First, I set out empirical concerns. Several assertions in the target article are the result of misunderstandings or the selective use of evidence. As someone who has led work in this field for over a decade, I disagree strongly that the goal of behavioral public policy has been “to provide an alternative to traditional s-frame policies” (target article, sect. 1, para. 6). Instead, the goal has been to integrate behavioral science into existing policy approaches, as shown by the development of a broader “behavioral insights” agenda (Strassheim & Beck, Reference Strassheim and Beck2019). Chater & Loewenstein focus too much on the political arguments made for nudges, and neglect the fact that a technocratic rationale for behavioral science was being advanced in parallel (Hallsworth & Kirkman, Reference Hallsworth and Kirkman2020). The technocratic strand took a much more systemic view of policy-making issues, and was embedded in institutional decision-making frameworks (Whitehead et al., Reference Whitehead, Jones, Lilley, Pykett and Howell2017).
The authors prominently feature a quote from David Cameron when campaigning to be UK Prime Minister in 2009. Yet there are several contemporaneous quotes that take a very different view. For example, a UK government document from around 6 months later, which directly informed the creation of the Behavioural Insights Team, explicitly states that the application of behavioral science “powerfully complements and improves conventional policy tools, rather than acting as a replacement for them… sustainable changes in behaviour will come from the successful integration of cultural, regulatory and individual change” (Institute for Government & Cabinet Office, 2010). The report explicitly rejects the idea that “‘behaviour change’ can be understood as simply a novel alternative to, say, legislation” (p. 52).
Moreover, the target article fails to engage in any depth with evidence of what “nudge units” actually do, nor the processes of public administration more generally. Frequently, the target article tries to draw a direct connection between academic papers on the one hand and specific policies on the other, while avoiding any serious engagement with how the policy-making process works – and the role that behavioral scientists actually play in it. Several ethnographic studies document the actual practices of “nudge units” and other practitioners (e.g., Ball & Feitsma, Reference Ball and Feitsma2020; Feitsma, Reference Feitsma2019). Such studies reveal that “behaviour experts are misrepresented by the behavioural policy frontstage with respect to the complexity of their endeavours” and that behavioral science is “also incorporated at earlier stages in the policymaking process,” rather than just being used as tweaks to implementation (Feitsma, Reference Feitsma2019).
What these studies also show is that practitioners attempt to enhance, support and, yes, critique (where appropriate) “traditional” policy-making options – a sharp contrast to the caricature of policy entrepreneurs who always shove nudges to the front of the queue. In the United Kingdom, the contribution that behavioral science made to national obesity policy was to emphasize the automatic and non-conscious dimensions of food consumption (e.g., Ello-Martin, Ledikwe, & Rolls, Reference Ello-Martin, Ledikwe and Rolls2005; Hollands et al., Reference Hollands, Shelmit, Marteau, Jebb, Lewis, Wei and Ogilvie2015; Levitsky & Pacanowski, Reference Levitsky and Pacanowski2012). I personally presented and discussed this evidence with the relevant civil servants over the course of several years. Behavioral science was used to show that individual-level solutions (e.g., relying on exercise alone) were likely to be ineffective because of the power of environmental cues for eating – the precise opposite of what the target article asserts.
I agree that corporate lobbying can be a powerful force, and that corporations are likely to have incentives to emphasize individual behaviors and personal responsibility. But I absolutely do not agree that behavioral scientists have meaningfully contributed to this effort, even unintentionally. The target article offers five main arguments in support: Each of them invites rebuttal on empirical and theoretical grounds, and which I have done at length elsewhere (Hallsworth, Reference Hallsworth2023).
My next group of comments concerns the target article's theoretical basis. The target article introduces a distinction between the “i-frame” and the “s-frame.” This distinction does not offer much clarity and holds up poorly under scrutiny. It does not take much time to think of “traditional” policies that incrementally tweak the rules of the game, rather than rewriting them – like not adjusting tax brackets for inflation. Moreover, several of the examples they give as “i-frame” or “s-frame” could easily be put in the opposite category. The moving to opportunity intervention, presented as an “s-frame” solution, clearly focuses on individual families and the social mobility effects on their children. Maybe you could resolve these taxonomic questions satisfactorily, with some effort. But it raises the question: Is this effort well spent?
Behavioral science should focus more on understanding the interactions between individual and system levels, rather than emphasizing a distinction between them. The target article often discusses the i-frame and s-frame relationship as if they are two separate domains that comment on each other. But it is often the interplay between individuals and systems that determines effects. This question is not purely academic: Some of the most exciting opportunities for behavioral public policy will come from a fusion with the insights offered by complex adaptive systems thinking. Complex adaptive systems show that system-level features of a system can emerge from the interactions of individual actors participating in the system, without direction (Boulton, Allen, & Bowman, Reference Boulton, Allen and Bowman2015). So, rather than a separate “i-frame” and “s-frame,” policy makers are often dealing with “cross-scale behaviors” (Schill et al., Reference Schill, Anderies, Lindahl, Folke, Polasky, Cardenas and Schlüter2019).
My final concerns are pragmatic. The result of all these choices is that the target article paints a picture that is both overly negative and also simplistic. Accordingly, the target article has been presented as a damning criticism of the whole enterprise of behavioral public policy – “What Nudge Theory Got Wrong” (Harford, Reference Harford2022). Already the target article has been used to argue that behavioral public policy should be reduced or even abandoned, rather than extended. This is all many readers will hear of the article, which is a shame: The target article's stated aim is to improve behavioral public policy and it offers solid recommendations, many of which are already in progress. These presentational and editorial choices undermine the target article's stated goals, making it curiously self-defeating – and actually impeding progress for the field.
First, I set out empirical concerns. Several assertions in the target article are the result of misunderstandings or the selective use of evidence. As someone who has led work in this field for over a decade, I disagree strongly that the goal of behavioral public policy has been “to provide an alternative to traditional s-frame policies” (target article, sect. 1, para. 6). Instead, the goal has been to integrate behavioral science into existing policy approaches, as shown by the development of a broader “behavioral insights” agenda (Strassheim & Beck, Reference Strassheim and Beck2019). Chater & Loewenstein focus too much on the political arguments made for nudges, and neglect the fact that a technocratic rationale for behavioral science was being advanced in parallel (Hallsworth & Kirkman, Reference Hallsworth and Kirkman2020). The technocratic strand took a much more systemic view of policy-making issues, and was embedded in institutional decision-making frameworks (Whitehead et al., Reference Whitehead, Jones, Lilley, Pykett and Howell2017).
The authors prominently feature a quote from David Cameron when campaigning to be UK Prime Minister in 2009. Yet there are several contemporaneous quotes that take a very different view. For example, a UK government document from around 6 months later, which directly informed the creation of the Behavioural Insights Team, explicitly states that the application of behavioral science “powerfully complements and improves conventional policy tools, rather than acting as a replacement for them… sustainable changes in behaviour will come from the successful integration of cultural, regulatory and individual change” (Institute for Government & Cabinet Office, 2010). The report explicitly rejects the idea that “‘behaviour change’ can be understood as simply a novel alternative to, say, legislation” (p. 52).
Moreover, the target article fails to engage in any depth with evidence of what “nudge units” actually do, nor the processes of public administration more generally. Frequently, the target article tries to draw a direct connection between academic papers on the one hand and specific policies on the other, while avoiding any serious engagement with how the policy-making process works – and the role that behavioral scientists actually play in it. Several ethnographic studies document the actual practices of “nudge units” and other practitioners (e.g., Ball & Feitsma, Reference Ball and Feitsma2020; Feitsma, Reference Feitsma2019). Such studies reveal that “behaviour experts are misrepresented by the behavioural policy frontstage with respect to the complexity of their endeavours” and that behavioral science is “also incorporated at earlier stages in the policymaking process,” rather than just being used as tweaks to implementation (Feitsma, Reference Feitsma2019).
What these studies also show is that practitioners attempt to enhance, support and, yes, critique (where appropriate) “traditional” policy-making options – a sharp contrast to the caricature of policy entrepreneurs who always shove nudges to the front of the queue. In the United Kingdom, the contribution that behavioral science made to national obesity policy was to emphasize the automatic and non-conscious dimensions of food consumption (e.g., Ello-Martin, Ledikwe, & Rolls, Reference Ello-Martin, Ledikwe and Rolls2005; Hollands et al., Reference Hollands, Shelmit, Marteau, Jebb, Lewis, Wei and Ogilvie2015; Levitsky & Pacanowski, Reference Levitsky and Pacanowski2012). I personally presented and discussed this evidence with the relevant civil servants over the course of several years. Behavioral science was used to show that individual-level solutions (e.g., relying on exercise alone) were likely to be ineffective because of the power of environmental cues for eating – the precise opposite of what the target article asserts.
I agree that corporate lobbying can be a powerful force, and that corporations are likely to have incentives to emphasize individual behaviors and personal responsibility. But I absolutely do not agree that behavioral scientists have meaningfully contributed to this effort, even unintentionally. The target article offers five main arguments in support: Each of them invites rebuttal on empirical and theoretical grounds, and which I have done at length elsewhere (Hallsworth, Reference Hallsworth2023).
My next group of comments concerns the target article's theoretical basis. The target article introduces a distinction between the “i-frame” and the “s-frame.” This distinction does not offer much clarity and holds up poorly under scrutiny. It does not take much time to think of “traditional” policies that incrementally tweak the rules of the game, rather than rewriting them – like not adjusting tax brackets for inflation. Moreover, several of the examples they give as “i-frame” or “s-frame” could easily be put in the opposite category. The moving to opportunity intervention, presented as an “s-frame” solution, clearly focuses on individual families and the social mobility effects on their children. Maybe you could resolve these taxonomic questions satisfactorily, with some effort. But it raises the question: Is this effort well spent?
Behavioral science should focus more on understanding the interactions between individual and system levels, rather than emphasizing a distinction between them. The target article often discusses the i-frame and s-frame relationship as if they are two separate domains that comment on each other. But it is often the interplay between individuals and systems that determines effects. This question is not purely academic: Some of the most exciting opportunities for behavioral public policy will come from a fusion with the insights offered by complex adaptive systems thinking. Complex adaptive systems show that system-level features of a system can emerge from the interactions of individual actors participating in the system, without direction (Boulton, Allen, & Bowman, Reference Boulton, Allen and Bowman2015). So, rather than a separate “i-frame” and “s-frame,” policy makers are often dealing with “cross-scale behaviors” (Schill et al., Reference Schill, Anderies, Lindahl, Folke, Polasky, Cardenas and Schlüter2019).
My final concerns are pragmatic. The result of all these choices is that the target article paints a picture that is both overly negative and also simplistic. Accordingly, the target article has been presented as a damning criticism of the whole enterprise of behavioral public policy – “What Nudge Theory Got Wrong” (Harford, Reference Harford2022). Already the target article has been used to argue that behavioral public policy should be reduced or even abandoned, rather than extended. This is all many readers will hear of the article, which is a shame: The target article's stated aim is to improve behavioral public policy and it offers solid recommendations, many of which are already in progress. These presentational and editorial choices undermine the target article's stated goals, making it curiously self-defeating – and actually impeding progress for the field.
Acknowledgments
I would like to thank George Loewenstein and Nick Chater for their courtesy and engagement, and to Katy Milkman, Michael Sanders, and Cass Sunstein for their comments on the draft of this paper. I also thank the organizers of the 2022 Annual Conference of the Center for Health Incentives and Behavioral Economics at the University of Pennsylvania.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interest
I am employed by The Behavioural Insights Team, which is mentioned in the target article and which has been advised by Nick Chater and George Loewenstein.