In their article, Chater & Loewenstein (C&L) support the preferential use of system-level interventions, largely because of their greater effectiveness. It is, however, unclear whether or not the apparent increased “effectiveness” of s-interventions is sufficient to justify their use in preference to individual level (i-) interventions. The authors note the risk of “heavy-handed ‘paternalism’” in adjudicating between “freedom-to-choose and freedom-from-temptation,” and suggest it should be tempered by “democratic processes.” I argue, here, that difficulties in striking a balance between different values are a core reason why s-interventions are controversial, and might limit the reach of C&L's conclusions. Although I agree with much of their characterisation of the unfortunate neglect – by both behavioural scientists and policy makers – of s- in favour of i-interventions, there remain important ethical reasons why it is not so straightforward to adopt s-interventions.

One reason for preferring choice-preserving interventions is they allow governments to avoid taking strong positions on the rankings of controversial values (such as health, environment, leisure time, religion, close relationships). Although it is impossible to govern without making some assumptions about what people (should) value, avoiding ranking such values or predetermining how people must trade them off with one another is often desirable. Yet a number of the interventions C&L appear to implicitly endorse take strong positions on the relative importance of such values, for instance, by banning plastic bag use, or addressing the “urgent” problem of misinformation through “dramatic tightening of regulation of social media.”

To better illustrate the problem, consider a helpful analogy C&L provide to assist the reader in seeing why it is foolish to get stuck in i-intervention thinking. C&L describe a slum landlord who claims his tenants’ poor health is a result of a lack of hand hygiene, rather than a lack of sanitary housing. Although we might readily accept that such a landlord should be required (morally and perhaps legally) to improve his housing stock, other examples might raise questions. For instance, skiing injuries could be reduced by encouraging people to wear helmets (i-intervention) or by requiring ski resorts to make their runs flatter (s-intervention). C&L's arguments suggest that, because the latter would surely be much more effective, we should endorse this s-intervention over the alternative i-intervention. Yet it is not obvious that this really is a preferable approach to the problem.

There are a number of issues at stake here. First, it is not as straightforward as is perhaps implied by C&L to judge what should count as the most “effective” intervention. Public health promoters will presumably judge whatever intervention results in the biggest increases in public health (however they choose to define this) as the most effective. But other groups and individuals might reject this – health is not the only thing that matters. Economists might focus instead on productivity; artists might think that cultural richness is more important. Returning to the ski slope example, if we make the ski slopes flatter, people go slower and get less pleasure. There are also fewer accidents. It is not self-evident whether or not this is a net improvement.

A second issue is who's job it is to shoulder the burden of making changes that will result in the desired improvements, and to what extent coercive force (or more moderate punishments or rewards) may be used in order to ensure adherence. C&L point to the enthusiasm for individual responsibility shown by corporate opponents of s-interventions. Yet freedom and responsibility are not purely the invention of commercial actors seeking to promote their own interests. The authors point to the (sometimes extensive) influence of the social and built environment on people's behaviour. Indeed, this might give us pause when considering the extent to which individuals are responsible for that behaviour. For instance, if the main determinant of whether or not one eats junk food is whether or not there is a fast-food outlet nearby, we might question whether people's dietary behaviour is a result of freely made choices, consistent with responsibility, or is instead “controlled” by the actions of others. But this “pause” is not equivalent to a robust conclusion that freedom and responsibility are absent, or unimportant. It is a far from settled topic within philosophy and interdisciplinary work in ethics (Brown & Savulescu, Reference Brown and Savulescu2019; Cavallero, Reference Cavallero2019; Davies, De Marco, Levy, & Savulescu, Reference Davies, De Marco, Levy and Savulescuforthcoming; Segall, Reference Segall2009).

The values at stake in the obesity case and other examples provided by C&L are not self-evident. Although it may be reasonable for states to take health as an uncontroversial “good,” this does not mean it may be pursued at all costs. The appeal of i-interventions is to avoid making too many controversial value weightings, and instead to facilitate individuals to weigh up their own values and act accordingly. In order to show that i-interventions are no good, it is not enough to simply show that they don’t reduce obesity or alcohol consumption to exactly the amount deemed by health promoters, behavioural scientists, or governments to be the “correct” level. It needs to be shown that i-interventions fail to facilitate decision making (or behaviours) by individuals that reflect their values and promote their interests. This might well be the case, particularly when commercial interests are unaligned with individual interests. It is not, however, enough to show that the greater “effectiveness” of s-interventions straightforwardly justifies their use. Nor does the relative enthusiasm of commercial interest groups for i-interventions show that individual choice has a dramatically reduced role to play in behavioural research or public policy.