Learning
Much of what is known about decisions under risk comes from studies in which risks are described—as numbers, graphs, or warnings. But in everyday life, people often learn about risks through personal experience. They make a choice—like investing in a new cryptocurrency or skipping a recommended vaccine—and then see what happens, whether the outcome is good or bad. Studies show that these two ways of learning—being told about risks versus learning from experience—can lead to very different choices. Our research on learning investigates how experience shapes decision making and asks how people mentally store and sometimes avoid information about risks and uncertain outcomes. We study how learned information is represented in memory, when and why people choose deliberate ignorance instead of seeking more knowledge, and how risk can be communicated more effectively through simulated experience rather than written descriptions.
Learning and the Description–Experience Gap
People tend to choose differently depending on whether they are told about risks or learn about the same risks through direct experience—by repeatedly making choices and seeing what happens. This systematic discrepancy, known as the description–experience gap, is one of the most robust findings in behavioral decision research.
In our laboratory studies, we investigate decisions from experience using the “sampling paradigm.” Participants see two buttons or options on a computer screen, but have no information about what will happen if they choose either one. To learn about these options, participants can “sample” them: They click on one option, and the computer immediately shows a result (e.g., win 0 or 32 points). They can click on each option as many times as they like, in any order. Each click on an option produces one result, drawn at random from the set of possible results for that option. When they have finished exploring, they make a real choice by selecting one of the options to actually play, and this final choice determines what they earn.
Our work shows that the description–experience gap is driven in part by insufficient exploration: People learning from experience often do not sample for long enough to encounter rare outcomes. As a result, they underestimate the importance of these outcomes when making decisions. Yet even when people sample extensively, differences between decisions from description and experience persist, suggesting that the two are fundamentally different teachers of risk.
Building on this work, we have proposed a broader conceptual framework for understanding how description and experience interact in shaping responses to real-world risks—from pandemics and climate change to drug use and financial decisions.
Mapping Memory and Public Understandings
A lifetime of learning shapes how memory organizes facts, concepts, and associations. In our research, we use free word associations to map the structure of semantic memory as a network, revealing how concepts are connected, which associations are strongest, and where individual differences emerge.
When asked what comes to mind when they think of “risk,” for example, people produce a rich web of associations that cluster into distinct themes—from danger and thrill-seeking to worry, deliberation, and reward. In the network shown here, each dot represents a word that people associated with risk, with larger dots indicating more central associations and colors marking clusters of related words. This work shows that “risk” is not a single concept but a multifaceted one, meaning different things to different people, and varying systematically with factors such as age and language.
We build open-source tools to make these methods accessible. For example, we have developed software for collecting, processing, and analyzing free association data using techniques from natural language processing and large language models.
The Psychology of Deliberate Ignorance
Knowing more is often seen as the key to better decisions, but people sometimes actively choose not to know. A patient may decline genetic testing; a citizen may decide not to read their secret police file. Our research asks when and why people opt for what we call “deliberate ignorance,” and how this shapes decision making in domains ranging from medical decision making to political and civic choices.
After the archives of the East German state security service (Stasi) were opened, many former citizens chose not to read their files—not out of indifference, but for reasons that included protecting relationships, managing emotions, and distancing themselves from the former surveillance regime. Ralph Hertwig and Dagmar Ellerbrock discuss their research on this phenomenon in an episode of the Bundesarchiv's podcast 111 Kilometer Akten (listen here; episode in German).
A comprehensive review of research of health decisions found that nearly one in three people avoid medical information. This avoidance is driven less by who people are—for example, their age, gender, or income—and more by factors such as mistrust, feeling overwhelmed, and fear of judgment. In other contexts, deliberate ignorance can also serve as a principled strategy—for instance, by shielding hiring committees or judges from information that could bias them in order to promote fairness.
Communicating Risks Through Simulated Experience
Most risk communication focuses on a narrow slice of information: the probability of possible outcomes. But probability is only one dimension of risk. How variable are the outcomes over time? Under what conditions do they occur? And how does it feel to experience them?
To capture this broader picture, we have developed a taxonomy of risk information dimensions. The table below illustrates these dimensions for smoking, contrasting the familiar focus on the possibility and likelihood of two outcomes—stress relief and lung cancer—with several other dimensions that are typically neglected.
We propose that risk communication should address these often-overlooked dimensions, and that one powerful way to do so is through simulated experience: interactive tools that let people “live through” the consequences of a risk before making a decision.
In a clinical trial with chronic pain patients, for example, those who engaged with an interactive simulation of opioid risks were more likely to reduce or stop their opioid use and to take up alternative therapies than were those who received standard descriptive information.
These ideas connect to broader questions about how people learn about risks to democracy. Citizens who have experienced decades of stability may have little personal experience of democratic breakdown and thus underestimate threats to democratic institutions. A similar pattern appears in vaccine hesitancy: When a disease is rare because vaccines have worked, people may feel safe and think prevention is unnecessary. Ralph Hertwig and Stephan Lewandowsky explore this “success paradox” in more detail in their article in The Conversation: The paradox of democracy’s success.
References
- Aeschbach, S., Mata, R., & Wulff, D. U. (2025). Mapping mental representations with free associations: A tutorial using the R package associator. Journal of Cognition, 8(1), 3. https://doi.org/10.5334/joc.407
- Ciranka, S., & Hertwig, R. (2023). Environmental statistics and experience shape risk-taking across adolescence. Trends in Cognitive Sciences, 27(12), 1123–1134. https://doi.org/10.1016/j.tics.2023.08.020
- Hertwig, R., & Ellerbrock, D. (2022). Why people choose deliberate ignorance in times of societal transformation. Cognition, 229, 105247. https://doi.org/10.1016/j.cognition.2022.105247
- Hertwig, R., Herzog, S. M., & Kozyreva, A. (2024). Blinding to circumvent human biases: Deliberate ignorance in humans, institutions, and machines. Perspectives on Psychological Science, 19(5), 849–859. https://doi.org/10.1177/17456916231188052
- Hertwig, R., & Lewandowsky, S. (2025). How citizens’ experience of democracy can actually pave the way to democratic backsliding. Communications Psychology, 3(1), 144. https://doi.org/10.1038/s44271-025-00327-1
- Hertwig, R., & Wulff, D. U. (2022). A description–experience framework of the psychology of risk. Perspectives on Psychological Science, 17(3), 631–651. https://doi.org/10.1177/17456916211026896
- Lejarraga, T., Ranganathan, K., & Wulff, D. U. (2024). Can simulated experience be harnessed to help people make investment decisions? Judgment and Decision Making, 19, e24. https://doi.org/10.1017/jdm.2024.17
- Lewandowsky, S., & Hertwig, R. (2025). Critical ignoring when information abundance is detrimental to democracy. Current Opinion in Psychology, 66, 102128. https://doi.org/10.1016/j.copsyc.2025.102128
- Offer, K., Oglanova, N., Oswald, L., & Hertwig, R. (2025). Prevalence and predictors of medical information avoidance: A systematic review and meta-analysis. Annals of Behavioral Medicine, 59(1), kaaf058. https://doi.org/10.1093/abm/kaaf058
- Schulte-Mecklenbeck, M., Wagner, G. G., & Hertwig, R. (2024). How personal experiences shaped risk judgments during COVID-19. Journal of Risk Research, 27(3), 438–457. https://doi.org/10.1080/13669877.2024.2350712
- Spektor, M. S., & Wulff, D. U. (2024). Predecisional information search adaptively reduces three types of uncertainty. Proceedings of the National Academy of Sciences, 121(47), e2311714121. https://doi.org/10.1073/pnas.2311714121
- Tiede, K. E., Hertwig, R., Mata, R., & Wulff, D. U. (2025). Communicating risks more comprehensively using simulated experience. Trends in Cognitive Sciences, 30(3), 200–202. https://doi.org/10.1016/j.tics.2025.12.004
- Wegwarth, O., Ludwig, W. D., Spies, C., Schulte, E., & Hertwig, R. (2022). The role of simulated-experience and descriptive formats on perceiving risks of strong opioids: A randomized controlled trial with chronic noncancer pain patients. Patient Education and Counseling, 105(6), 1571–1580. https://doi.org/10.1016/j.pec.2021.10.002
- Wulff, D. U., De Deyne, S., Jones, M. N., & Mata, R. (2019). New perspectives on the aging lexicon. Trends in Cognitive Sciences, 23(8), 686–698. https://doi.org/10.1016/j.tics.2019.05.003
- Wulff, D. U., & Mata, R. (2022). On the semantic representation of risk. Science Advances, 8(27), eabm1883. https://doi.org/10.1126/sciadv.abm1883
- Wulff, D. U., Mergenthaler-Canseco, M., & Hertwig, R. (2018). A meta-analytic review of two modes of learning and the description–experience gap. Psychological Bulletin, 144(2), 140–176. https://doi.org/10.1037/bul0000115