Computational modeling of moral decisions

The cognitive processes that give rise to moral decisions have long been the focus of intense study. This chapter reviews recent work illustrating new approaches to investigating moral cognition that borrow from methods traditionally employed in the domains of perceptual and reward-based decision-making. Computational methods have traditionally been employed in the domains of perceptual and reward-based learning and decision-making, but until recently had not been applied to the study of moral cognition. Decades of research on social preferences using economic games have demonstrated that when it comes to monetary exchanges, people do value others' outcomes to a certain extent although they care about their own outcomes far more. The concept of prediction errors may also provide a computational account of the phenomenon of moral hypocrisy, where people view themselves as moral while failing to act morally. Computational models advance theory by forcing researchers to formalize the components of cognition and how they operate at an algorithmic level.