Testing the future(s): behavioural insights for policymaking

How can policymaking become more rooted in how people think and decide? This was the starting point for the Discover Behavioural Insights for Policymaking workshop at the PolyFutures conference, organised by the EU Policy Lab’s behavioural insights experts Francesca Papa, Michal Krawczyk and Emanuele Ciriolo. 

Set in the iconic machine room at Flagey, the session introduced participants to Behavioural Insights (BI), as a practical approach to understanding decision-making. Drawing on psychology, economics and cognitive sciences, BI challenges the assumption that people respond to information, incentives or rules in fully rational and predictable ways. Instead, it focuses on how context, attention, expectations and social cues shape behaviour in practice. 

The workshop explored the application of this approach to design and foresight. Thinking about the future requires imagination, but also discipline. Behavioural insights can help bridge these needs. They invite policymakers to ask not only “what might people do?” but also “how can we know?” This was one of the central messages of the session: test, don’t guess. 

A core strength of behavioural insights lies in experimental methods. By randomly allocating participants to different treatments, controlled experiments allow researchers to isolate the effect of an intervention. This makes it possible to move from plausible stories to solid evidence. In complex policy contexts, where well-intentioned measures may have no effect, unintended effects, or effects only for certain groups, this distinction is crucial. 

Participants experienced this directly by competing in an experimental forecasting challenge. Using the Bayesian Truth Serum, the facilitators were able to give them incentives to think carefully. Because some of the questions pertained to the distant future, there was no direct way to tell which answers were correct. Yet the method, by also eliciting beliefs about others’ responses, helps identify who is likely to be right: expertise, lies not only in domain knowledge but also in the ability to reason about how others think. 

Many participants were struck by the idea that experimenters may incentivise correct answers even without knowing in advance which answers are correct. This illustrates a more general observation: behavioural scientists can make important contributions even without expert field knowledge. Their contribution lies in methods. They specialise in asking the right questions, eliciting truthful responses from citizens, gathering carefully calibrated judgements from experts. They can identify cases in which knowledge fails to be shared, confidence is misplaced, or a design that looks convincing in theory yields limited insight. 

Perhaps the most valuable moments of the workshop came from the questions raised by participants. These were not only technical questions, but critical and high-level ones:  

How realistic are lab experiments compared with the policy world?  

What can be learned from controlled settings, and where do they fall short?  

How should policymakers treat evidence generated with synthetic populations?  

Could such tools encourage “wishful thinking” if they are used without sufficient caution?  

What can behavioural insights achieve on their own, and where must they be combined with other forms of knowledge? 

These questions were central to the session because they captured the right spirit for using behavioural insights in policy. Behavioural methods are not a shortcut to certainty. They do not replace political judgement, democratic deliberation, domain expertise or institutional knowledge. They are most useful when they make assumptions visible, testable and open to revision. 

The workshop therefore highlighted both the promise and the limits of behavioural insights. Their promise lies in helping policymakers design interventions that are closer to real human behaviour. Their limits remind us that evidence must be interpreted carefully, with attention to context, ethics and implementation. 

A key insight emerging from the discussion was the complementarity between behavioural insights, design and foresight. Together, they form a more coherent approach to navigating uncertainty. 

The example of waste-sorting labels illustrated this interplay. Iterative design improved usability and clarity; experimental testing quantified behavioural impact; and system-level thinking revealed the limits of labels alone.  

The discussion on synthetic populations further underscored the need for caution. While such tools offer new ways to simulate future behaviours, they also carry risks, particularly if their assumptions remain opaque or untested. Their value depends not on their sophistication, but on the transparency and humility with which they are used. 

By the end of the session, the message was clear. Behavioural insights can enrich policymaking when they are used as a disciplined way of learning. They help policymakers move beyond intuition, expose assumptions to evidence and improve designs before they are scaled. In a policy environment shaped by uncertainty, complexity and rapid change, this experimental mindset is a practical resource for thinking more clearly about tomorrow.