Communicating uncertainty to lay audiences is as challenging as indispensible if people are to understand medicaltest results, gains from financial investments, or weather warnings.Compared to risk communication in the medical domain, there is so far only limited evidence on how to best communicateuncertainty for continuous quantities, such as financial returns or wind speeds (Spiegelhalter et al., 2011).The poster presents results from a longitudinal study investigating this question within a real-life setting. We implementeddifferent representations communicating probabilistic weather forecasts within an online information system operated by theGerman National Weather Service. The system is used by fire brigade coordination centers throughout Germany to prepare forsevere weather conditions.By analyzing web usage and search behavior, we investigate which representations users rely upon under real operationalconstraints. We link the analysis to tests which representations are best understood and could thus aid emergency managers intheir decisions.

People face consequential personal decisions throughout their lives. Immigrating to another country or separating froma life-partner are but two examples. How do individuals make such notoriously difficult decisions? Can they make themrationally? We posit that answering both these questions requires understanding a decisions transformative potential, ac-cording to which decisions range in (1) their perceived temporal impact (half-life), (2) the extent to which the decisionmaker can know whether a choice will generally make them better or worse off (valence uncertainty), and (3) the perceivedlikelihood of a decision to change the decision maker (personal change). We propose that under the conditions of incom-plete information that decisions with high transformative potential inevitably entail, people may make them by recruitingtheir social and cultural environment and by relying on heuristics. These conditions also render bounded rationality prin-ciples (e.g., satisficing) a more plausible rationality benchmark than maximizing expected utilities.

The key insight in this work is that the events people associate with risk is under systematic change over the past200 years. Leveraging Latent Dirichlet Allocation (Topics Modeling) and the Google Ngram Corpus, we identified historicaland newly-emerging events associated with risk and tracked their relevance over time. We also computed the probability ofrisk co-occurring with words associated with those identified events to capture a more accurate trend. Several highlights ofthe findings include: attention on risk has been spreading from one general domain (about losing life and war/battle) to a setof wider, more specific events and activities such as cancer, sex, HIV, smoke, and finance; in addition, the concept of risk hasrecently become more differentiated, incorporating both cost and benefits, long and short-term consequences. This approachcould be extended to study semantic history of a number of other concepts of interest.

To make inferences about the frequency of events in theworld (e.g., the prevalence of diseases or the popularity ofconsumer products), people often exploit observations ofrelevant instances sampled from their personal social network.How does this ability to infer event frequencies by searchingand relying on personal instance knowledge develop fromchildhood to adulthood? To address this question, weconducted a study in which children (age 8–11 years) andadults (age 19–34 years) judged the relative frequencies offirst names in Germany. Based on the recalled instances of thenames in participants’ social networks, we modeled theirfrequency judgments and the underlying search process with aBayesian hierarchical latent-mixture approach encompassingdifferent computational models. We found developmentaldifferences in the inference strategies that children and adultsused. Whereas the judgments of most adults were bestdescribed by a noncompensatory strategy that assumes limitedand sequentially ordered search (social-circle model), thejudgments of most children were best described by acompensatory strategy that assumes exhaustive search andinformation aggregation (availability-by-recall). Our resultshighlight that already children use instance knowledge to inferevent frequencies but they appear to search more exhaustivelyfor instances than adults. One interpretation of these results isthat the ability to conduct ordered and focused search is acentral aspect in the development of noncompensatoryinstance-based inference.

The emergence of deliberate ignorance, i.e. what influences children's deliberate decisions not to seek information, is a phenomenon so far notably overlooked. This project addresses this gap by investigating various factors that systematically modulate such decisions in children and adults. Across five studies, we presented participants with short stories illustrating social situations where a misdeed occurs, and measured participants' preference for knowing the identity of the wrongdoer. Studies 1-3 (N = 550) shows that both children and adults systematically manifest a preference for ignorance when the information value is low, and when the potential wrongdoer is suspected to be a friend. Studies 4-5 (N = 333) further investigate the role of information probability in this phenomenon. This first contribution shows that children's preference for deliberate ignorance is modulated by information value and the relationship frame proposed, suggesting a rational approach to ignorance.