Correcting misconceptions and shaping preferences about energy with reinforcement learning
Speaker: Stefano Palminteri, École Normale Supérieure Paris
2025/12/10 15:20-17:00
Location: Building S1|15 Room 133
Abstract:
Addressing the climate crisis requires citizens to make informed choices about sustainable energy policy, yet public debates are often shaped by inaccurate beliefs about the environmental and health impacts of different energy sources. Accurate knowledge of variables such as greenhouse gas emissions, mortality rates, and energy production is essential for rational decision-making, but these beliefs are frequently flawed and resistant to change. Here we investigate whether a brief, interactive, reinforcement-based intervention can correct such misconceptions and influence societal investment preferences. Across seven countries (UK, US, France, Spain, Italy, Germany and Japan), we show that participants held systematic misbeliefs – for instance overestimating the emissions and underestimating the safety of nuclear power- but that our intervention corrected these beliefs accuracy across all features. Furthermore, correcting beliefs about CO₂ emissions and mortality rates led to shifts in preferences towards objectively cleaner and safer energy mixes, and these changes were consistent across countries. Follow-up assessments indicated that the effects of our intervention persisted for at least x weeks. These results demonstrate an actionable pathway from knowledge correction to preferences in the environmental domain, and suggest that short, reinforcement-based interventions could be a scalable tool for improving public decision-making in climate policy and other domains where misconceptions pollute the debate.
