A walk through decision making under risk: the influence of experience and emotional outcomes on the perceived chances of a terror attack
Most of us know that the chance of dying from a heart disease or even in a car accident is much higher than in a terror attack. In 2015, there were almost ninety thousand reported deaths between the ages of 15 and 60 in Germany. Around 88% of those were due to health-related reasons, and 12% due to external reasons such as accidents or drug abuse. Heart disease, stroke, lung cancer and blood pressure diseases are the top causes of death, summing up to 40% of all death cases across all ages. Most of the top causes of death are strongly related to risky decisions we make daily, from smoking and eating habits through sports and even driving a car.
Yet, what do people fear? Although given as an option to choose from in different surveys, the fear of being seriously ill didn’t make it into the top 6 fears of people in Germany in 2016. Terror, on the other hand, was on the top of the list with 73% of Germans fearing terror (figure 1) and 64% scared of being victims of a terror attack. Why are we so frightened by something that causes a relative low number of casualties in the western world?
What is it about terror that scares us so much?
A single article could never give a complete answer to this question. Here, I will attempt to address two aspects of terror attacks that may contribute to an over-estimation of their probability: (1) The (fortunate) lack of experience we have with the risk of terror in our decisions and (2) the strong emotional aspect terror brings to the outcomes of our choices.
The indirect damage of Dread risks
A terror attack in the Western world, in which many people are killed or injured at once, is a low-probability event with high damage. These types of events are called Dread Risks. Dread risks seem to have a substantial influence on the decisions we make. For example, plane crashes are Dread risks. Interestingly, the chance in the USA of losing your life driving 12 miles in a car is higher than in a nonstop flight between Boston and Los Angeles. This means that usually, once you arrived at the airport, you finished the most dangerous part of your journey.
The influence of flight avoidance on society took a crucial turn after the 9/11 attacks. In the year to follow, many people chose driving their cars over a short flight. In a fascinating analysis, Gigerenzer showed that during that year 1,500 more people died in fatal car accidents just in an attempt to avoid the perceived danger of flying. Gigerenzer called this indirect damage of Dread risks.
Experience with a certain risk attunes our estimation of it happening
The outcomes of most of the decisions we face daily don’t come with attached labels of probabilities. Another source we can therefore use to make decisions is our own experience. Recent research has focused on “The Description-Experience Gap”, a difference in perceiving the chances of rare events, depending on how they are presented to us.
For many years, decision making research focused on decisions made from descriptions of probabilities. That means, people are shown the probabilities of a set of options and choose which one they prefer (see figure 3, left). When making decisions based on description, people make choices as if they overweigh the chances of rare events.
Recent research focused on the effect experience has on our decision making. In these types of decisions, people are first given the option to experience the choices, without being told about the underlying probabilities (figure 3, learning phase). After experiencing, they can then choose the preferred option (figure 3, choosing phase). When deciding from experience, the over-estimation of rare events seems to be reduced.
There is even some new evidence from non-human primates on the “The Description-Experience Gap”, questioning the necessity of language for the existence of the gap. In a recent study, male rhesus macaques had to choose between a described probability option and a known experienced option. For the description, probabilities were presented by a blue-fade to-red bar, indicating the probability of reward to the presented choice (the monkeys previously learned this method). For the experience option, they used choices that their probabilities were previously learned through a two-week learning period. The monkeys were more risk-seeking for experienced gambles than for described ones. Though much more research is needed for a full understanding of this gap, the authors claim for evidence that the effect of experience on our assessment might not be purely language dependent.
Experience with terror attacks lowers risk aversive behavior
Israel is a good example of a western society that (sadly) has a broad experience with terror. Between the years 2000 and 2003, there were fatal terror attacks in Israel at least once a month. At first, both foreign and domestic tourism suffered. However, the decrease was larger for foreign tourism (-60%) compared to domestic tourism (-10%). In the coming months, this difference increased even further: foreign tourism continued to drop while domestic tourism increased to a level even higher than before the attacks started.
A group of researchers noticed this pattern and decided to explore it in a controlled laboratory setting, using monetary decisions (figure 4). Participants had to choose between two options. Each option contained a rare outcome and a likely one. One option was safe, where both outcomes (rare and likely) had low-medium losses. The second option was risky, where the likely outcome had a low loss but the rare option had a very big loss.
In the description group, the players chose in advance how many times (out of 100 trials) each option would be played. In the experience group, they chose trial-by-trial while experiencing the result of each choice and only then choosing the next one. Those that had more experience with the choices ended up choosing the risky option far more often, compared to those that had to initially decide (only from the description) how many times that option should be played.
Coming back to the tourists’ behavior, the researchers suggested that when the risk was described to foreign tourists, they over-estimated the probability of the risk, resulting in an aversive reaction. As for the local tourists, they had to face these risky decisions daily (e.g. going to a supermarket, school, work). It is this experience, according to the researchers, that attuned the effect on their behavior.
How exactly experience affects our decisions is yet to be fully understood. How much experience is needed? What counts as an experience? Notably, it is not the experience with terror itself that would reduce the sensitivity to the risk, but rather the experience with decisions involving the risk of terror. There is no doubt that in this period in Israel most of the daily decisions were risky. Nevertheless, there was not only an increase in domestic tourism but also an increase in the income of coffee shops in that time. Yet, in most of the western world, terror attacks are still very rare and are less likely experienced as a daily risk. An argument can be made that this lack of experience contributes to (or at least does not moderate) the over-estimation of the chances of a terror attack.
Decisions with strong affective (emotional) outcomes produce not only different choices, but also reflect different brain networks
Traditionally, it is believed that risky decisions are based on a trade-off between the possible outcome and the probability assigned to it. In other words, we weight the outcome according to how probable it is. Let’s simplify it with an example: earning 2 cents for 90% (safe) or earning 80 cents for 5% (risk). How should we choose? One possible explanation states that the subjective value of the 2-cent option would increase due to its high probability. At the same time, the 80 cents’ value would decrease due to its low probability. At the end of the process, called expectation maximization, we choose the option with the highest subjective valuation. This process can be influenced by different weights we give to the probabilities and to the outcomes.
Most of the studies on risky decisions focused on outcomes with very low emotional value (low-affective), usually earning or losing small monetary rewards. In these low-affective choices we potentially give a stronger weight to the assigned probabilities than to the outcomes. In high-affective choices on the other hand, we might focus more on the emotional outcomes than on the probabilities themselves.
A recent study compared affective-rich and affective-poor decisions. The researchers created two choice-sets, high- and low-affective, and compared the choice-pattern of the participants for each set. At first, participants indicated how much they were willing to pay to avoid certain medical side effects, for example paying 30$ to avoid temporary memory problems. For the affect-rich set they used the medical side effects. The assigned monetary values were used for the affect-poor choice-set. In this way, the researchers could directly high- and low-emotional decisions. In our example, choosing to lose 30$ with 5% probability in one set and choosing memory problems side effect for 5% probability in another set – should be the exact same choice, only with different emotional level. The difference in the choices people make can show us the effect of emotional outcomes on the way we make decisions.
Participants behaved as if they over-estimated the affect-rich rare chances relative to the affect-poor ones. In other words: they chose the side effect much less than the equivalent monetary loss, even though they had the exact same probability assigned to them. The researchers showed that in affective-rich choices, the participants gave a smaller weight to the probabilities, focusing more on the emotional outcome.
The participants were also scanned in an fMRI while making these decisions. Indeed, it seems that there were qualitative differences in brain activation between the two types of decisions. The authors additionally preformed a formal reverse inference meta-analysis. This analysis showed that while low-affective decisions are based mainly on calculative brain processes, high-affective decisions are driven more by the outcomes’ emotional value and associated autobiographical memories.
So why do we fear terror so much?
It is likely that the fear of terror in the general public is partly caused by an over-estimation of the chances for a terror attack. Two aspects of terror attacks in the western world could help us understand part of this over-estimation:
(1) Most of the public in the western world has very little (if any) experience with decisions that involve the risk of a terror attack. Therefore, when estimating the chances for it, this lack of experience could lead to an over-estimation of this Dread risk.
(2) In decisions with high emotional outcomes, we tend to focus more on the outcomes themselves than on the probabilities. A terror attack is a Dread risk with a very strong negative emotional outcome. It is therefore likely that a focus on the outcome would lead to a stronger aversive behavior compared to other risks, even if they have higher probabilities.
At least one goal of terrorism is to create maximum damage with minimal effort. Looking at Germany’s biggest fears relative to the number of victims – it seems that terrorism has succeeded in this goal. The effect of this fear on our behavior could lead to different types of indirect damages: from economical influences (Where to travel, live or go shopping?) through political changes (Who to vote for? Who to give a visa to?) up until racism and discrimination (Who to sit next to on the bus? Who to hire for a job?), or as my colleague Caedyn Stinson has described, starting conspiracy theories. Yet hope is not lost! As Gigerenzer described, the indirect damage is not a necessary consequence of terror but rather a product of our psychology, hence it can and should be changed.