Judging responsibility under uncertainty

It has been said that “history is the story of events, with praise or blame” (Cotton Mather). But each of us tells the story of our own life in this way. We are evaluative creatures. We don’t just simply see What people do. We can’t help but think about why they did it, and whether it was a good idea. And when someone’s actions affect our welfare, we tend to hold them responsible. While judgments of credit or blame often come to us easily (and sometimes too quickly), they are founded on sophisticated inferences operating over a rich common—sense understanding of the world. 111 this project, we study how people hold others responsible. We develop computational models of responsibility attribution and test these models in large—scale online experiments. Specifically, we look at the question of how individuals are held responsible in cooperative and competitive team contexts. These group settings are interesting because they highlight the importance of determining what causal role a person’s action played in bringing about the outcome, and of inferring what the action revealed about the person, such as their abilities, beliefs, and intentions. Because people generally only have partial information about the situation, and cannot know for sure how others are going to act, decisions need to be made under uncertainty. Our study of how people hold others responsible for decisions under uncertainty will inform the current debates about responsible artificial intelligence, and help address challenges of responsibility that our society faces, such as global warming.