Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Paternalistic interventions restrict individuals’ liberty or autonomy so as to guide their decisions towards options that are more beneficial for them than the ones they might choose in the absence of such interventions. Although some philosophers have emphasised that there is a case for justifiable paternalism in certain circumstances (e.g., De Marneffe 2005; Wilson 2011), much of contemporary moral and political philosophy works from a strong presumption against paternalistic interventions: because paternalistic interventions restrict individuals’ liberties and treat them as less than fully capable of making decisions that are in their own best interest, they are generally considered impermissible, barring very exceptional circumstances.

Richard Arneson has argued that there are egalitarian reasons that support the case for paternalism: paternalistic interventions can protect poor decision-makers from making ‘bad’ choices, thus preventing inequalities between them and those with better decision-making skills. This line of argument can be applied to a range of contexts. For example, paternalistic restrictions on participation in biomedical research have been supported by concerns about equality: differences in individuals’ decision-making capacities result in an unfair distribution of the costs and risks associated with participation in such research.

This paper aims to clarify and advance our understanding of the egalitarian argument for paternalism. Arneson’s argument adds an important and often neglected dimension to the debate about paternalism. However, the argument also raises a number of questions about equality, paternalism and the relationship between the two.

I begin by restating Arneson’s argument (Sect. 6.2) before highlighting a number of complexities surrounding it (Sect. 6.3). First, with respect to what kinds of choices does Arneson’s argument hold? Second, what kinds of outcomes should we be concerned with when assessing whether or not a particular intervention is really in the interest of the person interfered with? Third, what types of paternalistic interventions lend themselves to Arneson’s argument? Section 6.4 reconsiders the concern that paternalistic interventions treat as less than equal those whose liberties they restrict. Arneson’s argument, with its focus on ‘good’ and ‘bad’ choosers seems particularly susceptible to this kind of concern. Any gains in distributive equality that can be garnered from paternalistic interventions must be weighed against possible negative effects on these ‘relational’ aspects of equality. Section 6.5 concludes.

2 Paternalism and Its ‘Distributive Dimension’

Like many discussions of paternalism, I will start from the definition proposed by Gerald Dworkin. According to this definition,

  • X acts paternalistically towards Y by doing (omitting) Z:

  • Z (or its omission) interferes with the liberty or autonomy of Y.

  • X does so without the consent of Y.

  • X does so just because Z will improve the welfare of Y (where this includes preventing his welfare from diminishing), or in some way promote the interests, values, or good of Y. (Dworkin 2011)

In two papers, Arneson (1989b, 2005) connects paternalism with considerations of distributive equality, emphasising that we should recognise what he calls the ‘distributive dimension’ of paternalism. He starts from the observation that there are significant variations in individuals’ decision-making skills: ‘people differ widely in their native capacities for deliberation about plans and skilful execution of them’ (Arneson 1989b: 412). Starting from a baseline of equality, if ‘bad’ options are available for people to choose, then these systematic differences in people’s decision-making skills will lead to inequalities between ‘good’ and ‘bad’ choosers, as the ‘bad’ options can be avoided by the good but not by the bad decision-makers:

A ban on paternalism in effect gives to the haves and takes from the have-nots. Left unrestrained in self-regarding matters, more able agents are more likely to do better for themselves choosing among an unrestricted range of options, whereas less able agents are more likely to opt for a bad option that paternalism would have removed from the choice set. (Arneson 1989b: 412)

Thus, paternalistic interventions that remove particularly bad options from people’s option sets can prevent inequalities. Anti-paternalism, Arneson concludes, ‘looks to be an ideology of the good choosers, a doctrine that would operate to the advantage of the already better-off at the expense of the worse-off, the needy and vulnerable’ (Arneson 2005: 276).

How might this argument work in practice? One application, which I will keep coming back to in this paper, is suggested by Jansen and Wall (2009), who focus on individuals’ participation in clinical research trials. Jansen and Wall defend hard paternalism in the case of trials that have an unfavourable balance of risk and benefit (in the case of therapeutic trials) or that impose more than minimal risk on participants (in the case of non-therapeutic trials). With respect to such trials, they argue, we must not simply rely on individuals’ consent when it comes to their participation as research subjects; a hard paternalistic approach to trial participation is justified because of considerations of fairness.

For the most part, their argument follows Arneson’s (though I discuss below one issue on which they deviate from Arneson). Like Arneson, Jansen and Wall emphasise that people differ in their decision-making skills and capacities:

Some are wise, some are foolish; some are careful in their deliberations, some are rash. Some are subject to cognitive and emotional deficiencies of one type or another, some are relatively free from such deficiencies. Some are too trusting of authority, some are too independent minded. And some are better able than others to take in and process the information relevant to the decision to participate in a given trial. (Jansen and Wall 2009: 176)

Thus, different approaches to restrictions on clinical research differ in their impact on potential research participants: stringent restrictions will benefit those with lesser decision-making skills, while loose restrictions will advantage those whose decision-making capacities are greater. Because those with lesser decision-making skills are likely to be (or become) worse off overall, considerations of distributive equality are relevant. A concern with distributive fairness, Jansen and Wall argue, ‘will require us to compare the likely distributive outcomes (in terms of the welfare impact on the population of potential research subjects) of different regulatory policies’ (Jansen and Wall 2009: 175).

The fact that paternalism can, as this line of argument suggests, be equality-promoting gives us one reason to support paternalistic interventions. At the same time, of course, equality is not the only consideration at stake and any distributive advantage of particular paternalistic interventions may well be outweighed by other relevant considerations. The objections to paternalism – that it involves undue interference with individuals’ liberties and that it interferes with individuals’ authority over important aspects of their lives – still stand and are not necessarily outweighed by distributive concerns. However, Arneson’s argument emphasises the distributive implications of paternalistic interventions as an important – and often neglected – consideration that must be weighed against other, possibly competing concerns when evaluating the case for particular paternalistic interventions.

3 The Scope of Arneson’s Argument: Choices, Outcomes and Interventions

In this section, I consider three questions that arise in connection with Arneson’s argument about the ‘distributive dimension’ of paternalism. First, we need an argument as to why the inequalities prevented by a particular paternalistic intervention would have been unfair inequalities; the fact that the inequality would have been the result of the agent’s choice could lead us to think that the inequality would have been fair. Second, the exact implications of the argument will depend on the metric we use to determine whether and to what extent someone is actually made ‘better off’ by a paternalistic intervention. Finally, while Arneson’s argument focuses on paternalistic interventions in which particular options are removed entirely, many paternalistic interventions rely on different mechanisms to shape individuals’ choices. These kinds of interventions can raise distributive concerns not addressed in Arneson’s argument.

3.1 Choice

Central to Arneson’s argument is that, under certain conditions, paternalistic interventions are beneficial from the perspective of distributive fairness because they can prevent unfair inequalities. Importantly, Arneson’s argument applies only if the inequalities that the paternalistic intervention prevents would in fact have been unfair: distributive fairness is not improved when we prevent inequalities that would have been fair. However, paternalistic interventions work by interfering with individuals’ choices, and choices are often considered to be a source of fair rather than unfair inequalities. How does this affect Arneson’s argument?

Arneson addresses this issue when he argues that his argument supports (at least some) hard paternalistic interventions. The distinction between hard and soft paternalism is frequently referred to in the paternalism debate. This distinction captures the difference between interventions that restrict voluntary choices (hard paternalism) and those that restrict non-voluntary choices (soft paternalism) (see Feinberg 1986). For example, if I hide my friend’s cigarettes after she has repeatedly expressed frustration about her nicotine addiction and inability to quit, this may be considered an instance of soft paternalism: to the extent that my friend’s decision to smoke is driven by addiction and not endorsed by her, the intervention is interfering with a non-voluntary choice. If, on the other hand, her choice to smoke is fully voluntary, my hiding her cigarettes should be described as an instance of hard paternalism.

In the literature on distributive equality, the criterion of voluntariness is also invoked, including perhaps most prominently in Arneson’s contributions to that debate. Arneson is one of the original proponents of (a particular version of) luck egalitarianism (Arneson 1989a, 1991), which he has since abandoned in favour of what he calls responsibility-catering prioritiarianism (Arneson 2000). However, both his egalitarian and prioritarian positions are meant to be responsibility-sensitive, which Anderson understands as the requirement that ‘it is morally wrong if some people are worse off than others through no fault or voluntary choice of their own’ (Arneson 1990: 177). Accordingly,

distributive justice does not recommend any intervention by society to correct inequalities that arise through the voluntary choice or fault of those who end up with less, so long as it is proper to hold the individuals responsible for the voluntary choice or faulty behavior that gives rise to the inequalities. (Arneson 1990: 176)

At first sight, Arneson’s endorsement of voluntary choice as a source of fair inequality sits uneasily with his suggestion that there is an egalitarian argument for hard paternalism: why would paternalistic interventions that prevent voluntary choices be advantageous from the perspective of distributive equality? In the remainder of this section, I discuss two possible strategies for making the case that the inequalities that are prevented by the interventions Arneson has in mind would have been unfair, even though they are the result of individuals’ voluntary choices: one provided by Jansen and Wall, the other by Arneson. Arneson, I suggest, provides a more appropriate response to this problem than Jansen and Wall.

Jansen and Wall argue that, even if individuals are responsible for the choices they make, it is society as a whole that is responsible for creating or permitting the opportunities in which particular choices become possible. Take gambling as an example:

a political society might decide to permit gambling houses in its territory in full knowledge that gambling will spell the financial ruin of many its members. In doing so, it would be vulnerable to the fairness objection we have been discussing. This remains true, even if it is also true that those who recklessly gamble are responsible, or at least partly responsible, for their fate. (Jansen and Wall 2009: 178)

On their account, then, ‘[i]t can be unfair to implement a regulatory scheme that is costly for bad decision-makers over one that is less costly to them, even if the bad decision-makers are responsible, or partly responsible, for their bad decisions’ (Jansen and Wall 2009: 178).

However, this argument does not, in fact, address the problem at hand. If – as Jansen and Wall assume – the choice to gamble is a choice that individuals are responsible for in the sense required by distributive fairness, then any inequalities resulting from those choices are fair. If we make this assumption, then opportunities to gamble are simply opportunities for fair inequalities. However, the move from an equal, fair distribution to one that contains inequalities where all of these inequalities are fair, is a move that – from the perspective of distributive equality – we must be indifferent towards. More generally, as far as equality is concerned, responsibility-catering versions of distributive equality cannot distinguish between, on the one hand, a distribution that is equal and fair and, on the other hand, a distribution that contains some inequalities, as long as these inequalities are fair.

Arneson addresses this problem in a different way. He emphasises that even if we accept that it is fair for individuals to be better or worse off than others to the extent that they are responsible for such (dis)advantages, many important differences in people’s decision-making capacities are due to luck and individuals therefore should not be held responsible for them:

Whatever conception of fault one adopts, inequalities of welfare that arise through the individual’s own fault as judged by that conception will neither violate the principle of equality nor count as unfair. But on anybody’s conception of fault the prudential disabilities that separate more and less able agents are surely in very considerable part due to accidents of genetic endowment and variously favourable early childhood circumstances that do not lie within the agent’s control and for which he cannot be either praiseworthy or blameworthy. So even if we accept that it is sensible to attribute some prudential failings of individuals to personal fault, these attributions cannot reconcile us to regarding as fair the great bulk of inequalities of welfare that separate more and less able agents. Paternalism remains in the running as one morally appropriate response to some of these pervasive and disquieting inequalities. (Arneson 1989b: 422–3)

Importantly, even if choices are ‘voluntary enough’ so that interference with them would be a case of hard rather than soft paternalism, that does not automatically make them the kinds of choices for which it would be fair to hold them responsible, for the purposes of distributive equality. Even though responsibility-catering egalitarians (including Arneson) often talk about ‘voluntary choices’ as the kind of choices that lead to fair inequalities, the conception of voluntariness they rely on is often very different from the notion of voluntariness that is used to draw the line between soft and hard paternalism. In fact, on more stringent interpretations of responsibility-sensitive egalitarianism, there will be few, if any, choices in the real world that would meet the requirements responsibility-sensitive egalitarians stipulate. Even ‘voluntary’ choices can be shaped by brute luck in such a way that any inequalities resulting from those choices would be considered unfair (Voigt 2007).

What does this imply for the egalitarian argument for paternalism that Arneson advances? Arneson’s argument only applies when we are talking about paternalistic interventions that prevent unfair inequalities. Paternalistic interventions that restrict choices that would have led to fair inequalities do not improve distributive equality. The scope of Arneson’s argument therefore depends on our ability to make the case that the choices restricted by particular paternalistic interventions are not the kinds of choices that would lead to fair inequalities – choices, that is, for which it would be fair to hold them responsible.

3.2 Outcomes

Another question we have to ask concerns the ‘metric’ we use to determine whether an individual is indeed made ‘better off’ by a particular intervention. Dworkin defines this aspect of paternalistic interventions broadly, suggesting that these interventions are to ‘improve the welfare of Y … or in some way promote the interests, values, or good of Y’ (Dworkin 2011). This idea can, of course, be fleshed out in very different ways. We may, for example, rely on the agent’s own judgements and preferences to determine whether or not an intervention improves his or her well-being: paternalistic interventions that are based on the preferences of the agents interfered with are arguably much less problematic than interventions that do not defer to agents’ preferences in this way (Goodin 1991).

Specifying the outcomes we are concerned with when we are intervening paternalistically can also help us respond to a line of criticism that has been put forward in this debate. In response to Jansen and Wall’s argument about participation in research trials and the idea that prohibitions on particularly unfavourable risk-benefit ratios will protect ‘poor’ decision-makers, Edwards and Wilson have emphasised that, contrary to what Jansen and Wall assume, people may participate in risky research trials not because they are poor decision-makers and unaware of the risks or incapable of accurately evaluating them but because they have altruistic motivations and are happy to accept risks that will benefit others (Edwards and Wilson 2012).

Jansen and Wall anticipate this argument. They emphasise that if we are thinking about the issue from the perspective of regulation, practical limitations will prevent us from designing policies that distinguish between, and treat differently, altruists and poor decision-makers: whatever regulation we come up with will affect both of these groups. More importantly, they argue, some risks would simply be wrong for individuals to assume, even for altruistic reasons: ‘Each person has a duty to respect herself’ and we fail to meet our ‘self-regarding duties’ if we accept such risks (Jansen and Wall 2009: 179). We can then think of legal requirements that clinical trials not subject research participants to excessive risks as giving effect to this concern.

Jansen and Wall do not provide a full defence of this (clearly controversial) argument. What matters for the present argument is that our understanding of, and response to, the altruism challenge also depends on how we define what the relevant outcomes are. We could, for example, define individual well-being in such a way that acting on an altruistic motivation makes the agent better off in some respects, for example because the agent gets pleasure from thinking about the beneficial effects of her actions on others, or because there is objective good in acting from altruistic motivation. If there are such positive effects on well-being, then these may outweigh, at least in some cases, the costs or risks that altruistic agents accept. We can also allow for the possibility that the influence of altruism on people’s choices can indeed be the kind of disadvantage that they should be compensated for. From the perspective of distributive justice, we could argue that altruistic choices can be problematic, even if they are ‘voluntary’ and reflect individuals’ preferences: if, for reasons of brute luck, some people are more altruistic than others and therefore more likely to forego benefits for themselves so as to aid others, it is certainly not obvious that the sacrifices they make as a result should not be considered unfair disadvantages. What this highlights is that it is important to specify what outcomes a paternalistic intervention is meant to improve; our answer to that question will also influence how we think about costs that people bear as a result of altruistically motivated choices.

This hints at a broader problem for the translation of Arneson’s argument into specific policy solutions: we may not be able to identify any options that are unequivocally bad for everyone who is affected by a particular paternalistic intervention, irrespective of their specific interests and preferences. Even in the context of health, which is often considered the kind of good that everyone needs, it is not the case that we can identify specific risks that would be ‘bad’ options for everyone. In many cases, people make trade-offs between health and other goods that may be important to them. For example, some women choose to continue with high-risk pregnancies because they place a very high value on having a child that is biologically related to them. Importantly, however, while this may limit the extent to which we can translate Arneson’s argument into policy prescriptions, it is not a problem with the theoretical underpinnings of his argument.

3.3 Interventions

Arneson implicitly assumes that we are working with paternalistic interventions that fully remove welfare-reducing options from people’s option sets. Is Arneson’s egalitarian argument relevant when we are looking at paternalistic interventions that rely on different mechanisms to shape individuals’ choices? In this section, I discuss two types of paternalistic interventions that rely on different mechanisms: those where information is withheld from individuals and those that change the relative cost associated with particular options rather than removing them entirely.

While most discussions of paternalism focus on interventions where particular options are blocked entirely, this is not the only – and perhaps not even the most common – way to interfere paternalistically with someone’s decisions. One type of intervention that could be guided by paternalistic motivations is that of withholding information. With this type of intervention, the concern is that the provision of information has a negative impact on individuals, for example by causing distress or leading individuals to make decisions that are likely to have negative consequences for them.

An interesting example of this kind of intervention is physicians’ decision not to disclose information to patients about unsubsidised medication. In some countries, medications that are not funded or subsidised through the health care system may nonetheless be available for patients to purchase at their own expense, often at very high prices. Sometimes, such drugs are not subsidised because they have only just become available and have not yet been approved for subsidised provision through the health care system. In other cases, drugs may not be subsidised because they are not considered good ‘value for money’: the drug may be perfectly safe but the expected patient benefit does not seem significant enough to warrant public provision at the price attached to it. When such drugs are available and suitable for particular patients, doctors will have to make decisions about whether or not to inform these patients. Some doctors appear to be reluctant to do so (Jefford et al. 2005).

Of course, a doctor’s decision not to let a patient know about such a drug is not necessarily paternalistic. For example, doctors may be withholding this information because they are really committed egalitarians: they oppose the idea that wealthy patients can gain access to medications that poorer patients cannot afford. Or, more realistically, they may simply not have the time to inform patients about the existence of such drugs. In such cases, the decision not to inform patients about an unsubsidised drug would not count as paternalistic. For the purposes of the argument, I am interested in scenarios where the decision to withhold information about unsubsidised treatments is paternalistic: doctors choose to withhold such information because they believe that this information would be detrimental to the patients’ well-being, for example because it would cause them distress.Footnote 1

One interesting complication in this particular example that does not arise for the scenarios Arneson discusses is that here people’s decisions – and the benefit they derive from the paternalistic intervention – depends at least in part on their financial situation. For wealthy patients, the cost of the drug may not make much difference, whereas poorer patients may find that they have to make significant sacrifices in order to purchase the drug. It may be the case that for these poorer patients, the case for paternalistic interference is greater simply because these patients have more to lose. If we are concerned with the distributive dimension of paternalistic interventions, this is certainly a relevant consideration – and perhaps more significant in its effects than the concerns about differences in decision-making capacities that are central to Arneson’s argument.

Similar concerns arise in connection with paternalistic interventions that, instead of removing ‘bad’ options entirely, make such options more expensive relative to their alternatives. We can, for example, attach a financial penalty to welfare-reducing options, or make welfare-enhancing options more attractive by using incentives. Depending on how strictly we define what it means to ‘remove’ an option, many paternalistic policies will in fact be based on this approach. Few interventions are able to block people’s access to particular options entirely; even legislation that enforces the use of seatbelts and safety helmets is effectively an intervention that changes the relative cost of the options involved: the option of not wearing a seatbelt or a helmet is not removed entirely but fines and penalties make it significantly more expensive.

In the health context, taxation is often used to make unhealthy products more expensive. Tobacco is perhaps the most prominent example here but taxes have also been introduced or considered to lower the consumption of alcohol as well as fatty and sugary foods. Positive incentives are also increasingly used. In the US, for example, many employers will lower health insurance contributions for employees who are non-smokers and whose weight is within the ‘normal’ range (Schmidt et al. 2010). As a matter of fact, of course, at least some of these policies are more likely to be motivated by considerations of cost rather than concerns about individuals’ well-being. However, these interventions could be paternalistically motivated: they could have been put in place so as to steer people towards choices that are ‘better’ for them.

A related type of intervention, which is also focused on the relative cost of particular options, is the regulation of financial inducements. This is a concern that is often raised in health contexts. For example, there are restrictions on how much people can be paid to participate in research trials, to act as pregnancy surrogates or to donate blood or organs.Footnote 2 Again, these policies do not necessarily have a paternalistic motivation but certainly paternalistic arguments could be made to support them.

As in the earlier example of the doctor deciding whether to disclose information about an unsubsidised drug, individuals’ financial situation is likely to influence what decisions they make and whether or not they will benefit from paternalistic interventions. With financial inducements, the concern is that they will have more of an effect on someone who is poor than on someone who is wealthy. In the US, for example, critics have noted that, in the context of phase 1 trials, the payments for participation have become ‘high enough to make participating in trials more lucrative than holding a minimum-wage job’ and as a result many poor people are relying on trial participation as a source of income (Elliott and Abadie 2008: 2317). From the perspective of equality, the worry is that because people are making choices against such unequal background conditions, people on low incomes would be more willing to take risks than people on higher incomes: ‘a sum of money that the wealthy can easily resist may be very tempting for poorer people’ (Elliott and Abadie 2008: 2316).

With incentives for healthy behaviour (and taxation on unhealthy products), one claim that is often made is that such interventions will have a greater effect on low-income groups, therefore helping to improve health outcomes among individuals who tend to be of poorer health. This mechanism would make incentives beneficial from the perspective of health equality. At the same time, we cannot be certain that these are the effects that incentives are in fact going to have: wealthier people are often in a better position to take advantage of available opportunities to adopt healthier behaviours, whereas for poor people, the existence of other constraints means that even with additional incentives in place, they may not be able to adopt healthier behaviours (Voigt 2012). If it is this second mechanism that becomes dominant, then incentives effectively create opportunities for inequalities that did not exist previously: for example, poor smokers may face the additional financial burden of higher taxes without reaping any of the health benefits that policy-makers sought to achieve. To the extent that such effects are likely to occur, it may actually be better – from the perspective of distributive equality – to completely remove options rather than change the cost attached to them. For example, from the perspective of equality, it may be better to completely ban tobacco than to increase tobacco taxes (see also Voigt 2010). Thus, even though approaches that merely change the relative cost of options are often considered more appealing because they are less restrictive, they may have distinct disadvantages as far as distributive equality is concerned (Voigt 2012).

4 Paternalism: Distributive Versus Relational Equality

The previous section highlighted some of the complexities surrounding Arneson’s argument and the importance of specifying how exactly specific components of the argument should be defined or interpreted. Arneson’s argument focuses on the distributive implications of paternalistic interventions: how do paternalistic policies increase – or decrease – equality in distributive outcomes? However, critics have pointed out that distributive concerns should not be regarded as central to egalitarian justice; instead, equality should be conceived of as concerned primarily with the nature and quality of relationships between individuals (Anderson 1999; Scheffler 2003).Footnote 3 Distributive considerations may still be relevant – because distributive inequality is likely to undermine relational equality – but distributive equality is only instrumentally, not intrinsically, important.

Similar ‘relational’ concerns are raised about paternalism. That paternalism may communicate disrespect towards the individuals interfered with has been a concern in the debate, even if the link to the relational equality literature is not made. For example, Seana Shiffrin underlines the way in which paternalistic interventions implicitly stipulate an asymmetry of knowledge and competency between the two agents involved. This makes the expression of disrespect a central feature of paternalistic interventions:

The essential motive behind a paternalistic act evinces a failure to respect either the capacity of the agent to judge, the capacity of the agent to act, or the propriety of the agent’s exerting control over a sphere that is legitimately her domain…. Paternalistic behaviour is special because it represents a positive… effort by another to insert her will and have it exert control merely because of its (perhaps only alleged) superiority. As such, it directly expresses insufficient respect for the underlying valuable capacities, powers, and entitlements of the autonomous agent. Those who value equality and autonomy have special reason to resist paternalism toward competent adults. (Shiffrin 2000: 220)

What does this mean for Arneson’s argument? If Arneson is right, then paternalistic interventions can have important benefits for distributive aspects of equality. At the same time, however, paternalistic interventions can be seen as problematic from the perspective of relational equality. Moreover, it is arguably one of its central assumptions – that some people are simply better or more competent decision-makers than others – that makes Arneson’s argument particularly susceptible to this line of criticism. Arneson’s argument is not unique in focusing on problems with individual decision-making. Richard Thaler and Cass Sunstein’s argument for libertarian paternalism, for example, is similarly based on concerns about choice heuristics and cognitive biases that affect individuals’ choices (Thaler and Sunstein 2008). However, while such biases will likely affect all of us to some extent in different situations, it is a central aspect of Arneson’s argument that some people are systematically worse than others when it comes to making choices. When an argument relies on drawing a clear line between ‘good’ and ‘bad’ choosers, concerns about disrespect cannot easily be dismissed.

Can we defend Arneson’s argument against this challenge? One possible response is that concerns about disrespect can – at least to some extent – be accommodated within the argument. As I suggested in Sect. 6.3 above, we need to specify in terms of what outcomes we decide that individuals’ welfare or interests are indeed served by a particular paternalistic intervention. Some metrics, such as welfare, may well capture some of the negative effects of individuals feeling that certain restrictions express disrespect towards them. Depending on how significant these negative effects are, they may outweigh whatever benefits we expect from the paternalistic interference. However, this response does not address the core concern about relational equality, which is a concern about how individuals treat and relate to each other independently of any effects such treatment may have on anyone’s well-being.

A second response is proposed by Jansen and Wall, who explicitly address this concern. They argue that paternalistic policies need not imply disrespect to anyone. They suggest that as long as these policies are sufficiently broad, they can avoid the kinds of effects anti-paternalists like Shiffrin are worried about. With respect to their own argument about participation in research trials, they emphasise that ‘fairness-based opposition to anti-paternalism has an impersonal dimension’ (Jansen and Wall 2009: 181). They conclude,

the paternalistic… restrictions that the fairness argument would justify must be formulated in general terms that apply broadly to the entire set of potential research subjects. They do not target specific individuals but rather groups of people. No person should conclude therefore that the paternalistic restrictions express the message that he or she lacks good judgement or good decision-making abilities. At most, the restrictions express the message that some (unspecified) members of the population of potential research subjects lack good judgement and decision-making abilities. And this message need not be insulting to any person in particular – indeed, it may be a message that nearly all would assent to. (Jansen and Wall 2009: 181)

This response is not entirely satisfactory. While it is possible to design policies that do not make reference to particular individuals and their decision-making capacities, whenever individuals find that particular choices are blocked for paternalism-for-equality type reasons, this will indicate to them that the options they would have chosen were considered to reflect poor judgement. Some people will find that their liberties are restricted by a paternalistic policy whereas for others, the restrictions do not interfere with the decisions they would like to make. If these restrictions are supported by an argument such as Arneson’s, the restrictions may well communicate disrespect to the former group; this is not the case for those whose choices remain unaffected.

Given the assumption of unequal decision-making capacities underlying his argument, it seems that Arneson’s position is susceptible to concerns about relational inequality. Of course, this does not invalidate Arneson’s argument; rather, it draws attention to a facet of equality that remains unexplored in his approach. Paternalistic interventions require careful weighing of the different considerations at stake. Distributive and relational concerns seem to pull in different directions in this case and it is far from clear how conflicts between relational and distributive equality should be resolved.

5 Conclusion

By drawing attention to the ‘distributive dimension’ of paternalism, Arneson highlights an important but often underappreciated aspect of paternalism. Distributive equality is an important goal and should be one of the considerations we take into account when considering whether or not particular paternalistic interventions are acceptable. Arneson’s argument does not (and is not meant to) provide an all-things-considered, knock-down argument to defeat anti-paternalism. Rather, it adds an important nuance to the debate; it brings another consideration to the table that we have to take into account as we evaluate the costs and benefits of particular paternalistic policies or interventions.

This paper considered Arneson’s argument in more detail so as to get a better sense for its scope, possible implications and the complexities it raises. Perhaps most importantly, Arneson’s argument only works to the extent that the choices curtailed by particular paternalistic interventions would have led to unfair inequalities. Since choice is involved, we may think that the resulting inequalities would have been unproblematic. The scope of the argument depends on our ability to make the case that differences in choice-making capacities are matters of brute luck whose influence on distributions should be considered unfair. It is also far from straightforward to determine when individuals are indeed made ‘better off’ by a particular intervention and to translate Arneson’s argument into policy proposals that would reliably make better off (in the required sense) those affected by the policies. Further, while Arneson focuses on paternalistic interventions that remove particular options, other – probably more common – types of paternalistic interventions (such as withholding information or changing the relative ‘cost’ of particular options) raise further issues of distributive equality not addressed by Arneson’s argument. Finally, whatever contribution paternalistic interventions can make to distributive equality, there arguably is more to equality than outcomes. Paternalistic interventions, especially when they rely on distinction between ‘good’ and ‘bad’ choosers, may express disrespect towards those whose interests they are meant to protect. Such relational inequalities must be weighed against whatever improvements in distributive equality we expect paternalistic interventions to achieve.