Caltech Researchers Study How Blame and Credit Shape Risky Two-Person Decisions

Researchers at Caltech in Pasadena have reported that people making risky decisions with another person often move toward compromise, but those who claim more credit for successful outcomes and less blame for failures tend to compromise less with their partners.

The findings were described in a July 2 Caltech article written by Katie Neith and outlined in a paper published June 29 in the Proceedings of the National Academy of Sciences. The study was conducted by researchers in the lab of Dean Mobbs, a Caltech professor of cognitive neuroscience, with Wenning Deng, a graduate student, and Ketika Garg, a postdoctoral scholar research associate.

The research examined how pairs of strangers handled shared risk in a video-game experiment, including an online social phase. Participants first played individually to measure their tolerance for risk. They were represented by avatars in a virtual environment with nine locations, or boxes, where they tried to collect rewards while avoiding a cartoon predator that could cost them points.

In the individual phase, participants played 120 rounds against two different predators, each with a different overall attack probability, according to Caltech. The goal was to learn the probability that a predator would attack a chosen box, allowing players to maximize rewards while avoiding capture.

In the social phase, they were paired online with a stranger for 60 additional rounds. Each person chose a location, and the two avatars moved to the middle box between their choices. A successful trial meant the player or players kept the reward points assigned to their chosen box, while capture cost them points. Caltech said players received rewards proportional to the points they collected.

The setup allowed researchers to measure whether participants moved toward a partner’s expected choice, which would indicate compromise, or tried to pull the shared outcome back toward their own preferred location. Participants were asked to predict their partner’s choice before making their own.

“We found that most people compromise, settling on something between their two preferences,” Garg said in the Caltech article. “But we also found that when outcomes are shared, people are biased in how they divvy up responsibility.”

After the shared rounds, researchers asked participants to take credit and assign blame for the outcomes of their collectively played games. They found that people were more likely to take credit for wins than blame for losses, even when feedback about both players’ choices made each player’s contribution clear.

Deng, the paper’s lead author, said in the Caltech article that the result was initially surprising because such self-focused responsibility judgments are often observed when evidence is ambiguous. In this study, Deng said, people showed the pattern even when the feedback was clear.

The researchers also found an association between that bias and compromise: People who claimed more credit for successes and deflected more blame for failures tended to compromise less with their partners, according to Caltech.

Mobbs said the study focused on two-person decision-making, or dyadic decision-making, to examine how people with different risk preferences coordinate when their choices are linked. Mobbs is also the Allen V. C. Davis and Lenabelle Davis Leadership Chair and director of the Caltech Brain Imaging Center, according to Caltech.

“We find that compromise is a primary mechanism, that it is reinforced by reciprocity, and that metacognitive biases in responsibility attribution predict who compromises and who doesn’t,” Mobbs said in the Caltech article.

The researchers said the bias has consequences for whether collaboration succeeds or breaks down. Partners who differ but are less biased tend to move toward a shared middle ground, while partners with strong biases can move into conflict by counteracting rather than accommodating each other, according to Caltech.

“Whether a team ends up aligned or in conflict may depend less on how different people are and more on how they assign responsibility when things go wrong,” Garg said.

The Caltech article does not state that Pasadena residents were among the participants. The local connection is that the work was led by researchers at Caltech, whose campus is listed by the institute at 1200 East California Boulevard in Pasadena. The study’s subject, however, may be familiar to readers who have been paired for joint tasks at school or work, examples Caltech itself noted in its article.

The researchers said they want to expand the work by studying larger groups and developing a decision-making game that more closely resembles real-world risks and rewards.

The PNAS paper is titled “Linking Compromise and Responsibility Attribution to Risky Decision-Making in Dyadic Foraging.” Caltech said the research was funded by the National Institute of Mental Health.