Nudging, shaming and stigmatising to improve population health

Nudges are small, often imperceptible changes to how particular decisions present themselves to individuals that are meant to influence those decisions. In his editorial, ‘Nudging by shaming, shaming by nudging’, Eyal highlights links between nudges and feelings of shame on the part of the ‘chooser’. In this commentary, I suggest two further distinctions between different types of shame-based nudges that should affect our assessment of such nudges.

T he idea of 'nudges' -small, often imperceptible changes to choice 'architectures' that are meant to exert a positive influence on people's choices -has gained a lot of traction in recent years. Its most prominent proponents, Thaler and Sunstein, argue that nudges are attractive because they can help individuals make decisions that they themselves would regard as better, without curtailing their liberties (1). Nir Eyal's paper draws a number of important links between nudges and feelings of shame on the part of the 'chooser' . Eyal suggests that, first, shame could play a significant role in certain types of nudges. For example, smoking bans may influence smoking behaviour by signalling that smoking is regarded as socially undesirable, thereby causing smokers to feel shame. Second, Eyal highlights that the role of shaming in certain types of nudges can make their use more problematic than other, more restrictive types of interventions (2). Eyal's discussion highlights an important problem with current debates about public health interventions. In many discussions, the central criterion for assessing such interventions is the degree to which they restrict individuals' liberties: more restrictive strategies, which limit or curtail what individuals can do, are generally regarded as more problematic than strategies that involve lesser interference with individuals' freedom of choice. The Nuffield Council's influential 'intervention ladder' , which suggests that we think of the range of policy options open to policy-makers as positioned along a ladder with increasing restrictions on citizens' liberties, is a case in point (3). Eyal argues that the focus on liberty restrictions misses important dimensions of how public health interventions can influence individuals' behaviour; shaming and its possible negative effects on choosers are one such dimension. Less restrictive interventions that induce feelings of shame to shape individual choices can, he argues, be more problematic than more restrictive ones that do not rely on shame-based mechanisms. As Eyal emphasises, the current focus on the relative restrictiveness of different health promotion strategies may cause us to lose sight of this aspect of health interventions. It is particularly important to explore the link between nudges and shaming because proponents of nudges seem to approve of the shaming effects some nudges rely on. Consider, for example, Thaler and Sunstein's discussion of how nudges can encourage people to use less energy. They consider a US study in which electricity bills informed customers how their energy use compared with the average energy use in their neighbourhood. This led above-average energy users to decrease their energy use but also led to a 'boomerang effect': below-average users ended up increasing their energy use. However, when the invoices also included an emoticon signalling approval (for those who were using less energy than their neighbours) or disapproval (for above-average users), the boomerang effect disappeared. Thaler and Sunstein speak approvingly of this kind of nudge, which goes beyond simply telling people how their behaviour compares to that of others [an approach sometimes referred to as 'social norms marketing' (4)] and instead relies on clear signals of social (dis) approval so as to ensure that behaviour changes in the desired direction. In this commentary, I want to highlight two distinctions not raised by Eyal that I think should be relevant to ongoing debates about public health interventions, nudges and the use of shame. First, some nudges are paternalistic-i.e. they interfere with choices so as to improve the health or welfare of the chooser-while others are meant to benefit third parties. Second, some nudges rely on shame arising in specific situation whereas others (also) create or contribute to broader stigmatisation in an effort to shape behaviours. While shaming strikes me as a prima facie problematic strategy for changing people's behaviours, it is not, I think, plausible to argue that shaming is never permissible, irrespective of whatever possible positive effects it may have. Even Martha Nussbaum, who takes a very strong position against the use of shaming because she regards it as inconsistent with basic commitments to equality, allows for the possibility that, under very specific circumstances, shaming could be 'constructive' (5). Shame can be used in a variety of different ways and different uses of shame may create different types of concerns. Our debates about the use of shame should therefore consider what, precisely, makes the use of shame problematic in particular instances and also aim to distinguish between more and less problematic instances of shaming. The two distinctions suggested here should make a difference to how we assess the acceptability of different kinds of nudging-by-shaming.

Paternalistic vs. non-paternalistic shaming
A first distinction that must not be conflated as we consider different strategies that are being employed to influence individuals' behaviours is whether the goal is to improve the health or well-being of the chooser or instead that of a third party. Sunstein and Thaler originally defined their approach as 'libertarian paternalism' , suggesting that nudges were primarily meant as a way to improve the well-being of the chooser: 'we argue for self-conscious efforts… to steer people's choices in directions that will improve their lives. In our understanding, a policy is "paternalistic" if it tries to influence choices in a way that will make choosers better off…' (1). At the same time, many of the nudges they discuss are meant to benefit third parties rather than the chooser and therefore are not paternalistic in this sense. For example, they make suggestions on how to increase organ donation, where the goal is to increase the supply of available organs, not to benefit potential donors (1). Similarly, some of the interventions Eyal discusses (such as Michael Bloomberg's failed super-size soft drink ban) seem to be primarily paternalistically motivated whereas others (such as nudges to increase patients' willingness to be treated by trainee doctors) are not. Whether a particular intervention is motivated by the wellbeing of the chooser or instead that of third parties should affect our assessment of its acceptability. When third parties are actively harmed by someone's choices, government intervention often strikes us not only as permissible but in fact required. When someone's choices harm only the chooser, however, we often think that the bar for interventions should be much higher. If we accept this reasoning, then this also has implications for nudging-by-shaming. When nudges are used to prevent choices that involve harms to third parties, they seem less problematic than when they are used to interfere paternalistically -even when they rely on shaming to influence the chooser. Of course, many choices affect both the chooser and third parties, making the distinction between paternalistic and non-paternalistic nudging tricky to maintain. Moreover, even with other-regarding actions, the seriousness of the harm the chooser might be imposing on others, or the benefit they are denying them, can vary greatly. It is not obvious that nudgingby-shaming would be acceptable with respect to any of these choices and I am not taking a position on this here; rather, I want to emphasise that shaming seems less problematic (though still problematic) in instances where people are not doing what they ought to do or are failing in certain moral duties that they owe to others than it does in contexts where choosers could be acting in ways that are better for themselves.

'Situational' shaming vs. shaming through broader social norms
The second distinction I would like to highlight focuses on the mechanisms through which shame operates to influence individuals' behaviour. In some of the nudges Eyal discusses, shame is induced in very specific situations (often interactions with another person). For example, in the case of the super-size soft drink ban, feelings of shame seem to arise if customers feel embarrassment when returning for drink refills 'under the gaze of other customers and vendors' (2). With other nudges, the shaming seems to work through broader social norms that influence behaviour and are not isolated to specific social interactions. Consider, for example, smoking bans. As Eyal suggests, these influence individuals' behaviour through 'subtle stigma' , (2) signaling to smokers that they are engaging in a behaviour that society disapproves of. This 'subtle stigma' may be situational: smokers may feel shame when they are required to leave buildings in order to smoke. [And indeed smokers report feeling like 'lepers' as a result of such bans (6).] At the same time, however, this kind of legislation is often intended to set in motion much broader denormalisation effects that in turn are expected to affect smoking behaviours (4). As Stanton Glantz puts it, smoking bans are effective because they 'implicitly defin[e] smoking as an antisocial act' (7). While the mechanisms are very different, in both cases social disapproval is signalled to smokers and it is the resulting shame that becomes part of the choice architecture, 'nudging' their behaviour towards a reduction in smoking. Why might this difference matter? With nudges where individuals are exposed to shame only if and when they engage in a particular behaviour-whenever they request a refill on their soft drink, for example-any feelings of shame the chooser experiences are isolated to a particular interaction and can be avoided quite easily. On the other hand, nudges that work through, or rely upon, the broader stigmatisation of particular behaviours are likely to be more pervasive in their effects, often including effects that are problematic for the chooser's health. There have been concerns, for example, that the stigmatisation of smoking (driven by a range of policies, including the kind of smoking bans Eyal discusses) can lead smokers to conceal their smoking status from health professionals and as a result lose out on the cessation advice those professionals might have been able to provide (8). Nudges in which shame is limited to particular situations and does not have 'spillover' effects may therefore be less problematic from a normative perspective -even if, of course, as Eyal highlights, even situational shaming can be very problematic.

Conclusion
Nudges have been met with significant interest from policy-makers as they consider possible strategies to improve population health. The possibility that nudges, like many other kinds of interventions, can induce shame-that, in fact, they rely on shame to effect behavioural changes-should be an important aspect of ongoing debates about strategies to improve population health. As Eyal highlights, the focus on restrictiveness that has dominated the debate on public health interventions is unhelpful; our assessment of public health interventions must be based on a much broader set of normative concerns.

Ethical issues
Not applicable.