Introduction
The past two decades have witnessed a growing reliance on behavioral insights in public policy. Interventions described as nudges or boosts—which alter choice environments or provide information, respectively (Thaler and Sunstein, Reference Thaler and Sunstein2008; Grüne-yanoff and Hertwig, Reference Grüne-yanoff and Hertwig2016)—have proliferated across governments, NGOs and industries worldwide (OECD, 2018; Sanders et al., Reference Sanders, Snijders and Hallsworth2018). Many of these interventions have shown promise for improving health, financial decisions and other socially relevant outcomes (DellaVigna and Linos, Reference DellaVigna and Linos2022; De Ridder et al., Reference De Ridder, Kroese and Van Gestel2022). Critics, however, have pointed out that focusing too heavily on top-down, individually oriented interventions can overshadow system-level sources of societal change such as changes in taxation, legal regulation, or grassroots mobilization (Chater and Loewenstein, Reference Chater and Loewenstein2022). Further concerns include the worry that paternalistic top-down steering may compromise individual agency (Noggle, Reference Noggle2018; Banerjee et al., Reference Banerjee, Grüne-Yanoff, John and Moseley2024), and behavioral public policy (BPP) might crowd out other policies that might more efficiently address large-scale social or environmental problems (Mols et al., Reference Mols, Haslam, Jetten and Steffens2015).
In response, a growing body of work has called for a more agency-centered BPP. Recent contributions argue that behavioral interventions should support, rather than bypass, people’s capacity for intentional action, reflection and endorsement of the policy goals (Banerjee and John, Reference Banerjee and John2024; Banerjee et al., Reference Banerjee, Grüne-Yanoff, John and Moseley2024). Related approaches such as nudge+ emphasize reflection and deliberation as complements to choice architecture (John et al., Reference John, Smith and Stoker2009). These developments mark an important shift away from purely technocratic, top-down models of behavior change. Despite these advances, however, two important gaps remain.
First, much of the agency-centered literature remains individualistic in scope. Agency is primarily discussed in terms of individual intentional action, reflection, or choice endorsement, with comparatively little attention to collective agency (Ostrom, Reference Ostrom1998; Cleaver, Reference Cleaver2007; Bratman, Reference Bratman2014; Rovane, Reference Rovane2019).
Collective agency is required for groups, communities, or publics to shape the policies that govern them. Many of the most pressing societal challenges targeted by BPP – e.g., pandemics, climate change and misinformation—are inherently collective problems, where outcomes depend on shared practices, coordination and social norms. Therefore, BPP truly seeking to both respect and promote agency must extend beyond individual choice to consider how agency can be exercised collectively within policy development.
Second, despite calls for bottom-up or participatory approaches, BPP has largely lacked operational guidance on how such approaches can be systematically implemented across the policy cycle. While extensive literature exists on co-design, co-production and participatory governance—particularly in public administration, policy design and health services research (Ostrom, Reference Ostrom1996; Voorberg et al., Reference Voorberg, Bekkers and Tummers2015; Blomkamp, Reference Blomkamp2018)—these insights have not been fully integrated into BPP practice. As a result, BPP continues to rely predominantly on top-down intervention design, supplemented by hearing-based consultations rather than genuine co-creation.
This paper pushes bottom-up BPP two steps forward. Step one is conceptual. We clarify what a genuinely bottom-up, agency-centered BPP requires by distinguishing individual from collective agency and by carefully articulating the relationship between the many aspects of human agency and participatory policy development. To do so, we deploy the concept of co-creation, understood as the active involvement of stakeholders throughout all stages of the development of policies and services (Ostrom, Reference Ostrom1996; Ind and Coates, Reference Ind and Coates2013). Drawing on philosophical and policy scholarship, we argue that concerns about autonomy and paternalism cannot be fully addressed without attending to whether and how citizens are co-creating the interventions that affect them.
Step two is methodological. We show how co-creation can be systematically integrated with a behavioral insights approach. Rather than proposing co-creation as a novel framework or governance tool, we synthesize and adapt existing participatory approaches to the specific epistemic, ethical and practical challenges of behavioral intervention design. We outline a behaviorally informed co-creation cycle that is systematically rigorous while embedding stakeholder agency at each phase, from problem definition to scaling. We conclude by discussing the limitations of co-creation, the conditions under which it may be inappropriate and directions for future empirical research.
From top-down steering to bottom-up participation
While top-down and individual-level interventions are warranted in a range of contexts, for example, in guiding health professionals to adhere to evidence-based best practices (Yoong et al., Reference Yoong, Hall, Stacey, Grady, Sutherland, Wyse, Anderson, Nathan and Wolfenden2020; Last et al., Reference Last, Buttenheim, Timon, Mitra and Beidas2021; Nwafor et al., Reference Nwafor, Singh, Collier, DeLeon, Osborne and DeYoung2021), these interventions may nonetheless suffer from reactance responses (Invernizzi et al., Reference Invernizzi, Falomir-pichastor, Muñntoz-rojas and Mugny2003), ineffective design (Imperial, Reference Imperial2021), or low stakeholder commitment (De Las Cuevas et al., Reference De Las Cuevas, Peñate and De Rivera2014). Furthermore, they face the well-known issue that policy developers themselves are cognitively bounded agents. Complex problems that require large-scale coordination or involve complex causal mechanisms run a particularly high risk of misrepresentation and mismanagement (Galesic et al., Reference Galesic, Barkoczi, Berdahl, Biro, Carbone, Giannoccaro, Gonzalez, Kandler, Kao, Kendal, Kline, Lee, Massari, Mesoudi, Olsson, Pescetelli, Sloman, Smaldino and Stein2023). To collectively adapt to large-scale crises, it is critical to understand how bottom-up change happens, how it interacts with top-down structures and how it can be best harnessed in practice.
Connolly et al. (Reference Connolly, Loewenstein and Chater2024) argue that while it is widely known that bottom-up change has been the driver of multiple large-scale societal changes, BPP has to a broad extent restricted itself to top-down interventions. Behavioral scientists, in their view, should instead take an interest in issues that have traditionally been the bailiwick of social scientists, such as the psychological drivers of protest and change—as well as psychological drivers of resistance to change. Their criticism reflects a widespread dichotomy where researchers see themselves as either merely describing how people change their views and practices, or as contributing to policy development. Yet, as with many apparent dichotomies, a range of shades of gray can be found in between the two extremes. In particular, hybrid forms of political engagement such as participatory governance (Heinelt, Reference Heinelt2010) reflect approaches where the boundary between experts and policymakers, and the people they serve, is relaxed. These practices endeavor to create channels for bottom-up change, striving toward the ideal that the people would not just be heard in the development of new policy, but instead the people would actively produce the policy in question.
While approaches that actively involve citizens in policy design and production of public services are nothing new (Brandsen et al., Reference Brandsen, Steen and Verschuere2018), they are only now being noted in the behavior change literature. Such approaches include, for example, participatory design, where stakeholders are involved in the design process (Robertson and Simonsen, Reference Robertson and Simonsen2012) and its expanded version co-creation, where stakeholders’ inclusion expands to cover the broader process (Blomkamp, Reference Blomkamp2018). Here, we use the term co-creation as defined by Voorberg and colleagues (Voorberg et al., Reference Voorberg, Bekkers and Tummers2015) to refer both to co-creation in the private sector to design products and services (Ind and Coates, Reference Ind and Coates2013), and co-production in the public sector to develop policies and services (Ostrom, Reference Ostrom1996). From the research perspective, a close relative is the Participatory Action Research approach, where researchers actively collaborate with communities to interactively generate data to collectively arrive at action solutions (Baum et al., Reference Baum, MacDougall and Smith2006; Cornish et al., Reference Cornish, Breton, Moreno-tabarez, Delgado, Rua, De-graft Aikins and Hodgetts2023).
Co-creating behavior change
As co-creation has gained popularity across research, design and policy co-production, there have been calls to deploy co-created approaches for behavior change, which invite citizens to help design the policies that ultimately shape their daily choices (Røiseland, Reference Røiseland2024). Co-creation can endow policy with added legitimacy and responsiveness to local needs (Brandsen et al., Reference Brandsen, Steen and Verschuere2018; Ansell et al., Reference Ansell, Sørensen and Torfing2023). Yet, BPP seldom emphasizes participatory or co-created policy mechanisms – even though these may more thoroughly engage the community, minimize paternalism and yield long-term changes in collective practices (Connolly et al., Reference Connolly, Loewenstein and Chater2024). For example, Banerjee et al. (Reference Banerjee, Grüne-Yanoff, John and Moseley2024), in highlighting the need for BPP to promote rather than undermine the individual’s capacity to act for reasons, restrict their discussion to the intentional action of individuals (Davidson, Reference Davidson2001) over collective agency (Cleaver, Reference Cleaver2007; Bratman, Reference Bratman2014; Rovane, Reference Rovane2019; Sotarauta and Grillitsch, Reference Sotarauta and Grillitsch2023). Co-creation can help address this limitation by supporting collective agency through facilitating community-led policy development.
There are many ways to involve stakeholders in the development of policies and services, but not all of them involve genuine bottom-up co-creation. For example, value-sensitive design (Friedman and Hendry, Reference Friedman and Hendry2019) is a participatory design process used chiefly in the contexts of technology and architecture. In these contexts, value-sensitive design involves mapping out stakeholders’ values concerning the development project in question, allowing designers to take these values into account. This helps circumvent the risk that designers would presume to know what stakeholder values are without asking them. However, value-sensitive design is not a robustly democratic process. Instead, it is an example of a hearing-based, top-down process where the stakeholders are heard but they do not directly participate in the design process itself. Likewise, approaches centered on the rhetoric of ‘listening’ (to citizens, diverse voices, cultures, etc.) imply a listener who ultimately may not engage the heard parties in bottom-up policy development.
In addition to hearing-based models of participation, there are multiple ways to embed genuinely bottom-up elements into the development of behavioral policy. Here, behavioral policy can borrow insights not just from co-creation of policy in other contexts, such as public health (Vargas et al., Reference Vargas, Whelan, Brimblecombe and Allender2022) and climate (Hofstad et al., Reference Hofstad, Sørensen, Torfing and Vedeld2023).
Co-creation has potential benefits across multiple domains. The practical and political benefits of co-creation range from the enhancement of innovation processes by bringing more viewpoints and resources to the table (Nambisan, Reference Nambisan2002; Prahalad and Ramaswamy, Reference Prahalad and Ramaswamy2004; Perks et al., Reference Perks, Gruber and Edvardsson2012). They can boost supply chain integration (Jüttner et al., Reference Jüttner, Christopher and Godsell2010) and increase employee engagement (Hatch and Schultz, Reference Hatch and Schultz2010), knowledge sharing (Kohlbacher, Reference Kohlbacher2008) and stakeholder commitment (Randall et al., Reference Randall, Gravier and Prybutok2011). Most importantly, though, co-creation can foster democratic participation, helping people influence decisions, policies and services (Schuler and Namioka, Reference Schuler and Namioka2017). In the context of BPP, co-creation can help increase legitimacy and trust in governance (Brandsen et al., Reference Brandsen, Steen and Verschuere2018; Ansell et al., Reference Ansell, Sørensen and Torfing2023).
Highlighting these issues, Ostrom (Reference Ostrom1996) uses the concept of co-production to describe how ‘citizens can play an active role in producing public goods and services of consequence to them’ (p. 1073). Co-production was developed to address the observation that public services fail when they treat citizens as passive recipients of policy actions rather than active contributors (Parks et al., Reference Parks, Baker, Kiser, Oakerson, Ostrom, Ostrom, Percy, Vandivort, Whitaker and Wilson1981; Ostrom, Reference Ostrom1996). Ostrom’s research program on common-pool resources (CPRs) constitutes an extensive empirical demonstration that communities can create, monitor and adapt rules to govern their own behavior (Ostrom, Reference Ostrom1990, Reference Ostrom1998), contradicting the first-generation rational choice models that predicted collective action failure without external enforcement (Ostrom, Reference Ostrom1998). Furthermore, Ostrom’s argument was explicitly a behavioral one: boundedly rational agents reliant on heuristics and norms, cooperating on the basis of trust, reciprocity and reputation. Crucially, and contrary to game-theoretical predictions, experimental evidence showed cooperation to increase in collective dilemmas following face-to-face communication (Ostrom, Reference Ostrom1998, Reference Ostrom2000). While mainstream BPP has developed adjacent individualist critiques on human design and error, e.g., cataloguing biases and designing choice-architecture interventions that harness them, Ostrom’s critique targets the level of collective decision-making. Our approach aims to reconnect these two branches.
Regardless of the similarities, there are also differences. CPR self-governance is endogenous and continuous, whereas BPP co-creation is commonly initiated by an institution or a public organization, is time-bound, and often still focuses on individual behavioral targets. In short, co-creation is not self-governance, but a means to reproduce within a policy cycle the conditions of successful collectives: face-to-face communication, participation and transparent cooperative behavior.
Finally, the common framing clarifies the necessity of collective agency for the key problems addressed by BPP. Climate change, pandemics and misinformation are all problems of the commons, and the cooperation required to address them is contingent on trust, reciprocity and normative beliefs. Participatory approaches are in a unique position to reveal these collective structures at depth beyond that of individual approaches.
Promoting agency across bottom-up policy development
Co-creation offers policymakers a way to systematically center citizen and stakeholder agency in policy development. In this section, we draw on the philosophical literatures on agency and action to discuss agency as a multidimensional concept, mapping out aspects of agency that are relevant for BPP. We then outline how co-creation can move behavioral policy from an approach that protects agency to one that promotes it.
BPP has always been motivated by a desire to safeguard human agency: part of its putative appeal, in contrast to bans and regulations, is to maintain individuals’ agency over their own choices (Thaler and Sunstein, Reference Thaler and Sunstein2008). Agency, for such an approach, is understood as individuals’ capacity to make choices about their own lives, while behavioral interventions are designed to increase the likelihood that people’s intentions and actions are aligned with aims of public policy, such as health or environmental considerations.
However, agency is not just about making choices in the absence of constraint. This point enjoys widespread consensus status in philosophical ethics and philosophy of action. This move is also present in discussions concerning the ethics of BPP, where critics have argued that, due to appealing to cognitive biases, nudges may in fact undermine agency despite not constraining choice (John et al., Reference John, Smith and Stoker2009; Hansen, Reference Hansen2016; Noggle, Reference Noggle2018; Kuyer and Gordijn, Reference Kuyer and Gordijn2023). The philosophical literature on agency is rich with different accounts of agency and action. In this section, we will refrain from endorsing any single theory. Instead, below, we map out aspects of agency drawing on a variety of philosophical perspectives and how they bear on BPP.
Reasons-responsiveness
On one perspective, a central aspect of agency is making choices in a way that responds to agents’ reasons for action (Bratman, Reference Bratman1999; Anscombe, Reference Anscombe2000; Davidson, Reference Davidson2001). Accordingly, perspectives highlighting that BPP can, and should, engage citizens in reflection and steer choice by appealing to reasons for action abound. These perspectives are a central motivation for boosts (Grüne-yanoff and Hertwig, Reference Grüne-yanoff and Hertwig2016; Grüne-yanoff et al., Reference Grüne-yanoff, Marchionni and Feufel2018), although Levy defends nudges, arguing that they, too, appeal to reasons (Levy, Reference Levy2019).
Reflection
While some question whether reflection is necessary for agency (see Shepherd, Reference Shepherd2021), it is widely agreed that pre-choice reflection displays agency (Frankfurt, Reference Frankfurt1971; Korsgaard, Reference Korsgaard1996). Accordingly, John, Banerjee and colleagues have proposed the nudge+ approach which incorporates reflection prompts into the intervention design (John et al., Reference John, Smith and Stoker2009; Banerjee and John, Reference Banerjee and John2024; Banerjee et al., Reference Banerjee, Grüne-Yanoff, John and Moseley2024).
While the above aspects of agency are well-embedded in the BPP literature, below we outline further aspects of agency that have not received adequate attention in the literature.
Efficacy beliefs
An individual’s capacity to exercise their agency is undermined if the individual does not believe they are capable of meaningful action (Koi, Reference Koi2023; Morton, Reference Morton2024; Martikainen, Reference Martikainen2025). Accordingly, self-efficacy beliefs (Bandura, Reference Bandura1986) emerge as an important aspect of agency.
Sense of agency and ownership
A distinctive aspect of human agency is that it feels like something to exercise one’s agency. For example, this can include subjective experiences of ownership, effortfulness, efficacy and control: the senses that ‘this is my action’, ‘I am doing something difficult’ and ‘I am in control of this’ (Cleaver, Reference Cleaver2007; Gallagher, Reference Gallagher2007; Pacherie, Reference Pacherie2007; Lukitsch, Reference Lukitsch2020). Some forms of BPP are felt to change our behavior without sustaining senses of ownership and control, so it is unsurprising if these forms leave some feeling that their agency has been sidestepped (Hansen, Reference Hansen2016; Ivanković and Engelen, Reference Ivanković and Engelen2019; Banerjee et al., Reference Banerjee, Grüne-Yanoff, John and Moseley2024). By contrast, co-creation centers stakeholders as active producers of the policy or intervention in question. Active participation in the co-creation process fosters a sense of authorship and ownership of the resulting policy and innovation, improving commitment (Hatch and Schultz, Reference Hatch and Schultz2010).
Self-control
In contrast to mere sense of control, self-control entails the actual regulation of one’s own behavior despite motivational conflicts (Kennett, Reference Kennett2001; Mele, Reference Mele2001; Koi, Reference Koi2023). This can involve acting as a ‘citizen choice architect’ (Reijula and Hertwig, Reference Reijula and Hertwig2022), altering one’s environment to steer one’s future behavior (Koi, Reference Koi2026).
Social embeddedness
Beyond the individual framing above, agency is also a robustly social phenomenon. Increasingly, theories of agency recognize that agency is, at least in part, a relational phenomenon built on recognizing each other as agents (Honneth, Reference Honneth1996; Buss, Reference Buss2005; Mackenzie, Reference Mackenzie2008; Darwall, Reference Darwall2009). Our agency is then in part constituted by being seen and treated as an agent, and experiences of belonging and participation therefore become crucial for agency (Malik and Obhi, Reference Malik and Obhi2019; Martikainen, Reference Martikainen2025).
Collective agency
While social embeddedness highlights that the agency of individuals emerges in mutual recognition, it is also widely appreciated that groups and collectives also have agency, where the actions taken as a group or collective (e.g., the actions of an organization, a group of friends or a corporation) may not be fully reducible to the actions of individuals (List and Pettit, Reference List and Pettit2011; Bratman, Reference Bratman2014; Gilbert, Reference Gilbert2017; Rovane, Reference Rovane2019).
While the above overview is not exhaustive, it highlights how multifaceted agency is. In particular, it highlights how moving beyond a focus on individual choice and toward an s-frame approach (Chater and Loewenstein, Reference Chater and Loewenstein2022; Connolly et al., Reference Connolly, Loewenstein and Chater2024) can help address agency as a social phenomenon, where participation is founded on recognition and active inclusion.
Hearing-based approaches explicitly invite participants to articulate their own values and preferences, minimizing paternalistic top-down assumptions. This helps mitigate concerns about covert forms of behavioral influence (Noggle, Reference Noggle2018; Ivanković and Engelen, Reference Ivanković and Engelen2019). However, while the heard parties are engaged in reasons-responsive reflection, other aspects of agency remain with the choice architects—much like, in a letter to Santa Claus, a child can deliberate on what they wish for, but they ultimately have no agency over whether and how their wishes are fulfilled. Instead, efficacy, sense of agency and ownership, and control all lie with Santa. The child is recognized as an informant with desires, but not as a co-producer of the desired change. Co-creation can help address this imbalance.
Ideally, co-creation involves collaborating with stakeholders from the outset, engaging them both in identifying the problems to be addressed and in the development of the intervention. This can be continued through all stages of intervention development; we expand on this below. By engaging participants in creating the intervention, co-creation goes beyond engaging participants in reflection. Instead, it fosters a sense of agency and ownership of the intervention under development: it becomes ‘our intervention’, something participants have made efforts toward and feel shared control over. It signals to participants that they are welcome at the table, bolstering self-efficacy, fostering social embeddedness and allowing stakeholders and practitioners to act as a group, forming shared commitments to action. Beyond ethics, facilitating participants’ sense of agency and ownership promises practical benefits: it can lead to higher adherence, commitment and potential for sustained behavior change.
Accounting for these aspects of agency, co-creation may help deliver a categorical change in how BPP addresses agency: not just by safeguarding and respecting it (through non-interference), but also by actively scaffolding and promoting agency and democratic participation (Cleaver, Reference Cleaver2007; Heinelt, Reference Heinelt2010).
Co-creation is uniquely positioned to avoid the often-cited worries that BPP would violate personal autonomy (Engelen and Nys, Reference Engelen and Nys2020; Schmidt and Engelen, Reference Schmidt and Engelen2020; Wachner et al., Reference Wachner, Adriaanse and De Ridder2020). Roughly, a person’s autonomy is compromised if their choices are being shaped by another party rather than by themselves. Co-creation addresses the root cause of the autonomy concern. By actively producing the intervention, stakeholders make it something shaped by themselves, rather than by another party. If a co-creation process has been successfully carried out, the resulting intervention will therefore not be autonomy-undermining but will instead be an instance of collective self-regulatory action.
Ethical and practical advantages of co-creation
Top-down policymaking has faced concerns that policymakers’ and practitioners’ views on what people should prefer and choose might not align with those of the people themselves, and top-down interventions may be unduly paternalistic, not respecting people’s autonomy and agency (Scoccia, Reference Scoccia2019). Further concerns include that interventions may not be transparent to their targets (Ivanković and Engelen, Reference Ivanković and Engelen2019), making behavioral interventions manipulative (Noggle, Reference Noggle2018).
As discussed above, co-creation can help public policy recognize, respect and support agency. It can be a step toward seeing and treating people affected by policy as agents and active participants rather than as passive targets of an intervention. In addition, co-creation offers other potential advantages, both ethical and practical. We overview them below.
Improving intervention relevance and fit
Expert choice architects may miss features of the intervention context. By engaging target populations early and at depth, co-creation fosters the creation of interventions that resonate with participants’ needs, values, practices and beliefs.
Identifying externalities and spillover effects
Intervention development faces the challenge of identifying and addressing externalities and spillover effects before they occur (Claassen, Reference Claassen2016; Kuyer and Gordijn, Reference Kuyer and Gordijn2023). Co-creation creates a space where externalities can more readily be identified in advance. Hearing-based approaches achieve this in a more limited fashion. In those approaches, stakeholders are consulted, e.g., on their views on a specific intervention under development, and they may be able to identify effects that the intervention would have beyond the target behavior. By enlisting stakeholders in the development and ideation processes, co-creation fosters a more open-ended discussion. As stakeholders consider different options and generate ideas, this helps work out the broader impact of the proposed alternatives.
Promoting fair participation and addressing power imbalances
While nudges and other behavioral interventions are supposed to be ‘easy and cheap to avoid’, this aim does not always pan out as planned. What seems easy and cheap for policymakers may not be easy and cheap for everyone (Roberts, Reference Roberts2018; Fowler and Roberts, Reference Fowler and Roberts2019; Siipi and Koi, Reference Siipi and Koi2022). For example, Ghesla et al. (Reference Ghesla, Grieder and Schubert2020) found that while default nudges are effective in reducing greenhouse gas emissions, they result in low-income households paying more for electricity than they would prefer, contributing to social inequalities. More broadly, socioeconomically disadvantaged populations may be nudge-proof in contexts where they lack the requisite resources to uptake a given behavior, whereas in other contexts, the same population may be more ‘nudgeable’, lacking the resources to monitor for nudge presence or avoid a default behavior (Roberts, Reference Roberts2018).
Here, co-creation facilitates tailoring the intervention to these social realities. How easy or hard this is to accomplish depends crucially on the stakeholder groups involved. Deliberative democratic processes often fail to engage the entire citizenry: disadvantaged people can experience barriers to full participation in long-term deliberation and democratic institutions that other social strata do not encounter, a phenomenon sometimes termed ‘political poverty’ (Çelik, Reference Çelik2023; Morton, Reference Morton2024; Martikainen, Reference Martikainen2025). In developing interventions that target the whole citizenry or disadvantaged groups specifically, policymakers must invest time and resources in enabling participation for these groups, and the desired representativeness may not be reached within a single initiative, although it may cumulatively improve over time.
Even as we emphasize the difficulty of achieving desired representation where systemic socioeconomic inequalities are at stake, co-creation remains valuable in these contexts. Even an uneven participant representation in the co-creation process can help improve citizens’ experiences of agency and ownership of the intervention and can help not just identify but also address ethical risks.
Improved transparency
Transparency is a key ethical concern for BPP. For many forms of nudging, it is not obvious to the target population that they are being nudged, which makes them harder to avoid, and, some argue, undermines nudges’ legitimacy (Hansen and Jespersen, Reference Hansen and Jespersen2013; Ivanković and Engelen, Reference Ivanković and Engelen2019; Schmidt and Engelen, Reference Schmidt and Engelen2020). By inviting citizens and stakeholders to actively contribute to policy design, co-creation facilitates stakeholder awareness of the policy under development.
Transparency has two prongs: transparency about the presence of an intervention (e.g., the people are informed that a default rule is used) and transparency about the purpose of the intervention, that is, informing the people about what the policymakers aim to accomplish with the default (Hansen and Jespersen, Reference Hansen and Jespersen2013; Ivanković and Engelen, Reference Ivanković and Engelen2019; Siipi and Koi, Reference Siipi and Koi2022). The latter prong is tightly connected to concerns about manipulation: on some accounts of manipulation, roughly, a person is being manipulated if the manipulator’s real intentions are not made known (Bělohrad, Reference Bělohrad2019).
Appreciating the two-pronged nature of transparency, transparency is not a given even for co-creation. While co-creation processes effectively improve stakeholder awareness that an intervention is being developed, in principle facilitators could neglect to inform participants of the aims of the co-creation process, which may make a difference to stakeholder values (Ivanković and Engelen, Reference Ivanković and Engelen2019; Siipi and Koi, Reference Siipi and Koi2022). Ensuring the transparency of the aims of the process is paramount to ensuring the process and resulting intervention will not be experienced as manipulative.
Improved intervention scalability and sustainability
When communities play an active role in shaping an intervention, they are more likely to champion it and refine it over time. Co-creation thus extends beyond a one-off trial, creating capacity for continual adaptation to new conditions (Brandsen et al., Reference Brandsen, Steen and Verschuere2018).
Overall, co-creation promises improved participation and agency for stakeholders, but it is not a silver bullet for ensuring equitable participation. In co-creation, practitioners intentionally center stakeholders, fostering a sense of equality where both groups engage as peers. However, achieving a truly egalitarian process can be challenging. This is because practitioners also play a crucial role in identifying the relevant stakeholders and facilitating interactions that productively and impartially produce insight into stakeholder values and enable stakeholders’ active participation. This is a sensitive process, and deliberative approaches to governance have sometimes been criticized for their propensity for performativity and bias (Solomon, Reference Solomon2006; Fishkin, Reference Fishkin2009; Sunstein, Reference Sunstein2017; Setälä and Smith, Reference Setälä, Smith, Bächtiger, Dryzek, Mansbridge and Warren2018). While a deliberative process usually requires some structuring, some putatively deliberative processes have been criticized for being ‘stage-managed’ and falling short of a genuine dialogue (Maginn, Reference Maginn2007). Even when the dialogue is not overly managed and scripted by facilitators, stakeholders with the most resources, charisma, or outspokenness may dominate conversations, potentially drawing disproportionate attention to their perspectives. These power imbalances often replicate broader societal structural biases (Solomon, Reference Solomon2006; Fishkin, Reference Fishkin2009).
Stakeholder groups can also fall into ‘groupthink’ where one perspective becomes a shared perspective, and dissenting views or alternatives are not explored (Solomon, Reference Solomon2006; Sunstein and Hastie, Reference Sunstein and Hastie2015). This behavior is driven by a need to conform and to avoid the social cost of conflicts and disagreement (Terry and Hogg, Reference Terry and Hogg1996). Several solutions on how to avoid groupthink and ensure a balance and plurality of perspectives have been proposed and applied. These include strategies of discussion facilitation (Manz and Neck, Reference Manz and Neck1995; Russell et al., Reference Russell, Hawthorne and Buchak2015; Sunstein and Hastie, Reference Sunstein and Hastie2015) and the creation of dedicated spaces for reflection and listening (Ercan et al., Reference Ercan, Hendriks and Dryzek2019). More recently, psychological safety—the belief that interpersonal risks in a group can be taken without deleterious consequences—has been linked with effective group functioning (Edmondson, Reference Edmondson1999; Baer and Frese, Reference Baer and Frese2003; Frazier et al., Reference Frazier, Fainshmidt, Klinger, Pezeshkan and Vracheva2017; Newman et al., Reference Newman, Donohue and Eva2017). In what follows, as we describe how co-creation can be deployed in the context of BPP, we will refer to specific practices designed to help structure the co-creation process in a manner that helps mitigate the risk of harmful group phenomena.
Integrating co-creation into intervention development
In this section, we propose a multi-step co-creation cycle that builds on both participatory design (Robertson and Simonsen, Reference Robertson and Simonsen2012; Schuler and Namioka, Reference Schuler and Namioka2017) and community-based participatory research (Baum et al., Reference Baum, MacDougall and Smith2006; Bergold and Thomas, Reference Bergold and Thomas2012; Friesen et al., Reference Friesen, Lignou, Sheehan and Singh2021). The aim is twofold: (1) to give policymakers and researchers a clear blueprint for incorporating stakeholder voices early and often, and (2) to preserve rigor in the form of structured diagnostic, design and evaluation methods.
We describe ways to integrate a co-creation approach into a typical development process for behavioral interventions. While no single template will suit every policy domain, the framework below can be flexibly adapted. Implementation details—such as how to solicit stakeholder feedback or which data-collection tools to use—can be calibrated to local contexts, resource constraints and the complexity of the target behavior. The guiding principles of transparency, psychological safety and mutual respect inform the process. Taken together, these elements help guard against performativity and bias.
Importantly, while bottom-up co-creation ideally is integrated throughout the intervention process, it is also possible to deploy co-creation strategically in select phases of the process, or to incorporate it in later phases even when the process has started out as top-down or hearing-based (see Table 1).
An idealized schematic of stakeholder integration in different stages of the development of behavioral public policy
Table 1 Long description
Stakeholder roles are compared across five policy development phases under three approaches: top-down, bottom-up, and co-creation. In defining the target behavior, top-down has experts or decision-makers set the target, bottom-up adds stakeholder input through town halls, interviews, or polling, and co-creation has stakeholders initiate or jointly define the target. For understanding the problem and choice architecture, top-down may consult stakeholders on specific issues, bottom-up uses interviews or polls for insight, and co-creation involves stakeholders in a structured process to identify choice points plus barriers and drivers. When developing interventions, top-down keeps design with experts, bottom-up gathers reactions to proposed interventions, and co-creation uses deliberative group work with stakeholders and experts as peers. In validation and pilots, top-down and bottom-up mainly use stakeholders as participants, while co-creation includes stakeholders in implementation and refinement. During scaling, top-down is led by experts and decision-makers, bottom-up seeks stakeholder feedback before rollout, and co-creation has stakeholders actively scale and tailor interventions to new contexts.
Defining the intervention target behavior
The first phase of the co-creation process is the clarification of the aims and the overall desired result of the intervention. For example, in the health context, these may include reducing the number of unnecessary infections or increasing adherence to treatment, whereas in the climate context, these may include reducing household energy consumption, increasing the proportion of sustainably produced goods consumed, or decreasing city-wide carbon emissions from traffic.
In top-down interventions, these aims are defined by policymakers, researchers and practitioners. Top-down interventions can implement hearing-based interaction with stakeholders via, e.g., town halls, interviews, surveys or polling. Bottom-up approaches to defining the target behavior can take multiple forms. Sometimes, the target behavior emerges from a stakeholder initiative. In other cases, policymakers can facilitate the bottom-up production of a target. In those cases, policymakers still predefine the target domain, as they need a preliminary idea of the behavioral domain to recruit relevant stakeholder groups. Facilitators should also be prepared to shift their ideas of relevant stakeholders as the target domain gets better defined.
Risks at the definition phase include poorly defined target or operationalization, which lead to cumulative problems down the line. Poorly managed stakeholder selection risks leading to a lack of relevant viewpoints and biased decisions.
Outside purely bottom-up citizen initiatives, in practice, the intervention target behavior is co-created within a predefined broader scope. Here, however, co-creation can still define the specific target behaviors for an intervention. For example, when the broader scope is cutting emissions from commute, co-creation can still help address which specific behaviors an intervention should promote (e.g., active commute versus public transport or working from home).
Understanding problem structures and charting choice-architecture
Understanding the choice architecture within which the behavior takes place requires identifying both the relevant choice points, and the drivers and barriers that affect each choice point. Here, top-down interventions are likely to produce suboptimal processes, as they may lack insight into how various choice points are interlinked in the context of the intervention (John and Stoker, Reference John and Stoker2019; Galesic et al., Reference Galesic, Barkoczi, Berdahl, Biro, Carbone, Giannoccaro, Gonzalez, Kandler, Kao, Kendal, Kline, Lee, Massari, Mesoudi, Olsson, Pescetelli, Sloman, Smaldino and Stein2023).
Still, co-creation is not necessary in all contexts. In some situations, such as the design of an organ donor list, simple survey and interview methods are often more convenient than co-creation. Similarly, hearing-based methods where experts use interviews and town halls to chart out the choice architecture and identify behavioral bottlenecks may be used when the experts have a clear enough view on the problem structure. However, surveys, interviews and town halls often lead to either the majority or the most motivated viewpoints being emphasized and may sideline alternative or minority perspectives (Solomon, Reference Solomon2006; Fishkin, Reference Fishkin2009; Sunstein, Reference Sunstein2017).
For cases with high heterogeneity and complexity, a bottom-up co-creation workshop allows stakeholders to collectively carry out the process of defining the crucial choices and their interrelations. The charting begins with a general collective ideation of possible choices related to the target. Here, the facilitator should actively engage the whole group to minimize the formation of active-passive group polarization.
Once enough choice points have been identified, each will be considered for specific barriers and drivers. This process can be aided, for example, by asking for or situating the choices within concrete examples, or by asking directly about specific barriers (e.g., ‘How do you think others would react to you changing your behavior?’, ‘Are there any features in the physical environment that make this behavior difficult?’), or using visual timelines. When workshops are not feasible, a similar process can be implemented online using either a purpose-built iterative process or a variant of the Delphi method (Donohoe et al., Reference Donohoe, Stellefson and Tennant2012).
Crucially, whether a hearing-based or co-creation -based approach is selected, we recommend that no explicit reference to nudges or possible intervention examples should be made at this point. This is to ensure the relevant choices and their architecture are charted, and not only those considered to best fit a preconceived intervention (for example, only listing choice points related to default options). Finally, the selected choices should be assessed for both impact and feasibility. In principle, the target choice points for intervention development should be selected based on both of these, prioritizing impact where possible.
The main risk of co-creation in charting problem structures and choice architectures is that despite a lengthy process, actionable information is not gleaned. This could be either because the choice points that are identified are either inconsequential or too situation-specific to generalize; because choice point dynamics have been misunderstood; or because constrictive group processes limit ideation and insight, leading to lukewarm general information not applicable in practice. Facilitation should ensure the process remains focused enough to yield actionable results. Co-creation can also be complemented by standard evidence-based methods, such as drawing on existing research literature on the determinants of behavior, or by hearing experts.
For example, in the context of sustainable land use, a Finnish study used co-creation to map out barriers to land reforestation among farmers. Generational attachment to fields cleared by their ancestors was identified as a key barrier (Ruuskanen et al., Reference Ruuskanen, Godenhielm, Haltia, Horne, Kekkonen, Kujala, Maidell, Melin, Niemi and Sahari2025). This insight would have been difficult to glean through methods like polling.
Developing behavior change interventions
After the identification of target choice points and the drivers of choice, co-creation can be continued in the intervention development phase.
In introducing intervention development to participants, facilitators can briefly introduce what behavioral interventions are. However, they should take care not to constrict group ideation by providing generic examples (e.g., the fly in the urinal nudge), which bear the risk of constricting creativity to focus on a narrow set of similar solutions. It is also best to avoid sharing examples from the target behavior domain. This process can be structured by considering existing intervention groupings in turn. For example, what, if any, are the default choices for this behavior and how they are presented? Do social comparisons and group relationships produce barriers to the target behavior? Is choice-relevant information available and provided in a fun, accessible and timely manner? What changes in the physical environment would make the behavior more likely? How to make the behavior more salient in the environment?
Once the group has outlined possible intervention options, facilitators can provide more specific information about existing nudges and research literature to help the group refine their suggestions. The group discussion should also review the intervention(s) for possible ethical issues. This is complemented by an expert ethics review when the target domain warrants further scrutiny or when potential issues are flagged.
The risks in this stage include either stifled or unmotivated ideation or stakeholder selection manifesting at this stage as biased solutions that serve specific stakeholders’ aims or do not account for heterogeneity. For example, our colleagues (Salo et al., Reference Salo, Suomalainen, Halonen, Honkatukia, Horne, Karvonen, Koi, Koponen, Lanki, Maidell, Norkio, Ruuskanen, Saastamoinen, Sandman, Siipi, Sinokki, Sivonen and Tuominen2026) studied active commuting among municipal employees. However, in some events, participants who self-selected into the co-creation workshops were cycling enthusiasts with limited insight on barriers to active commuting among those who did not actively commute, leading to a partial view.
Here, a good example is the failure of the World Health Organization’s first response to the 2014 Ebola outbreak in West Africa. Top-down bans on traditional burial rites clashed with the collective significance of those rites, driving burials out of officials’ sight and without safety procedures, thereby exacerbating the outbreak (World Health Organization, 2015; Gostin, Reference Gostin2021). An effective response emerged only in co-creation with the local community, which allowed the community representatives and WHO to jointly develop culturally appropriate interventions.
Validation and pilot studies
Interventions should be validated to ensure their feasibility and to gain preliminary insight into their efficacy. One way to validate an intervention is to carry out a pilot study. For a thorough overview on how to design an effective pilot study, see Pearson et al. (Reference Pearson, Naylor, Ashe, Fernandez, Yoong and Wolfenden2020). The role of the pilot study is not only to evaluate whether the intervention is likely to be effective or whether it can be implemented in practice. These are, of course, essential considerations. One often neglected issue is the use in safeguarding against opaque or uninformative research outcomes (Zarin et al., Reference Zarin, Goodman and Kimmelman2019; Taylor and Kowalkowski, Reference Taylor and Kowalkowski2021).
However, pilot studies do not replace stakeholder involvement. On the contrary, continuous feedback loops help the group fine-tune interventions in real time. During the pilot, short stakeholder meetings help gauge initial impressions, satisfaction and any logistical problems. Ideally, quantitative indicators (e.g., usage rates and sign-ups) are combined with qualitative insights (e.g., focus-group feedback and diaries).
Not all proposed interventions easily yield to experimentation (Ewert et al., Reference Ewert, Loer and Thomann2021). Here, validation of a proposed policy and intervention can also take other forms than that of an experimental design: the community can be an active producer of knowledge in assessing the feasibility, acceptability and cultural relevance of an intervention before implementation. By involving community members, stakeholders and practitioners in co-designing and reviewing the intervention, researchers can validate the proposed policy measure with qualitative insights that may not emerge in a strictly controlled experimental setting.
Risks in the validation phase are familiar from intervention research in general. These include pilot fatigue, where participants may lose interest if the pilot drags on without visible adaptations or results. The question of measurement is also essential: If the target effect is operationalized too narrowly, qualitative changes may remain uncaptured.
Scaling the intervention
Most intervention designs never make it to the scaling stage. However, if the pilot is deemed successful, the next step is broader rollout.
Success at a small scale does not guarantee impact at scale, in part due to variance in choice contexts. However, we emphasize that it is never too late to incorporate co-creation. Co-creation allows for adaptive scaling, where participants from new areas or demographic groups are invited to adapt the intervention to their circumstances before it is fully implemented.
Just as in piloting and validation, multiple methods of gaining insight into intervention effect are available. Here, methods for assessing the effect of the intervention should be chosen based on the intervention, rather than selecting interventions based on a given method. Randomized controlled trials (RCTs)—cornerstones of BPP—have enjoyed paradigmatic status as part of evidence-based policymaking (DellaVigna and Linos, Reference DellaVigna and Linos2022). Regardless, undue insistence on RCT-ready interventions can sometimes ignore more fundamental, system-wide changes that resist easy experimental quantification or that require multi-stakeholder coordination (Ewert et al., Reference Ewert, Loer and Thomann2021; Chater and Loewenstein, Reference Chater and Loewenstein2022). In short, methods of intervention evaluation are chosen in light of available resources and capabilities, as well as on the intervention and the target behavior.
For example, scaling COVID-19 vaccine communications required tailoring nudges to target demographics (Baur et al., Reference Baur, Saperstein, Griffis, Vazquez, Barnes and Fish2025). Baur et al. used iterative co-creation with stakeholder organizations to tailor nudges to the distinct barriers faced by target demographics. For example, vaccine communications for LGBTQ+ community centers addressed specific barriers to vaccination, such as concerns about being misgendered at the vaccine clinic (Baur et al., Reference Baur, Saperstein, Griffis, Vazquez, Barnes and Fish2025). The interventions were assessed with informal feedback, which also helped iterate them further.
A key advantage of co-creation is that it can continue past the official ‘launch’ of an intervention. For this reason, even as the program grows, contact with local stakeholders should be maintained. Here, also, one should confirm that higher-level authorities (city councils, government ministries and hospital boards) are supportive and willing to adapt rules or budgets to accommodate the new approach.
Risks for broader rollout include structural obstacles. For example, larger systems may have strict rules or departmental silos that hinder adoption, requiring explicit strategies to navigate or change bureaucratic inertia. As scale increases, some managers may revert to top-down decisions. This undermines trust and can reduce local stakeholders to passive ‘recipients’ rather than co-creators.
Co-creation can be integrated across the development and scaling of an intervention, or in only part of the process, for example, if feasibility or confidentiality constraints prohibit use of co-creation in certain phases, or if the process has already begun. Throughout the process, co-creation helps center stakeholder agency and integrate predictable behavioral and social mechanisms—such as identity obligations, distrust and incentive structures—that shape feasibility and compliance.
Limitations of co-creation in behavioral policy
Above, we have argued that co-creation allows policymakers to create channels for bottom-up change. But it is not without its limitations.
First, even when co-creation processes are deployed, decisions often ultimately rest with elected officials and professionals. Reviews of co-creation in practice find that initiatives are frequently evaluated on process rather than outcomes, and that outputs often fail to bind subsequent decision-making (Maginn, Reference Maginn2007; Voorberg et al., Reference Voorberg, Bekkers and Tummers2015; Ercan et al., Reference Ercan, Hendriks and Dryzek2019).
Accordingly, our claim is a bounded one: while co-creation moves from safeguarding agency to promoting it, it only rarely reflects shared authority over final decisions. Its value depends on decision-makers committing to uptake in advance with a defined mandate for the process, buy-in to its outputs and earmarked resources. Absent such commitments, the commons literature suggests that the stakes are higher than mere ineffectiveness: just as externally imposed control can crowd out the intrinsic motivations that sustain cooperation (Cardenas et al., Reference Cardenas, Stranlund and Willis2000; Ostrom, Reference Ostrom2000), participation that is solicited and then ignored can become an obstacle to the very trust and sense of recognition that co-creation exists to build.
Second, the most bottom-up form of co-creation, the citizen initiative, is often stifled by bureaucratic siloes. For example, in a citizen initiative, municipal representatives may be motivated to develop and implement a bottom-up suggestion but be prevented from doing so because municipal funds have already been earmarked for other costs. Resources should be earmarked for citizen initiatives, and great care should be taken that these resources are not disproportionately funneled into the projects of the most vocal citizen groups, as citizen initiatives rely on networking and can reflect social privilege (Connelly et al., Reference Connelly, Bryant and Sharp2020; Denters, Reference Denters2023).
Third, co-creation must prove practicable under existing political and technological conditions, including budget and time constraints. Here, as noted above, co-creation can be deployed in selected phases rather than across the whole cycle. Also, co-creation can harness stakeholder organizations and their resources; for an example of multi-stakeholder integration under severe time constraints, see Baur et al. (Reference Baur, Saperstein, Griffis, Vazquez, Barnes and Fish2025).
Despite these risks, the potential returns—in legitimacy, adaptability and citizen-driven innovation—make a strong case for pursuing co-creation. One strategy to mitigate risk is a dual-track approach, in which a top-down plan proceeds in parallel with a co-creation track. Where the co-creation process uncovers alternative solutions or improvements, these can be integrated into subsequent policy cycles.
From an ethical standpoint, while citizen agency and bottom-up engagement are ethically desirable, all policy development is subject to risk assessment and balancing ethical considerations. As a result, co-creation is not always the ideal choice, and in some cases, it can be ethically better not to rely on co-creation. Consider, for example, immunization programs during COVID-19, where public health professionals needed to ensure adequately many get vaccinated as rapidly as feasible. The pandemic raised a range of policy questions, only part of which were apt targets of co-creation. Whether a vaccination program should take place was clearly a question where policy should rely on experts. But co-creation still occupied vital roles. During the COVID-19 pandemic, one of the drivers of vaccine hesitancy and refusal was distrust of top-down policy, distrust that may be reflective of a desire to be recognized as an agent or to exercise one’s own agency by refusing to comply (Butler and Sorell, Reference Butler and Sorell2022; Kärki, Reference Kärki2022). Co-creation can also be well suited for addressing part of the problem cluster, even if it would not be the best approach tout court. During COVID-19, co-creation approaches helped address a range of issues, such as how to respond to misinformation in specific patient populations (Lum et al., Reference Lum, Fischer, Ytell, Scherer, O’leary, Elk, Hurley, Washington and DeCamp2024) and how to tailor nudges to target demographics (Baur et al., Reference Baur, Saperstein, Griffis, Vazquez, Barnes and Fish2025).
Even as co-creation and co-design have gained a foothold, multiple knowledge gaps on them remain. We need further research on co-creation as a methodology. To ensure that co-creation processes are inclusive of marginalized voices—especially those lacking resources, time or cultural capital to participate—future research should explore methods for broad-based recruitment, the extrinsic and intrinsic incentives for sustained engagement, and further development of facilitation techniques that mitigate status differences within groups. More insight is also needed on how co-creation can be evaluated and assessed for impact in both the short and long term. Here, we need ways to develop rigorous expert assessment, as well as research on stakeholder perspectives. Most case studies are told from the perspective of practitioners and researchers, and it is practitioners and researchers who decide whether a co-creation process has been successful. As a result, we do not have sufficient empirical evidence on whether co-creation in fact makes participants feel their agency is recognized and respected, or whether stakeholder perspectives on whether the process was successful match those of the facilitators. Furthermore, rigorous comparative evidence on the cost-effectiveness of co-created versus top-down behavioral interventions does not yet exist, and producing it should be a priority for further research.
Finally, making a case for co-creation may seem naïve in a time where decision-making power is accumulating amongst the few rather than the many, and where anti-democratic attitudes that critics label as ‘technofascism’ or, more moderately, as ‘the new authoritarianism’ (Coeckelbergh, Reference Coeckelbergh2026) are driven in part by the AI transition.
The technocratic transition comes with challenges to legitimacy and institutional trust (Kreps and Kriner, Reference Kreps and Kriner2023; Coeckelbergh, Reference Coeckelbergh2024; Robles and Mallinson, Reference Robles and Mallinson2025). This makes this also a pivotal time for BPP, which relies heavily on public acceptance and psychological compliance. As noted above, top-down policy risks reactance even without AI input. Shifting governance to human-AI ensembles, as some have suggested (Keppeler et al., Reference Keppeler, Borchert, Pedersen and Lehmann Nielsen2025), may exacerbate reactance and erode trust. In light of this, fostering and developing genuinely bottom-up forms of democratic participation and co-creation is not just ethically desirable, but also of practical value for governance.
Conclusions
Above, we have outlined how co-creation can be deployed to move from safeguarding individual agency from constraints such as bans and regulations, to actively facilitating collective agency. This yields a more collaborative, effective and ethical way to design behavioral change interventions. Our proposal is in line with and expands on both the nudge+ approach as proposed by John and colleagues (John et al., Reference John, Smith and Stoker2009; Banerjee and John, Reference Banerjee and John2024), as well as with Evans and Terrey’s (Evans and Terrey, Reference Evans, Terrey, Stoker and Evans2016) view of co-designing with citizens and other stakeholders. Furthermore, a bottom-up approach can be instrumental in moving us beyond the i-frame toward a more system-aware s-frame (Chater and Loewenstein, Reference Chater and Loewenstein2022).
Top-down nudges and boosts have shown promise in many areas, yet an exclusive focus on centralized solutions risks ethical pushback, reduced local ownership and insufficient attention to underlying systemic factors. Co-creation offers a complementary and potentially transformative path, empowering communities to engage with policymakers, researchers and practitioners to design interventions that better reflect local contexts and produce enduring behavior change. Embedding co-creation throughout the intervention development process balances rigor with responsiveness, and—done well—can generate innovations that no policymaker or practitioner could have devised alone.
Co-creation thus takes bottom-up BPP two steps forward. It allows citizens and stakeholder groups to exert agency over intervention development, anchoring bottom-up policy to the multifaceted nature of human agency. And it allows bottom-up dynamics to permeate the entire lifecycle of an intervention, from mapping desired aims to intervention design, piloting and scaling.
Even as co-creation makes policy development a matter of collective agency, its targets still often are individual choices. More research is needed on how co-creation could help refocus BPP to collective agency and better collective choices beyond the co-creation process itself.
While co-creation presents an agency-first way forward for BPP, bottom-up approaches are not a panacea. To make the most of co-creation, we must confront its inherent challenges. Nonetheless, the payoff can be far-reaching: stronger trust, heightened legitimacy and interventions that center collective agency in tackling the critical social and environmental problems of our time.
Acknowledgements
The authors thank Olli-Pekka Ruuskanen, Paula Salo, Jani Sinokki, Helena Siipi, the research groups of the TUIMA and Climate-Nudge Projects and Sanchayan Banerjee.
Funding statement
This work was funded by the Academy of Finland, Strategic Research Council (#335186).
Competing interests
The authors declare none.
Open access
