact that most smokers pick up the habit as children or teenagers, when informed consent can hardly be assumed. And if we assumed it, the highly addictive nature of tobacco would still cast doubt on the voluntariness of the choice to continue smoking. Even among adults, well-known cognitive errors and biases – the tendency to engage in wishful thinking, the propensity to discount future risks, etc. – suggest a degree of irrationality in the formation of smokers’ preferences (Goodin 1989; Thaler and Sunstein 2008; Conly 2012). These observations can be taken to support the view that government should help satisfy ‘reconstructed’ preferences, that is, the preferences that people would hold if they were perfectly rational, fully informed, entirely autonomous and so on. Under these circumstances, smoking would not be valued more highly than health, longevity and other important goals. This argument goes further than the last because in this case, paternalistic interference is based on what we could call hypothetical latent preferences.

Again, the question is whether this is still a form of means paternalism. In Mill’s famous bridge scenario, intervention is justified to correct a simple case of misinformation: once the passer-by is informed about the danger of the broken bridge, he immediately acknowledges that being prevented from crossing it advanced his own goals. By contrast, the argument we are considering here goes much further because the gap between expressed and assumed preferences is significantly larger. When that gap it is too large, paternalistic subjects may simply not recognize the reconstructed preferences ascribed to them as their own (see Brännmark 2023: 50, citing Rosati 1995). Indeed, we often take preferences that are less than fully informed and rational – say, our choice of career or romantic partners – to be central to our idea of the good, and we would reject any attempt by government to ‘correct’ the errors in their formation. Granted, preferences to smoke will rarely be as central to a person’s life plan. But it seems equally hard to argue that by appealing to highly idealized, reconstructed preferences we are promoting a person’s good by their own lights, as opposed to by some external or less subjective standard.

What if, after initial resistance, smokers do eventually come to see tobacco control measures as promoting their own ends? This brings us to a fourth argument to place these measures under the banner of means paternalism. Interventions may be justified on the basis of ex post preferences, that is, the preferences that people come to hold as a result of the intervention. Like the passer-by in Mill’s broken bridge scenario, smokers may retrospectively welcome the intervention – perhaps not immediately, but after some period of adaptation. Studies indeed suggest that a large majority of former smokers are happy they stopped (Shahab and West 2009). Hence, tobacco control measures could be seen as a form of paternalism that respects people’s preferences, albeit those they are yet to endorse.

However, there are several issues with the argument. First, it is not clear that the empirical evidence supports the premise that ex-smokers are happier when they have been made to stop, as opposed to when they have stopped of their own accord. But even leaving such practical worries aside, what is missing is an explanation for why people’s later preferences should be more authoritative than their present ones. Surely the justifiability of a preference-shaping intervention cannot derive solely from its success. If that were the case, government could force or nudge people to adopt the official religion, say, and claim that satisfaction with the change at some later point in time justified the policy. At the very least, then, the argument would need to explain why appealing to future preferences is permissible in some cases, even when it seems off-bounds in so many others.

All the argumentative strategies I have sketched contain valuable insights, and my aim is not to dismiss them entirely. Instead, I want to suggest that while some of these arguments may be compelling in their own right, they do not effectively show that tobacco control measures either do not involve shaping preferences, or that they do so in ways that are consistent with the idea of means paternalism.

3. Preferences and Neutrality

Given the difficulties of placing tobacco control measures under the umbrella of means paternalism, one might conclude that they run afoul of neutralist liberal commitments. I want to resist this conclusion. The focus on means paternalism, I will suggest, is misleading because it relies on an incomplete account of what it means for government to respect people’s preferences, and a simplistic understanding of preferences themselves. Once we begin to unpick these ideas, it becomes clear that it is not generally impermissible for liberal states to shape people’s preferences; indeed, they are even required to take an active interest in the development of people’s desires, values and goals. The rationale for tobacco control policies is best understood in this light, or so I will argue.

The literature on paternalism is full of examples where goals and values are imposed on people against their will: an adult’s innocuous drug habit is criminalized, a woman seeking an abortion is forced to undergo assessments designed to dissuade her, a terminal cancer patient is denied assisted suicide. In cases like these, we assume that individuals have clearly defined and settled preferences which are subsequently frustrated, overridden or substituted. As the previous discussion indicates, it is not always easy to identify an individual’s true preferences. The same individual may hold different preferences at different points in time, or she may have goals and desires at different levels of generality and specificity (that is to say, some ends are means to further ends). Yet despite these complications, there persists a presumption that it is possible to identify preferences that can serve as the basis for means paternalistic interventions. And the crucial criterion is that these preferences are genuinely those of the individual herself, as opposed to preferences imposed externally. Only when they are authentic can a liberal government claim to respect people’s own ideas of how to live their lives.

This picture assumes that it is straightforward to ascertain the authenticity of preferences. It simply takes for granted that a clear line can be drawn between our own ideas about how to lead our lives, and those of others. This assumption is almost axiomatic in much of the field of economics. Neoclassical economists assume that people act rationally to satisfy their preferences, where preferences in turn are understood as stable and independent ‘primitives’ whose origin is not scrutinized (Stigler and Becker 1977; on preferences as primitives, see Angner 2018). Behavioural economists have chipped away at assumptions of rationality, but the underlying notion of preference remains much the same. ⁵ Put simply, ‘most economists have not asked how we come to want and value the things we do’ (Bowles 1998: 75). In truth, of course, preferences are not formed in a vacuum. Individuals form their own views of what matters in life against a complex background of social relations, norms, institutions and arrangements. This is not a controversial observation in most social sciences. As Barry (1965: 75) once remarked, ‘the dependence of wants on the social environment is a commonplace of sociology, which could almost be defined as the intellectual consequence of taking it seriously’.

Paying attention to the formation of preferences sheds light on what the liberal commitment to neutrality can and cannot require. It is tempting to oppose any official influence on people’s ideas of the good, but closer reflection shows that this is an impossible standard. If preferences are influenced by the social environment and the social environment in turn is influenced by politics, then preferences cannot be exogenous to politics. Consider some of the ways in which preference-shaping is unavoidable. At the most fundamental level, governments shape preferences by setting the legal, political and economic frameworks within which people pursue their aims. The permissions and prohibitions defined by these frameworks create expectations of entitlements, and with them, views about what is worth pursuing. ‘Whether people have a preference for a commodity, a right or anything else’, Sunstein (1991: 8) observes, ‘is in part a function of whether the government has allocated it to them in the first instance’. Various psychological mechanisms are at play here, such as the ‘exposure effect’ that leads us to like what we know, or the ‘endowment effect’ that leads us to attach higher value to what we already have (see Lichtenstein and Slovic 2006). Existing institutions and arrangements also influence preference-formation by setting social norms, harnessing the human desire to conform to them. Norms affect individual motivations by creating incentives and disincentives, effectively acting as taxes on preferences that diverge from social expectations (Lessig 1995; Sunstein 1996). These mechanisms together influence broader cultural learning processes, by transmitting values, desires and goals across generations through processes of socialization (Wildavsky 1987; Bowles 1998).

Consider, for example, the legalization of same-sex marriages. Recent changes in various jurisdictions have affected public opinions, with decreased homophobic attitudes being partly explained by the way policies, laws and institutions legitimate and naturalize same-sex partnerships (Flores and Barclay 2016; Tankard and Paluck 2017). In other words, whether someone holds preferences to associate or dissociate with gay and lesbian people is partly a result of the existing legal framework. Governments influence preferences on a wide range of issues that can be regulated in more ways than one, ranging from genetically modified foods (GMOs) to gun ownership. The broader point is that rules must be set one way or another: buying GMO foods and carrying guns are either made easy or difficult for consumers by the chosen legal and regulatory frameworks. These rules have an expressive dimension that shapes social norms, and these norms end up influencing what people come to view as their own preferences. This is the sense in which preference-shaping is unavoidable: if ‘ought implies can’, neutrality of effect seems a hopeless ideal. But, of course, this does not absolve government from seeking neutrality in its aims and justifications (Rawls 1996: 191–195; Quong 2011: 18).

At this point, a critic might object that there is an important difference between a government that acknowledges the passive effect of the social environment on human development, and one that actively attempts to inculcate and instil certain preferences. Indeed, the significance of the social environment might be taken as precisely the reason to insist on minimal official intervention into said environment. On this view, government should limit itself to establishing a minimal institutional and legal framework that defines basic entitlements and rights, leaving the rest to the organic interplay of individuals in the family, civil society and – perhaps most importantly in advanced economies – the market. Having established that framework, government retreats from the business of shaping people’s preferences and lets the chips fall where they may. People turn out the way they turn out, and government may at most help them attain what they already want. Call this the laissez-faire view.

It is worth emphasizing that this view does not amount to the rejection of preference-shaping as much as the choice for it to occur in one particular way. There are important differences between a highly interventionist government and one that takes a laissez-faire approach, but the latter must be justified just like the former. Any choice of institutional design has consequences for preference-formation: there is no natural default. The market, in particular, is not simply a neutral mechanism to discover and satisfy pre-social preferences. As Rawls (1971: 259) notes when discussing the design of the basic structure of society:

an economic system is not only an institutional device for satisfying existing wants and needs but a way of creating and fashioning wants in the future. How men work together now to satisfy their present desires affects the desires they will have later on, the kind of persons they will be … Since economic arrangements have these effects, and indeed must do so, the choice of these institutions involves some view of human good and of the design of institutions to realize it.

The choice of a laissez-faire framework needs to be justified just like any other form of preference-shaping. But closer reflection reveals several problems, which the case of tobacco exemplifies. For one thing, the laissez-faire approach can leave people vulnerable to the preference-shaping efforts of other, more powerful parties. It is well-known that large corporations expend considerable resources to exploit the informational and cognitive shortcomings of consumers, thereby manipulating preferences for private gain (e.g. Akerlof and Shiller 2015; on ‘market nudges’ see Ivanković and Engelen 2024). The tobacco industry is a case in point, having invented much of modern marketing, from colour billboards to product placement in films and television (Proctor 2011: 59). These marketing efforts encourage the formation of life-long habits by appealing to the emotional and psychological needs of different consumer groups, for example associating smoking with daredevilry among boys and young men, or empowerment among girls and young women. Liberals may be suspicious of the idea that government take an active role in shaping preferences because they value a sense of authenticity or independence in the way people choose and pursue their life plans (e.g. Dworkin 2011: 211). But it is precisely this value that we jeopardize when we allow powerful market actors to shape people’s preferences for their own ends.

What is more, preferences that are formed through market interactions can lead to harmful outcomes, both collectively and individually. Few deny that government should help solve collective action problems and market failures. In the face of environmental harms, for example, governments attempt to inculcate norms against littering and wastefulness, and to promote preferences for environmentally conscious forms of consumption. Similar considerations regarding collective harms and externalities may justify interventions in markets for tobacco products, including marketing restrictions or smoking bans in public places. It is also plausible that government should sometimes intervene in the process of preference-formation for the sake of individuals themselves. Consider the problem of ‘adaptive preferences’, where people form their preferences against an unjustly constrained set of options (Elster 1983; Sen 1987; Nussbaum 2001). Classic examples include people who reject the value of education because they have never had the opportunity to experience it, or women who adapt their expectations to what is feasible in a patriarchal society. These preferences, which undermine the attainment of important goods, are formed under conditions that do not guarantee voluntariness or autonomy. The same is true in the case of many smokers: smoking has been identified as a mechanism to deal with stress among socially disadvantaged groups, it is highly addictive, and many smokers underestimate its risks in an attempt to reduce cognitive dissonance (Goodin 1989: 582). Even if one disagrees with attempts to override or frustrate these kinds of preferences, it is hard to deny the desirability of reforming social and economic institutions so that people do not develop these preferences in the future.

This is hardly an exhaustive list of problems with the laissez-faire view. But it suffices to show that removing government entirely from the process of preference-formation is unlikely to be a desirable solution, even if it were possible. To be sure, none of this is to suggest that government should always take a highly interventionist role in the process of individual preference-formation. It may well be that leaving markets, families and civil society to their own devices is, for the most part, the most justifiable course of action. My point is that a requirement to justify the role of government remains whether it is active or passive actor in this process. Because it is responsible for setting the institutional framework through which the social environment shapes preferences, it must also assume responsibility for the outcomes. Hence the question is not whether it should avoid shaping people’s preferences, but rather how it can do so in a way that is consistent with liberal principles.

4. Liberal Preference-shaping

When and how should a liberal government intervene in the process of preference-formation? It is not disputed that intervention can be warranted to discourage preferences detrimental to the justice and stability of a democratic society (say, desires to harm others). What is controversial is the idea of shaping people’s preferences to advance the good. Those sympathetic to this idea are often associated with perfectionism, the view that government should create the conditions for human flourishing by promoting some objective account of value. Liberal perfectionists champion freedom – in particular the ideal of an autonomous life – and argue that individuals should be helped to lead worthwhile lives in line with this ideal. Neutralists, on the other hand, insist that government should not take a stance on what makes one way to live more worthwhile than another. Hence, it should justify its actions without appealing to any contested conception of the good. I will not delve into this debate (for an overview, see Quong 2011: Ch. 1). Instead, I want to show that even if we confine ourselves to neutralist premises, government must be in the business of shaping preferences.

A government committed to neutrality should not promote any particular conception of the good, but it cannot avoid questions regarding the good altogether. One reason for this is the need to compare individuals and social groups to assess claims of distributive justice. Who is relatively disadvantaged in society? What do they lack that others have? And why does it matter? It is difficult to answer these questions without presupposing the value of certain goods and conditions. It is possible, however, to create a public standard of value that people with divergent views of the good life can agree on. Rawls’ theory of justice solves the problem by proposing a ‘thin’ theory of the good endorsed by contract parties behind a ‘veil of ignorance’ that obscures their own identities. This enables the construction of an index of primary goods, understood as ‘various social conditions and all-purpose means that are generally necessary to enable citizens adequately to … pursue their determinate conceptions of the good’ (Rawls 2001: 57). Primary goods include liberties, opportunities, wealth, health and intelligence. Another option, first proposed by Sen (1980), is to appeal to capabilities, understood as real freedoms to be or do things such as being well-nourished, receiving an adequate education, enjoying good health and so on. If these capabilities are not conceptualized as elements of human flourishing but as political demands, they similarly serve to identify goods that allow each individual to pursue their own views of what matters in life, without taking a position on the merit of these views.

These generic goods, as I will call them, serve not only to assess claims of distributive justice that citizens make on one another; they also provide the yardstick for liberal preference-shaping. Given that government inevitably influences preferences through the design of fundamental social arrangements, it must ensure the protection of generic goods and conditions essential for any reasonable life plan. This involves intervening in the process of preference-formation whenever these goods are under threat. For instance, if the free market allows corporations to manipulate consumers into favouring unhealthy foods, these preferences may be checked through education campaigns, regulation or taxes. If existing infrastructure overly promotes sedentary lifestyles, ‘active travel’ policies may encourage preferences for healthier forms of transport. These measures need not assume the superiority of certain preferences or conceptions of the good. Nor do they require imbuing health (or any other generic good) with intrinsic value, encouraging people to think of it as a goal to be pursued for its own sake. Instead, they can be justified as interventions into the process of preference-formation to safeguard holdings of generic goods that people have reason to value for any specific life plan. ⁶

This argument for preference-shaping has an intergenerational dimension. Social arrangements persist over time, so they must be justifiable not only to currently existing people, but also to the people who come to be. Much of the literature on intergenerational justice has focused on questions of distributive justice among generations, for instance the just rate of saving and the availability of resources across time. Sometimes neglected is the fact that our choices in the present also influence the kinds of people who will exist in the future – their desires, values and goals. The two questions are intertwined. For example, neglecting investments in education today may limit opportunities for future generations. But by the same token, it would be unfair to encourage future generations to form preferences the satisfaction of which would require more resources than justice would allow them. Preferences that overly prioritize material consumption, for example, may not be environmentally sustainable within the constraints of a fair social system. Hence, to ensure each person has a fair opportunity to pursue their individual conception of the good, government must help shape preferences in a way that secures the all-purpose means that enable this pursuit. Consider some historic examples of health-limiting social practices: footbinding, using arsenic as a skincare treatment, drinking methanol-based spirits. Governments sought to discourage these practices not only for the good of the individuals who engaged in them, but, perhaps more importantly, for those who would have otherwise formed preferences for them. The denormalization of these practices thus removed significant health risks that undermined the pursuits of future people. The same reasoning plausibly underlies contemporary behaviour change efforts in the field of public health.

Of course, none of this guarantees that preference-shaping interventions will be uncontroversial. Few people reject freedom, wealth, health or leisure in the abstract, but in practice they may weigh their relative importance differently. It follows that any given intervention aimed at inculcating preferences for a healthy lifestyle may invite the charge of ‘healthism’ despite protecting a generic good. I address this concern in more depth in the following section. The point I wish to emphasize here, however, is that worries concerning ends paternalism or perfectionism would not be among the reasons to reject the intervention. Liberal neutralists often object to efforts to interfere with how individuals form, prioritize and pursue their ends because such interventions may imply a lack of confidence in individuals’ capacity to make these decisions competently (Quong 2011: 100–101). But preference-shaping interventions need not express a judgement of any given individual’s ends, let alone their ability to choose the right ends (in the case of future people, there aren’t even formed ends that could be the object of such a judgement). Instead, they simply reflect an account of the goods and conditions that individuals tend to require in a fair society. Compare: interventions aimed at discouraging unfair gender norms need not involve a judgement of any individual woman’s choice to exercise a traditional family role; instead, they may simply follow from a requirement of justice to ensure that social arrangements uphold a conception of citizens as free and equal. Similarly, discouraging preferences that undermine the realization of generic goods need not constitute an official statement of what is worth pursuing in life. ⁷

Granted, like paternalistic interventions, these preference-shaping efforts can affect people’s ability to form their life plans independently and autonomously. However, it is not controversial that such interferences can be justified in some circumstances. In the case of children, even ends paternalism is commonly thought unobjectionable because parents are said to have wide-ranging prerogatives to instil preferences – for example, religious commitments – that they judge will make their children’s lives go well. It is also widely held that society at large must play a role in the ethical development of children in ways that may restrict parental rights. For example, governments are justified in intervening when the inculcation of religious commitments severely undermine children’s lifetime expectations for generic goods such as health. In the case of adults, autonomous agency raises thornier questions, and several factors will affect the justifiability of shaping preferences. For one thing, it matters how important the targeted preferences are likely to be within the overall structure of people’s conceptions of the good. For example, religious commitments tend to reflect much deeper values than consumer preferences, and a health-promoting intervention that influences the former faces a higher justificatory bar than one that targets the latter. The means by which preference are shaped also matter. Interventions can be coercive (e.g. bans, taxes) or persuasive (e.g. incentives, nudges), and they can appeal to reason (e.g. information, education), or attempt to bypass rational agency (e.g. subliminal messaging, manipulation) (for a related discussion, see MacKay 2019: 444–445). I do not aim to fully resolve the question of where to draw the line between permissible and impermissible interventions. But other things being equal, showing respect towards autonomous agents clearly requires the least intrusive and most transparent means.

5. Shaping Preferences Through Tobacco Control

I argued that we can construct a neutral, non-perfectionist argument for preference-shaping built around the idea of generic goods. Tobacco control policies aimed at discouraging unhealthy preferences are plausible examples of this form of preference-shaping. But as I mentioned before, the fact they protect a generic good cannot be sufficient to demonstrate their permissibility. People have interests in health, but also in all sorts of other generic goods, such as liberties and freedoms, material prosperity, social status and so on. Hence it is not enough to base specific policies on a neutral footing. It also needs to be shown that the balance between the different competing interests is justifiable, and that no other moral constraints are violated in the pursuit of the good in question. After all, the ‘healthism’ worry is not just that the state promotes health, but that it elevates it above all other concerns.

What is required, then, is a judgment on whether the proposed intervention strikes a reasonable balance among competing goods. Various considerations need to be reflected in this judgement, including how much the intervention advances or undermines various types of goods; how central the affected preferences are likely to be within the overall structure of people’s conceptions of the good, the nature of the means to bring about the intervention, and various pragmatic considerations surrounding the effectiveness of the intervention and the risk of unintended outcomes. In practical terms, a plausible way to proceed is to examine the preference-shaping policies in question and consider them from the point of view of all affected parties. By examining how the adoption of these policies would affect the interests of social groups impacted by them, we assess the strength of potential complaints against the policy, and thus its overall justifiability. Now, abstract normative reasoning may not always provide a definitive answer as to whether a given intervention is justified. The relevant considerations may be indeterminate, and different societies may reasonably attach different weights to them in their specific local contexts. Such hard cases require democratic consultation on collective priorities (Kniess 2019: 419–420). Even here, however, a general normative analysis can reveal the merits of the relevant considerations and help define the range of justifiable policies. Although a comprehensive analysis for the case of denormalization and endgame proposals is beyond the scope of this article, it is worth sketching the relevant considerations as they apply to the three primarily affected groups: non-smokers, future people and existing smokers. ⁸

1.1. Non-smokers

Tobacco control policies are frequently justified with reference to third-party effects, in particular the impact of smoking on the interests of non-smokers. The burdens imposed on non-smokers are said to include higher healthcare costs, reduced productivity in the economy, and exposure to second-hand smoke and other environmental harms. By alleviating these burdens, tobacco control policies protect interests in generic goods such as health and wealth that benefit not only smokers, but all members of society, including non-smokers. As I suggested in section 2, how significant these benefits are is an empirical matter. It seems unlikely that they alone would justify some of the more demanding endgame proposals, such as banning all forms of tobacco use. This is because any harm inflicted on non-smokers could arguably be significantly limited without completely eliminating the use of tobacco and the underlying preferences of smokers.

However, non-smokers typically do not have very strong interests in the continued availability of tobacco products either. It can be argued that they have an abstract interest in freedom of choice, which would include the option to start smoking. Endgame proposals would eliminate this option (at least for those born after the specified cutoff date), and denormalization would make it significantly costlier. Yet the value of this option is likely to be negligible. Smoking is not a particularly ‘fecund’ choice, that is, it does not normally increase freedom by enabling other choices (cf. Feinberg 1980: 36; on the abstract value of the choice to smoke, see also Schmidt 2022: 80). It is also worth noting that very few adult non-smokers make use of the option to start smoking. ⁹ Indeed, if we imagined that cigarettes had been invented only recently, so that the only affected parties were non-smokers, then it seems clear that regulatory agencies would not allow them on the market: the value of the option to smoke is very unlikely to outweigh the associated risks and costs (Schmidt 2016: 3). Even if freedom of choice is an important generic good, then, tobacco control is not a significant threat to it. In short, denormalization and endgame proposals are comparatively easy to justify to the group of non-smokers.

1.2. Future people

Some of the same considerations apply to future people. Future potential smokers would be deprived of the opportunity to smoke as a result of endgame policies. But like current non-smokers, future people have not (yet) formed preferences for smoking. Thus, it is not an option they are attached to. It is true that at least some future people will value the option to smoke, and therefore would have a potential complaint against efforts to eliminate it. But if smoking were no longer associated with its contemporary social meanings, this would likely be a small minority (Grill and Voigt 2016: 300–301) – perhaps akin to those who today regret that duelling or driving without a seatbelt are no longer legal. In addition to taking away freedom, preference-shaping aimed at discouraging smoking may also limit the autonomy of future people, that is, their ability to form their preferences without outside pressure or interference. However, as argued above, without tobacco control policies, this ability would also be constrained, for example through tobacco advertisements or social norms that make smoking desirable. Hence, endgame and denormalization proposals would not significantly reduce the freedom or autonomy of future people.

On the other side of the ledger, future people have very weighty interests in changes to the social environment that would minimize or eliminate the possibility of developing preferences for smoking. Without more stringent tobacco control policies, it is likely that many future people would be tempted to pick up the habit – most likely as children or teenagers, when voluntariness or autonomy are not guaranteed. Many would come to regret it. In addition to interests in the generic good of health, there are financial considerations: smoking can be an expensive habit, and tobacco control can increase expectations for wealth and material prosperity for future people. The magnitude of these benefits is hard to overstate when the effects of tobacco control policies are projected several generations into the future: there are simply many more future people than existing people (Grill and Voigt 2016: 300–301). Thus, if endgame and denormalization proposals can be justified to current non-smokers, future people seem to have even stronger reasons to endorse this form of preference-shaping.

1.3. Current smokers

Unsurprisingly, the interests of current smokers are most directly impacted by tobacco control policies. We can begin by pointing out that the same generic interests in health and wealth matter for smokers too. Now, some are committed smokers who value the freedom to smoke over these goods. Many others would like to stop smoking, suggesting they value these goods (and others associated with smoking cessation) above the benefits derived from smoking. Even in this latter group, admittedly, many continue to attach some value to smoking. And clearly, abandoning any addictive habit will often involve considerable pains of deprivation. Hence, unlike with non-smokers or future people, it is not enough to argue that the abstract value of the option to smoke is low. Moreover, some may argue that these tobacco control policies are not only taking away the option to smoke, but also valuable lifestyles associated with it. It is sometimes said, for example, that the smoking ritual can play an important role for bonding and identity-formation in friendship groups, youth cultures or certain occupations. ¹⁰ Worries over autonomy also loom much larger than in the case of future people: for those who have already developed preferences, constraints on self-direction are felt much more acutely. Finally, they also raise issues of self-respect. Although discouraging preferences held by current smokers need not involve a perfectionistic or paternalistic judgement, it risks undermining the social status of members of this group, as well as their confidence in their own capacity to choose their life plan.

However, the weight of these considerations will depend on the precise nature of the tobacco control measures in question. For example, endgame proposals that ban the sales of cigarettes after a given date of birth do not affect their availability to those born before. While this may reduce the value of the option to smoke – for example, there may be fewer social settings where it would be common or socially acceptable – it does not eliminate it, thus preserving freedom of choice. This significantly reduces the strength of the complaints that committed smokers can mount. Or consider denormalization proposals and their expressive effects on smokers’ self-respect. These effects could be moderated by placing the focus on social norms and perceptions of the tobacco industry, rather than a moralized discourse that condemns the behaviour of individuals (Voigt 2013: 11–13). These strategies can also be combined with harm-reduction strategies that show care towards smokers, for instance by providing support for cessation. In short, the strength of reasonable complaints by current smokers depends on the details of the policy.

Nevertheless, it remains the case that tobacco control policies will negatively impact important generic goods that some smokers will value above others. How can such policies be justified? Ultimately, if they are justified, it will primarily be on account of the interests of future people, who, depending on how many generations we count, would outweigh those of existing smokers. Beyond the mere aggregation of interests, there is also an important question of distributive justice: because smoking typically follows a social gradient, the benefits in health and wealth would primarily accrue to the disadvantaged (Voigt 2010: 91–106). Although I have here only sketched the argument, partly to exemplify the method, it seems plausible that in light of these considerations the tobacco control policies in question can be justified as form of liberal preference-shaping. None of the considerations I have discussed here is itself novel, but it is worth highlighting that I have here invoked them not as justifications for paternalistic intervention, that is, interventions that override people’s choices for their own good. Rather, I have invoked them as reasons for governments to intervene in the process of preference-formation, independently of whether any person’s preference can be overridden or frustrated for her own good.

6. Conclusion

Many liberals believe that government should be, in some relevant sense, neutral among competing views of flourishing or the good. A government guilty of ‘healthism’ violates a commitment to neutrality by privileging one way of life over others – namely one that values health above all. From the vantage point of those who press this objection, efforts to denormalize and eliminate smoking objectionably enlist the institutions of the state to create ideal citizens who endorse the ‘correct’ values. Thus, tobacco control measures appear in tension with liberal neutrality. Some theorists have argued that these measures can be justified by reference to the preferences of those targeted by them, thus constituting a form of means (but not ends) paternalism. But appeals to people’s latent, reconstructed or ex post preferences seem mostly unconvincing, or not true instances of means paternalism. A more honest starting point is to accept that the measures in question are attempts to change people’s preferences. Yet this raises difficult questions about how tobacco control can be reconciled with liberal commitments to neutrality.

By shining a light on processes of preference-formation, this article sought to show that the tension is not as great as might initially appear. Since individual preferences are inevitably influenced by the social environment, governments cannot avoid – and in fact often must – play a role in the process of human development. This can be justified if they are guided by a notion of generic goods valued by people in pluralistic societies. A duty to engage in liberal preference-shaping is supported by intergenerational considerations of justice and the interests of future generations. However, any given instance of preference-shaping must strike a reasonable balance between competing goods, even if the importance of any single good can be justified from a neutral perspective. Hence, the challenge is to show that tobacco denormalization and endgame strategies strike such a reasonable balance. I have illuminated the kinds of considerations that the argument would need to engage, mounting an initial case in favour of their justifiability. A conclusive argument requires more in-depth analysis, as does the evaluation of different public health issues that do not share the same features as tobacco (including its highly addictive nature, extreme health risks and socioeconomic effects). Nevertheless, by shifting the focus away from means paternalism and the notion of preferences that informs it, the general approach suggested here provides a more fruitful framework to continue to test the justifiability of behaviour change interventions in the field of public health.