I do not disagree with Chater & Loewenstein's (C&L's) central claim, that nudge-style interventions have been oversold as a way of achieving behavioral change, resulting in excessive focus on the i-frame in recent policy debates. I would like to insist, however, that the libertarian paternalism project as originally conceived by Sunstein and Thaler (Reference Sunstein and Thaler2003) is not vulnerable to this objection. The problem arose because of a gradual expansion of their claims over time, along with a loosening of constraints on the circumstances of application. In the book Nudge (2008), Thaler and Sunstein offered inconsistent definitions of what constitutes a nudge, illustrated through policy interventions that extended far beyond the domain of libertarian paternalism. By offering a clear account of the relevant distinctions, I hope to show that there are circumstances in which the privileging of the i-frame over the s-frame in behavioral public policy remains appropriate. My central criticism of C&L is that they are insufficiently attentive to this point about paternalism, and so gloss over important differences between policy problems like obesity, where a concern for individual autonomy limits the range of s-frame interventions that can be undertaken, and climate change, where the full menu of s-frame options is available.

It remains the default assumption in most domains of public policy that the state should promote its objectives by adjusting the incentives faced by individuals, making behavior that it seeks to promote less costly and making behavior that it wants to discourage more costly. This standard approach is considered problematic, however, when the objectives being sought involve promoting the putative good of the same individual whose behavior is being incentivized. In this case, as John Stuart Mill (Reference Mill1867) observed, the intervention is either superfluous or will involve the state second-guessing the individual's own evaluation of the objective. Thus, Mill's argument categorizes interventions that involve interference with individual choice through reference to the individual's “own good” as objectionably paternalistic. This analysis has been sufficiently influential in liberal states that in many policy domains – such as nutrition and dietary choice – public officials are reluctant to engage in coercive intervention (understood broadly, to include both taxation and subsidization of options).

As the material circumstances of human life have improved and the number of unresolved collective action problems has dwindled, the proportion of human misery that is self-inflicted has steadily increased. This accounts for much of the enthusiasm that greeted the original proposal for libertarian paternalism. Sunstein and Thaler argued that, even in cases in which it would be impermissible for the state to change the incentives faced by the individual, it may be permissible to present the choice in a way that generates a psychological effect that favors some preferred option. In particular, if the effect is achieved through exploitation of a cognitive bias, one cannot really object to it as an infringement of individual autonomy, because the intervention only works on individuals who are failing to exhibit rational autonomy in their actions (Hansen, Reference Hansen2016). So long as the intervention does not change the rational structure of the decision problem, it cannot be construed as interfering with individual autonomy. It follows that the state need not stand by idly while individuals engage in self-destructive behavior, it can use insights from behavioral psychology to devise interventions that will be psychologically powerful and yet non-coercive (Sunstein, Reference Sunstein2014). Or in the terms favored by C&L, even when the state cannot change the s-frame, it can still manipulate aspects of the i-frame.

In its initial formulation, libertarian paternalism was specifically aimed at the problem of developing policy in domains where standard forms of state interference were considered impermissible. Over time, however, Thaler and Sunstein, along with many enthusiasts, began introducing modifications that blurred these boundaries. The idea of a nudge was introduced, along with the suggestion that it might be permissible for the state to modify the incentives faced by individuals just a little, or that coercive imposition of costs might be permissible so long as they were not that large, or non-pecuniary, or “easy and cheap to avoid” (Thaler & Sunstein, Reference Thaler and Sunstein2008, p. 6). The Cameron government in the United Kingdom took things a step further, expanding the use of nudge strategies to cater to what amounted to merely a political preference against state coercion. This lent support to the notion that psychological manipulation might be favored simply on the grounds that it was cheaper than the provision of incentives. Certain widely publicized efforts in the United Kingdom, such as the attempt to use the “watching eyes” effect to deter tax evasion, reflected this preoccupation. Paternalism was clearly not the issue here, because the state has no qualms about throwing people in prison to deter tax evasion as well.

Although C&L mention the paternalism issue briefly, they ignore the issue throughout the bulk of their discussion. Thus one finds climate change being discussed side-by-side with public health problems related to obesity, despite important structural dissimilarities between these policy domains. Greenhouse-gas emissions are a classic negative externality, and so only the most extreme libertarian doubts that states are entitled to act coercively to deter their production (e.g., through carbon taxes). Certain states have of course failed to implement appropriate charges, but this is because of political challenges largely unrelated to any neglect of the s-frame. With obesity, on the contrary, even the mildest liberal has reason to be cautious about giving state officials license to interfere with dietary choices. It is not merely corporate lobbying that has prevented action in this domain, there is also an important matter of political principle at stake. Yet in their discussion of obesity, C&L ignore the possibility that an aversion to paternalistic intervention might lead policymakers to favor i-frame interventions. This is unfortunate, because it was the promise of a more sophisticated approach to these questions that was responsible for much of the excitement generated by Sunstein and Thaler's initial proposal.