We offer a philosophical perspective on this important programmatic article, focusing on three related constructive critiques and suggestions.

Our first point focuses on the background normative framework for public policy evaluation. Following the majority of work in this area, Chater & Loewenstein (C&L) primarily evaluate policies in terms of how effectively they promote welfare. They present a range of evidence that i-frame interventions often fail to benefit their targets, relative to s-frame alternatives. This forms the basis of their central critique of the i-frame agenda, which they argue has been coopted and exploited by private interests opposed to s-frame reforms.

This is a powerful line of criticism. Welfare promotion is obviously one extremely important dimension of policy evaluation. If i-frame interventions yield meagre welfare gains, that is a good reason to oppose them and the private influence that promotes them. But welfare is not the only dimension of evaluation. Alongside maximising human welfare, we also care about a plurality of other goods, such as fairness, equality, prioritising the worst off, as well as democratic values. Crucially, we are often prepared to trade-off some degree of welfare promotion for the sake of these other values (few of us are full-blown utilitarians). This is important because advocates of i-frame interventions (such as “nudge” proponents) often appeal to non-welfarist values, such as liberty and autonomy, in defence of an i-frame approach. So, a complete normative evaluation will need to compare i-frame and s-frame interventions in terms of a plurality of values.

More positively, we want to suggest that C&L's welfare-based critique of i-frame interventions (and the private influence behind them) may be bolstered by reflecting on these wider values. To illustrate, consider two plausible (and non-exclusive) ways of understanding the value of democracy. On one view, democracy is valuable because it gives citizens control over their collective lives, thereby promoting their autonomy (Lovett & Zuehl, Reference Lovett and Zuehl2022). On another view, the value of democracy inheres in the fact that it gives citizens similar levels of political influence, thereby avoiding objectionably inegalitarian social relations (Christiano, Reference Christiano2008; Kolodny, Reference Kolodny2014; Viehoff, Reference Viehoff2014). If, as C&L persuasively argue, i-frame interventions are a means by which private interests exert influence on public policy, this plausibly undermines the values of democratic autonomy and equality (Bartels, Reference Bartels2016, Ch. 11; Christiano, Reference Christiano and Estlund2012; Lovett Reference Lovett2024). When private actors are able to leverage their wealth to influence policy making, this both undermines regular citizens’ control over policy decisions (thereby undermining their autonomy) and places them in a subordinate relationship to the wealthy (thereby undermining equality). Hence, the case against private influence over public policy need not be restricted to its effects on welfare. C&L's critique of the i-frame research agenda can be waged on multiple fronts.

Our second point is that private influence not only comes with normative disadvantages, but has also been shown to be epistemically harmful in a variety of contexts. C&L's critique focuses on how private interests can affect which research questions are asked, what studies are thus carried out, and how they are adapted for policies. However, private influence penetrates deeper, often affecting the actual results of whatever research is carried out and thus how the research questions are answered. For instance, it is a well-known problem in pharmaceutical research that researchers with industry ties are much more likely to produce studies that draw proindustry conclusions, even without any obvious biasing of the research methods used (see, e.g., Lexchin, Bero, Djulbegovic, & Clark, Reference Lexchin, Bero, Djulbegovic and Clark2003). These effects on study results are part of a well-studied suite of mechanisms by which industry influence has subverted the scientific pursuit of truth. These include deliberate (and very subtle) strategies designed to maintain ignorance (Pinto, Reference Pinto2017) or to sustain self-serving consensuses that may diverge from the best evidence (Holman & Bruner, Reference Holman and Bruner2015).

Take, for example, a tactic from the tobacco industry's war on cancer research (Oreskes & Conway, Reference Oreskes and Conway2011) that is evidently still in operation today (Adams, Reference Adams2011). Industry conducts proprietary meta-research into which methods are reliable and which not. With this knowledge in hand they fund less reliable research that investigates (matters pertinent to) policies they wish to subvert. This prevents consensus forming simply by ensuring that enough erroneous results are disseminated to perpetuate academic debate (Weatherall, O'Connor, & Bruner, Reference Weatherall, O'Connor and Bruner2020). Importantly, these strategies for directly influencing and perverting the process and outcomes of scientific enquiry will be available to private interests whether the research questions pertain to s-frame or i-frame interventions.

Finally, in light of these wider problems with private influence (and the subtle and subversive ways in which it operates) we believe the solutions suggested by C&L are insufficient. The overall tenor of their proposal is to stress greater personal awareness of, and willingness on the part of the researcher to personally take steps to counter-act, psychological or methodological biases in favour of i-frame and against s-frame interventions. Somewhat ironically, these recommendations have a rather i-frame flavour: Give researchers information and encourage them to make better individual choices. And, for that reason, they cannot hope to counter the systematic ways in which private interests bias not only the choice of interventions to study and implement, but also the outcomes of research. Moreover, the methods and results of industry science are often proprietary and not shared with the broader community (Bright & Heesen, Reference Bright and Heesen2023). Industry may thus have more information about how to exploit human psychology than outsiders, further undermining the efficacy of C&L's proposed solution. We suggest – turning the article's main contention back on itself – that we should also consider s-frame interventions that target the research field. First and foremost: What can be done to better shield research from the influence of private interests?