Amy Rodger

Video games reach billions of people worldwide and may offer a scalable opportunity to engage players in climate and environmental action. However, little empirical research has assessed whether video games can influence the pro-environmental behavioural determinants or actions they aim to promote. To address this gap, we evaluated a 21-day ‘green activation’ embedded in Love & Pies, a popular mobile game participating in the 2024 Green Game Jam. The activation encouraged players to adopt pro-environmental behaviours in food consumption, waste management, and habitat restoration through techniques like educational messaging and stories from activists and players. Using a pre–post survey design ( N = 33,910), we assessed players’ self-reported pro-environmental behaviours and key behavioural determinants (e.g. knowledge), linking responses to objective gameplay data. At baseline, players reported moderate levels of subjective knowledge, perceived capability, and outcome efficacy across all three environmental issues. However, one in five players showed low levels of capability and knowledge, suggesting potential targets for intervention. Engagement with the activation, measured by milestones completed, was associated with small increases in subjective knowledge and, to a lesser extent, perceived capability. No consistent associations were observed for other determinants or self-reported behaviours. These findings align with prior research suggesting that games may more readily influence behavioural determinants than actual behaviours. However, video games may be a useful tool for engaging broad audiences with climate issues, particularly when grounded in behavioural theory and designed for repeated exposure to messaging. This study also highlights several design and evaluation challenges in commercial studio environments, as well as future opportunities for collaborative research at the intersection of behavioural science, climate communication, and interactive media.

Generative artificial intelligence (AI) is increasingly used across the behavioural research lifecycle. It may improve the efficiency and quality of research, but rapid, uncritical adoption carries risks to research integrity: algorithmic bias, ethical concerns around consent, privacy and plagiarism, erosion of researchers’ skills and public trust, and environmental costs. This Open Letter presents Behavioural Research UK’s (BR-UK) statement on using generative AI effectively and responsibly in behavioural research. It was informed by existing guidance and expert consultation. The statement sets out BR-UK’s four key propositions: (1) research ethics and integrity principles must include AI concerns; (2) AI should assist, but not replace, researchers through a human-in-the-loop approach; (3) rigorous evaluation of AI use is essential; and (4) coordinated stakeholder action is needed to move from identifying risks to practical solutions. It then provides recommendations and points to practical tools, such as decision aids and checklists for declaring AI use, for seven stakeholder groups, namely research-generating institutions, funders, publication outlets, behavioural research networks and communities, AI platform and tool providers, individual researchers, and research users. The statement will be reviewed annually until 2028, the duration of BR-UK, as technology and evidence evolve.

Behavioural research is a multidisciplinary field essential for addressing societal challenges, but it suffers from inconsistent terminology that impedes knowledge synthesis and communication across disciplines. This study used consensus methods to develop informal and ontological definitions for seven key behavioural research terms: individual behaviour, population behaviour, individual behaviour change, behavioural theory, behavioural research, behavioural science, and leadership. Using a seven-step iterative method involving a literature review, expert consultations, and feedback from diverse behavioural researchers and users, the team refined definitions to enhance clarity, inclusivity, and cross-disciplinary understanding. The resulting definitions were accompanied by elaborations and examples to provide context and concrete illustrations of the terms. These shared definitions provide a standardised vocabulary to improve study and intervention design and reporting, evidence synthesis, and translation of evidence into policy and practice. While acknowledging limitations such as sample bias, this work lays a foundation for clearer communication and future advances in behavioural research.

Artificial intelligence systems increasingly shape how people think, feel, and act, particularly in high-impact domains such as healthcare, education, and social support. While significant attention has been paid to technical performance, bias, privacy, and security, the behavioural impacts of AI systems remain under-evaluated and under-governed. This Open Letter outlines why behavioural science should be treated as a core component of responsible AI practice. Even when technically accurate, AI systems can influence behaviour in ways that are harmful, misleading, or misaligned with people’s interests. These effects are often predictable consequences of repeated human–AI interaction, yet they are rarely assessed systematically. We describe where behavioural risks arise in practice and outline what good practice looks like, drawing on established behavioural science methods. We then propose practical steps for funders, researchers, and developers to embed behavioural science expertise and behavioural evaluation across the AI lifecycle, including behavioural testing pipelines, longitudinal evaluation, and interdisciplinary governance. Addressing behavioural impacts is essential to ensuring that AI systems are not only effective, but safe, trustworthy, and appropriate for real-world use.

Public support is crucial for the success of policy interventions that aim to change behaviour. While communicating evidence of policy effectiveness can increase support, it remains unclear which type of evidence is most effective. Statistical evidence is often seen as objective and persuasive, yet personal anecdotes can strongly influence beliefs. We examined how statistical and anecdotal evidence affect policy perceptions. In three online experiments with representative UK samples (N = 901), we showed participants different types of evidence (statistical, anecdotal, or both) that argued for or against six policies, such as meat taxes (climate change), banning e-cigarettes (public health), and 20 mph speed limits (community safety). We measured policy support and perceived effectiveness before and after exposure and explored participants’ reasoning through open-text responses. Results showed that positive statistical and anecdotal evidence did not significantly increase perceived policy effectiveness or support, even when combined. However, negative anecdotes significantly reduced both, though this effect was sometimes mitigated when paired with statistical evidence. Qualitative results found that participants have broader concerns beyond policy effectiveness, such as fairness. Our findings suggest that communicating evidence on policy effectiveness alone may not increase support, as it does not address broader public concerns.

Artificial intelligence systems increasingly shape how people think, feel, and act, particularly in high-impact domains such as healthcare, education, and social support. While significant attention has been paid to technical performance, bias, privacy, and security, the behavioural impacts of AI systems remain under-evaluated and under-governed. This Open Letter outlines why behavioural science should be treated as a core component of responsible AI practice. Even when technically accurate, AI systems can influence behaviour in ways that are harmful, misleading, or misaligned with people’s interests. These effects are often predictable consequences of repeated human–AI interaction, yet they are rarely assessed systematically. We describe where behavioural risks arise in practice and outline what good practice looks like, drawing on established behavioural science methods. We then propose practical steps for funders, researchers, and developers to embed behavioural expertise and behavioural evaluation across the AI lifecycle. Addressing behavioural impacts is essential to ensuring that AI systems are not only effective, but safe, trustworthy, and appropriate for real-world use.

Public support is crucial for the success of policy interventions that aim to change behaviour. While communicating evidence of policy effectiveness can increase support, it remains unclear which type of evidence is most effective. Statistical evidence is often seen as objective and persuasive, yet personal anecdotes can strongly influence beliefs. We examined how statistical and anecdotal evidence affect policy perceptions. In three online experiments with representative UK samples ( N = 908), we showed participants different types of evidence (statistical, anecdotal, or both) that argued for or against six policies, such as meat taxes (climate change), banning e-cigarettes (public health), and 20 mph speed limits (community safety). We measured policy support and perceived effectiveness before and after exposure and explored participants’ reasoning through open-text responses. Results showed that positive statistical and anecdotal evidence did not significantly increase perceived policy effectiveness or support, even when combined. However, negative anecdotes significantly reduced both, though this effect was sometimes mitigated when paired with statistical evidence. Qualitative results found that participants have broader concerns beyond policy effectiveness, such as fairness. Our findings suggest that communicating evidence on policy effectiveness alone may not increase support, as it does not address broader public concerns.

Energy, transport, and food sectors contribute substantially to climate change and are key areas where behaviour change can have the greatest impact. International policy learning and transfer are central to government approaches in these sectors. However, policies often show limited effectiveness when applied in varied contexts, likely due to assumptions of population homogeneity and over-reliance on socio-demographic characteristics. Using Multilevel Latent Class Analyses (MLCA) on over 17,000 respondents across 61 regions in nine OECD countries, we found that the majority fell into population segments with the least sustainable behaviours across the three sectors, and that socio-demographic characteristics failed to predict segment membership consistently. Individuals engaging in sustainable behaviours in one sector were more likely to do so in others, and unsustainable behaviours had similar positive associations across sectors, suggesting coordinated cross-sector interventions are crucial. Regional distribution of population segments typically grouped geographically by country, underscoring the importance of contextual influences. However, we identified significant within-country variations in regional distributions, with some regions showing substantially different proportions of people in sustainable segments than neighbouring regions under identical national policies, showing the importance of regional context. Moving beyond traditional ad-hoc policy transfer approaches, we present a systematic framework for identifying evidence-based transfer opportunities, showing policymakers not only who needs interventions but where successful patterns emerge and which contexts appear to enable sustainable behaviours. These findings provide actionable insights for the international policy community seeking to accelerate behavioural change at the scale and speed required for climate mitigation.

Background The Behaviour Change Intervention Ontology (BCIO) can be used to annotate reports of intervention evaluations for evidence synthesis. Such annotation requires a framework for entering and storing data (a ‘data schema’). This paper assessed the feasibility of using a prototype schema in a systematic review aimed at identifying promising behaviour change techniques (BCTS) to include in interventions designed to reduce speeding by motorists. Methods A systematic search identified randomised trials on interventions to reduce speeding through drivers’ ability and motivation. A prototype schema was created to annotate these papers, capturing information about Behaviour Change Techniques (BCTs), outcomes, and other study features. A team of annotators worked in pairs, independently annotating study reports before discussing their annotations to reach a consensus. The annotators noted issues concerning schema feasibility and identified gaps in the BCIO and possible associations between BCTs and outcomes. Results A total of twenty-nine eligible publications, encompassing 30 studies, were annotated, resulting in 88 study conditions and 2,097 individual annotations. The provided schema established a useful framework for categorising study features, although some inconsistencies and errors were noted in the annotations produced. Sixty BCIO classes effectively represented all but three of the annotated BCTS, yet none of the 70 behavioural outcomes were linked to ontology classes. The most frequently cited BCT in interventions was the provision of information regarding the adverse health effects of speeding, followed closely by BCTs offering feedback about speeding. The BCTs that were annotated as effective more often than ineffective involved devices and signage that delivered immediate feedback about speeding. Discussion A prototype schema for annotating behaviour-change intervention evaluations using BCIO could synthesise evidence on interventions to reduce speeding but needs a user interface for real-time validation. BCTs that raise immediate awareness of speeding are promising candidates for future interventions.

Commercial video games reach billions of people worldwide and may offer a scalable opportunity to engage players in climate and environmental action. However, little empirical research has assessed whether these games can influence the pro-environmental behavioural determinants or actions they aim to promote. To address this gap, we evaluated a 21-day “green activation” embedded in Love & Pies, a popular mobile game participating in the 2024 Green Game Jam. The activation encouraged players to adopt pro-environmental behaviours in food consumption, waste management, and habitat restoration through techniques like educational messaging and stories from activists and players. Using a pre–post survey design (N = 33,910), we assessed players’ self-reported pro-environmental behaviours and key behavioural determinants (e.g., knowledge), linking responses to objective gameplay data. At baseline, players reported moderate levels of subjective knowledge, perceived capability, and outcome efficacy across all three environmental issues. However, one in five players showed low levels of capability and knowledge, suggesting potential targets for intervention. Engagement with the activation, measured by milestones completed, was associated with small increases in subjective knowledge and, to a lesser extent, perceived capability. No consistent associations were observed for other determinants or self-reported behaviours. These findings align with prior research suggesting that games may more readily influence behavioural determinants than actual behaviours. However, commercial games may be a useful tool for engaging broad audiences with climate issues, particularly when grounded in behavioural theory and designed for repeated exposure to messaging. This study also highlights several design and evaluation challenges in commercial game environments, as well as future opportunities for collaborative research at the intersection of behavioural science, climate communication, and interactive media.