We aim to investigate how moral judgments differ for acts committed by the self versus another person, at the behavioral, cognitive, and neural level. **Project Background** In past work, we found that people condemn immoral acts slightly *more* when they themselves have committed them, versus when they judge acts committed by others - but only when the act is accidental (Cushman & Young, unpublished data). This finding stands in contrast to a large body of work in social psychology on self-serving biases, including the fundamental attribution error. Though a small effect, we have replicated it several times. As this difference manifests only for acts that are unintended, we suspect that this difference in judgment may reflect an underlying difference in representations of one's own mind, versus the mind of another. Our preliminary work supports a subtle version of this hypothesis. Participants seem to attend just as much, or more, to their own interior states when considering self-generated acts as when considering other-generated acts. However, they do not seem to use this information to mitigate their moral judgments - i.e., they know they did not intend to cause harm, but they condemn themselves anyway. With neuroimaging data, we hope to better understand how neural representations of one's own and others' mental states play a role in how people process and respond to moral violations. **Current Directions** I (Emily Wasserman) have extended this project in the form of a novel card-choosing game. A behavioral version of this game has been tested on Amazon Mechanical Turk and piloted on behavioral participants. We have also adapted the task to the scanner, and collected functional MRI data during this task, as well as a theory-of-mind localizer and self-versus-other localizer task. **Project Status** We have finished collecting behavioral pilot data (see "Pilot Study") and have almost finished collecting MRI data (see "MRI Study").