Research Lead: The Illusion That’s Impossible to Shake, a 21st Century Trolley Problem, and More

We are excited to introduce a new feature, the Research Lead. Each month, we will highlight a handful of academic papers that we find interesting and important. Our goal is to provide a portal into original academic research, whatever your level of expertise. We also hope the Research Lead will foster interdisciplinary thinking and collaboration. If you have a suggestion for a future Research Lead, please email us at with your recommendations.

#1: The Illusion That’s Impossible to Shake

“One cannot escape the conviction that one’s views capture reality better than some other set of views,” writes Stanford psychology professor Lee Ross, in a recent article in Perspectives on Psychological Science. “Indeed, any departure from that conviction would be tantamount to the adoption of the conviction that one’s new views capture reality.”

“One cannot escape the conviction that one’s views capture reality better than some other set of views.”

It’s a simple observation that has, for the better part of five decades, in its various forms, iterations, and applications, captured Ross’s imagination. In Perspectives, Ross reflects on the origins and development of his research into how people come to understand their own and others’ behavior. The article is a whistle-stop tour of his most interesting and important work. This includes his research on the fundamental attribution error. A staple of any introduction-to-psychology course, the fundamental attribution error is the observation that people often overestimate the influence of dispositional factors, and underestimate the influence of situational factors, as the reason for someone else’s behavior. But for our own behavior, we are more apt to take into account situational influences. For example, Look at that jerk pushing through the train versus I was late to pick up my kids from school so I had to get off the train quickly. Ross has focused his original idea and now refers to the truly fundamental attribution error: “the illusion of superior personal objectivity.” (You don’t leave yourself much room for adjusting the name, when the original contains the word fundamental.)

This illusion, Ross explains, tends to get us in trouble, working like blinders on how we understand our world. In the article, he reviews the illusion’s relationship to media mistrust, its role in peace and conflict negotiations, and its influence in how we interpret religious teachings. If it’s impossible to avoid the illusion, we can at least be aware of how it influences our thinking and behavior toward ourselves and others. In the current tense social and political climate, this certainly feels like a worthwhile task. As Ross observes, “Citizens of good will are generally open, even eager, to engage in dialogue with the other side, but they do so with the goals of enlightening the other side about what is true and what justice demands, and they do so with little thought that they themselves might be in need of enlightening.”

Ross, L. (2018). From the Fundamental Attribution Error to the Truly Fundamental Attribution Error and Beyond: My Research Journey. Perspectives on Psychological Science, 13(6), 750-769. (Link)

#2: A Trolley Problem for the 21st Century

Imagine you’re a programmer for a self-driving car. You’re writing the accident avoidance algorithm and you’ve got to make some moral decisions. There are two children on one side of the street and three grandmas on the other, and you have to choose to save one group and kill the other. Who lives and who dies? How about a runner and her dog versus a two homeless people? A woman versus three cats?

In an update to the classic trolley problem, researchers at the Massachusetts Institute of Technology, developed an online game to research moral decisions, like the ones above, all over the world. In total they collected 40 million moral decisions, from several million participants, from 233 countries. (Give it a try here.) Three fundamental principles emerged in people’s preferences: save people over animals, save the greater number, and save children. They also identified three clusters of moral profiles. The first was made up of many Western countries, the second many Eastern countries, and the third comprised Latin America and many former French colonies. An example of differences that emerged among these clusters includes the finding that Eastern countries don’t prefer to save young lives as strongly as the other clusters. The researchers also found that the more economic inequality in a country, the more likely people were to save a businessperson over a homeless person.

The findings provide us with an interesting look at moral decision-making in what could soon be our technological future. However, don’t expect developers to rush to program driverless cars according to the preferences observed in the study. For one, using public preferences to set ethics can be tricky territory as prejudice and bias come into play. Second, for programmers, the problem is not who to kill or save, but how to avoid dangerous scenarios in the first place. Nevertheless, the study provides an interesting and entertaining update to the trolley problem at a scale previously unimaginable, forcing us to reflect and debate on our moral decisions.

Awad, E., Dsouza, S., Kim, R., Schulz, J., Henrich, J., Shariff, A., … & Rahwan, I. (2018). The Moral Machine experiment. Nature, 563(7729), 59. (Link)

#3: Research for the Good of Soccer and Open Science

Recently, 29 research teams from around the world set out to answer the same question: are soccer referees more likely to award red cards to darker-skinned players? As intriguing as the question is (we’ll get to the results in a moment), the real purpose of this project was to understand how different research teams approached, analyzed, and justified their scientific decisions. All 29 teams used the same data set from four major soccer leagues. Throughout the project, the research teams shared feedback on their analyses and reviewed and provided feedback on the analysis of the other teams. Each team also had the opportunity to update their analytical approach based on the feedback.

What did they find out about referee decisions, and what did they learn about crowdsourcing data analysis?

Overall, 20 of the 29 teams found that darker-skinned players were more likely to receive a red card, whereas 9 teams found no significant relationship. The median result was that compared to lighter-skinned players, darker-skinned players were 1.3 times more likely to be on the receiving end of a red card; findings certainly worth looking into for soccer’s governing bodies, like FIFA, who’ve led public campaigns to help the sport “Say No to Racism.”

“Had any one of these 29 analyses come out as a single peer-reviewed publication, the conclusion could have ranged from no race bias in referee decisions to a huge bias.”

From a methodological perspective, the project’s unique crowdsourcing setup provides a much-needed look at how rigorous science can still produce a range of results. “Had any one of these 29 analyses come out as a single peer-reviewed publication, the conclusion could have ranged from no race bias in referee decisions to a huge bias,” wrote Raphael Silberzahn and Eric Uhlmann, two of the co-authors. The article provides an inventory of the methodological approaches of the 29 teams, as well as a detailed timeline of how the ambitious effort was coordinated.

Crowdsourcing, however, is not without costs. There is the opportunity cost of 29 teams putting their time and resources into answering a single question. There is also the effort and time needed to coordinate a project with so many people and parts—a significant undertaking in its own right. That said, for complex and important questions, crowdsourcing may have a key role to play in helping us understand a particular phenomenon or issue, as it leads to a more holistic understanding of that phenomenon or issue. It also provides researcher with a space in which they can test and compare their methods with their peers’, strengthening the knowledge of the community. The lesson is that crowdsourcing data analysis is a powerful tool for scientists, but when it’s used, just be sure to bet on nuance rather than consensus.

Silberzahn, R., Uhlmann, E. L., Martin, D. P., Anselmi, P., Aust, F., Awtrey, E., … & Carlsson, R. (2018). Many analysts, one data set: Making transparent how variations in analytic choices affect results. Advances in Methods and Practices in Psychological Science, 1(3), 337-356. (Link)

#4: A Glimpse into the Evolution of Human Cooperation

A new study, published in Current Biology, provides interesting evidence for how human cooperation may have evolved. The research team, led by Coren Apicella, conducted research among the Hadza hunter-gathers in Tanzania. The goal was to understand the factors that influence cooperation over time, as the hunter-gathers move camps and interact with different sets of people. For instance, do cooperators form their own super groups of pro-sociality? Do people have a disposition toward acting selflessly versus selfishly?

“The Hazda provide an important test case for evolutionary models of cooperation,” the authors write. “Their daily life is marked by widespread sharing of food, labor, and childcare. And their lifeways more closely approximate pre-Neolithic populations compared to samples drawn from industrialized settings.”

“The strongest predictor of an individual’s level of cooperation is the mean cooperation level of their current campmates.”

Over the course of six years, the researchers employed a public-goods paradigm to measure cooperation from 383 individuals at 56 camps. In the public goods game, participants contributed sticks of honey to a public pool, where contributions were multiplied by three, and then divided among the group. The researchers measured the rate at which individuals at different camps were willing to cooperate with others and increase the public good.

“Consistent with social norms, culture, and reciprocity theories, the strongest predictor of an individual’s level of cooperation is the mean cooperation level of their current campmates,” the authors state. This means that rather than a predisposition to acting selfishly versus selflessly or super cooperators clustering together, individuals tend to act like those around them act. This, the authors suggest, provides insight into how cooperation evolved: “These findings highlight the flexible nature of human cooperation and the remarkable capacity of humans to respond adaptively to their social environments.”

Smith, K. M., Larroucau, T., Mabulla, I. A., & Apicella, C. L. Hunter-gatherers maintain assortativity in cooperation despite high-levels of residential change and mixing. Current Biology, 28(19), 3152-3157. (Link)