The influence of uncertainty: How a psychologist is improving disaster decision-making

In an increasingly uncertain world, resilience depends not only on the hazards we face, but on how we understand and respond to them. From extreme weather warnings to public health crises, individuals are constantly required to make decisions based on incomplete or probabilistic information—yet these judgments are often shaped as much by perception and communication as by the risks themselves.

This challenge sits at the heart of the work of Dr Susan Joslyn, a cognitive psychologist at the University of Washington whose research focuses on how people interpret uncertainty and make decisions under risk.

Bridging psychology, climate science, and risk communication, Susan's work offers critical insights for resilience practice: improving how uncertainty is conveyed can directly influence how individuals and communities prepare for and respond to crises. In this instalment in the Careers in Resilience series, PreventionWeb interviewed Susan, reflecting on her career and exploring what it takes to support better decision-making in the face of uncertainty.

What drew you to focusing on decision-making based on weather forecasts within the field of psychology?  

Unlike many who work in this area, I began my career as a traditional academic, cognitive-behavioural scientist. I learned to do careful, well-controlled experiments, randomly assigning participants to control and experimental conditions to ask theory-driven questions about how the mind works - what is referred to as “cognitive architecture”.

But then, I became interested in how these principles applied to real world situations. I joined a multi-disciplinary team seeking to understand how to effectively communicate weather forecast uncertainty information (30% chance of a tornado) to emergency managers (EMs) and eventually, to members of the public.

As the psychologist on the team, I began by searching the psychological literature for previous experimental work on this topic. However, at the time, except for the well-known work in behavioural economics showing that most people do not treat probability in an economically optimal way, there was none. I was looking for evidence addressing whether probabilistic information helped--at all. In other words, do probabilistic forecasts improve decisions compared to traditional forecasts that provide a single value (Daytime high of 70°F, Rain, Tornado), even though the decisions based on probabilistic forecasts might not be “optimal”?

Your work on the Decision Making with Uncertainty Lab focuses on the understanding of complex scientific information by non-experts. Could you share some of the findings on this subject?

To fill that gap in the literature, my lab and I began testing the impact of probabilistic forecasts on decisions. We assigned participants realistic decisions (e.g. whether to issue a freeze warning) and provided them real archived forecasts and outcomes—we informed participants of the outcome after each decision. All participants received the same basic forecast, but half received a single value version (e.g. 33°F) while the other half also received an uncertainty estimate (e.g. 40% chance of temperatures < 32°F).  

We found that, compared to the single value forecast, numeric uncertainty information does improve decisions to take protective action. Moreover, the careful experimental design allowed us to make a causal inference: As everything else about the decision was the same for all participants and random assignment ensured that abilities and prior knowledge of those in control and experimental groups were similar, we could conclude that probabilistic forecasts caused the improvement in decisions.

Importantly, participants had greater trust in the forecast when uncertainty was acknowledged. A study we did in the US showed that most people know that forecasts involve uncertainty—a forecast that acknowledges uncertainty therefore seems more honest. In addition, we found that level of education did not matter—one did not have to understand probability theory to benefit from probabilistic forecasts.

Are there any risk communication techniques that you have identified that increase the understanding of complex disaster risk information?

We learned that, in designing communication to increase understanding, it is important to first know what information is needed by decision-makers and how it will be used to make key decisions. For instance, uncertainty information is much easier to understand if it can be applied directly to the decision context. If an EM needs to post a high wind warning when windspeeds will exceed 30 mph, the forecast informing that decision should say “30% chance of wind speeds greater than 30 mph”. The opposite phrase “70% chance of wind speeds less than 30 mph,” despite meaning the same thing, is confusing because it does not match EMs “greater than” concern. To get it to match, the EM must subtract 70 from 100 and change “less than” to “greater than”. These transformations are time consuming and increase the mental workload as well as the chance of errors (e.g. skipping the subtraction) that lead to misunderstanding.

It seems obvious, but you would be surprised at how often this principle is violated. Similarly, EMs concerned with posting a wind warning care less about the movement of the low-pressure system (a major concern of the forecasters producing the forecast) than they do about when, where and to what degree wind speeds will increase. These are just a few examples—the general principle has a much broader application:  first understand the user’s decision, then provide information that can be applied directly to that decision.

How would you say that decision making under uncertainty can be improved, especially considering the impact of mis/disinformation?

In addition to understanding the information needs of the recipient, in many cases it is important to also know how they understand the situation—referred to as their “mental model”.

Mis/disinformation is often the result of capitalizing, intentionally or accidentally, on cognitive short cuts that are used to streamline the process of building the mental model. For instance, people tend to think that global average temperature will decline as soon as greenhouse gas emissions are reduced—they assume the patterns will match (e.g. both decline). In fact, global average temperature will continue to rise for decades, even centuries (depending on the rate of reduction, etc.), after we start to reduce greenhouse gas emissions. 

The mismatch in the patterns of change in these two processes (emissions decline/temperatures rise) is very difficult for people to conceptualize.  Unless they are told otherwise, most people assume a match. We have shown that describing this mismatch directly improves people’s understanding and can engender a sense of urgency to address the issue.  The good news is that cognitive psychologists have been documenting these common mental short cuts, such as pattern matching, for decades so we know when and where they are likely to crop up. Communication strategies can be designed to block them.   

Sometimes, as we’ve shown with respect to misunderstandings about the mRNA vaccines during the COVID pandemic, it helps to refute misunderstandings directly, e.g. “Some people think the vaccine contains the COVID19 virus, it does not…”. When this direct refutation was followed by an explanation of how the vaccine actually works, both misunderstanding and reluctance to get vaccinated were reduced.

How would you say that your research can inform the work of civil protection and emergency management professionals?

Our research can inform the development of effective communication strategies. Providing information that can be applied directly to the decision at hand, addresses how people think about the situation and the common misunderstandings, is key. Using this approach can make it possible for members of the public to understand more complex scientific information than that for which they are generally given credit.

Not only will better-informed users have greater trust in the information provided, they can also make decisions tailored to their own risk tolerance and are less likely to retain harmful misunderstandings. This line of research suggests that less is not always more—sometimes over simplified messaging opens the door to misunderstandings and eventually to mistrust.

What advice would you give to young professionals interested in careers at the intersection of psychology and disaster risk reduction?  

There are so many interesting and important questions that we have yet to tackle in this field. It takes researchers from many different approaches and backgrounds to conduct work that will allow us to fully understand these complex and critical issues.

Always start with the needs of the user, test new strategies carefully before implementing them and don’t be afraid to try something new! 

Dr Susan Joslyn is Professor Emeritus at the University of Washington, working within the Department of Psychology. Her interest lie in decision making and communicating uncertainty information, and works in the 'Decision making with uncertainty lab', a cognitive-behavioral laboratory conducting survey and experimental research to answer applied questions about decision-making under uncertainty and the communication strategies that best support those decisions.