Chater & Loewenstein's (C&L's) conclusion that the effects of “i-frame” interventions have been disappointing depends on assumptions about their potential effectiveness. An oversimplified conception of individuals as passive actors – averse to mental effort and reliant on “automatic” responses – is likely to lead to unrealistic expectations regarding what i-frame solutions can achieve. An account that sees agents as active participants making decisions within varied choice environments lends itself towards more realistic expectations (and conclusions) about what can be achieved by i-frame solutions (Hertwig & Grüne-Yanoff, Reference Hertwig and Grüne-Yanoff2017; Newell & Shanks, Reference Newell and Shanks2023; Sher, McKenzie, Müller-Trede, & Leong, Reference Sher, McKenzie, Müller-Trede and Leong2022).

Several commentators have noted that – because of the heterogeneity of policy problems, interventions, populations, and target outcomes – a more nuanced assessment of effectiveness is required. For instance, Simmons, Nelson, and Simonsohn (Reference Simmons, Nelson and Simonsohn2022) conclude that considering only the average effect of nudges can obfuscate the effects of particular nudges (see also Hallsworth, Reference Hallsworth2022). Although DellaVigna and Linos (Reference DellaVigna and Linos2022, p. 83) found an average impact of 1.4 percentage points (not per cent, as reported by C&L) over a control group take-up of 17.3 per cent, the usefulness of such an average when predicting the effectiveness of specific nudge interventions is questionable.

Notwithstanding contentions regarding the impact of i-frame interventions, the critical question is whether, overall, such interventions do harm by detracting from s-frame policies. Although this may sometimes occur – and so C&L provide a salutary warning – the case studies did not fully convince us of the extent of the problem. For example, although the BP carbon footprint campaign may have been successful in undermining efforts for systemic responses to climate change, we are unsure whether those efforts were further undermined by the i-frame interventions of behavioural scientists.

Moreover, because opportunities for systemic reform only occur infrequently (for instance, following a change of government, a policy review, or a major system failure), there are lengthy periods where important but less dramatic policy improvements can be made, and where there are limited i-/s-frame trade-offs because there is little political support or momentum for broader reform.

For instance, the Australian Government's central behavioural policy team, BETA, has worked on a wide range of i-frame interventions related to the details of regulatory design such as: Consumer bills or activity statements, product labels, consumer information sheets, and registration processes. These measures are pertinent because they all work within the existing regulatory framework and thus could detract from broader s-frame changes to that framework. In our judgement, however, this was rarely if ever the case.

Indeed, in some instances, i-frame interventions supported broader, systemic changes. For example, to combat harm from online wagering, Australian federal, state, and territory governments adopted a national framework that included prohibitions on lines of credit, inducements, and advertising of payday lending (Australian Government, 2018). It also included measures to help individuals manage their gambling, such as a requirement that customers receive meaningful activity statements from online gambling providers. A collaboration between academics and government officials tested statement designs on a simulated gambling platform and found the statements had a modest but material impact on the amount bet (Australian Government, 2020).

Such instances of i-frame solutions being introduced as part of broader systemic change question conceptions of i- and s-frame solutions as competing with one another. C&L, in their discussion on how s-frame reforms reduced smoking in the United States, point out that some policy solutions (i.e., mandated changes to package labelling) have an i-frame “flavor” thereby highlighting that delineations between the two frames can be somewhat artificial. Meaningful change is likely to emerge through a combination of changes to both the system itself and to the interface between the individuals and the system; as both represent a change to the environment in which decisions are made; they are likely to blend together as part of a battery of solutions aimed at a particular problem.

Furthermore, there are numerous i-frame interventions that do not compete with s-frame reforms. For example, employers have an important but challenging role in supporting workers to return to work after an extended illness or injury. Workplace regulations (system-level policies) alone are insufficient because supporting the return to work is an infrequent and atypical management challenge. Consequently, well-designed, timely materials for managers are likely to improve the return-to-work experience, without undermining any system-level reforms in the process (Australian Government, 2019, 2022). We expect that C&L would not quarrel with such work but we feel it deserves greater attention.

Nevertheless, we agree that behavioural scientists should consider how they can contribute to systemic reforms. We offer several suggestions that build on those offered by C&L. First, modesty – behavioural scientists should not overhype the potential impact of i-frame interventions beyond what is justified by their typically modest results. This will reduce the risk that i-frame interventions detract from broader s-frame measures. Second, trade-offs – behavioural scientists should be mindful of any trade-offs between i-frame and s-frame initiatives because of the political context or scarcity of academic or bureaucratic resources. Third, the stalking horse – working with government agencies on i-frame interventions can provide behavioural scientists with a valuable avenue to advise on behavioural insights relevant to systemic reforms. For example, designing consumer comparison sites may reveal complexity in financial products that the comparison site cannot readily simplify, and where regulation may be warranted. Fourth, problem diagnosis – BETA and other behavioural policy units have already contributed to s-frame interventions through better diagnosis of the policy problem. As C&L suggest, there is likely more that behavioural scientists could contribute here.

A final thought is the role that i-frame interventions can play in simply raising awareness in the general public of the need for behaviour change, and in turn increasing the potential for the public to support (i.e., vote for) system-wide reforms. Returning to the carbon-footprint example, the evidence that carbon-footprint calculators actually reduce personal emissions is limited and mixed (Dreijerink & Paradies, Reference Dreijerink and Paradies2020), but knowledge about how our personal actions can collectively make a difference in tackling environmental problems can be a powerful motivator for supporting proenvironmental action (Newell & Moss, Reference Newell and Moss2021; Xie, Brewer, Hayes, McDonald, & Newell, Reference Xie, Brewer, Hayes, McDonald and Newell2019).