How do people make judgments about appropriate punishment? How do they translate their moral judgments into more tangible penalties? What is the effect of group discussion? And what does all this have to do with social norms?

In this essay I attempt to make some progress on these questions. I do by outlining some of the key results of a series of experimental studies conducted with Daniel Kahneman and David Schkade, and by elaborating, in my own terms, on the implications of those studies. Among other things, we find that the process of group discussion dramatically changes individual views, most fundamentally by making people move toward higher dollar awards. In other words, groups often go to extremes. The point has large implications for the role of norms in deliberation and the effect of deliberation in altering norms. We also find that people's judgments about cases, viewed one at a time, are very different from their judgments about cases seen together. Making one-shot decisions, people produce patterns that they themselves regard as arbitrary and senseless. The point has large implications for the aspiration to coherence within the legal system.

More particularly, our principal findings are as follows:

• In making moral judgments about personal injury cases, people's judgments are both predictable and widely shared. The judgments of one group of six people, or twelve people, nicely predict the judgments of other groups of six people, or twelve people.

• In making punitive damage awards for personal injury cases, people's judgments are highly unpredictable and far from shared. People do not have a clear sense of the meaning of different points along the dollar scale. Hence dollar judgments of one group of six people, or twelve people, do not well predict the dollar judgments of other groups of six people, or twelve people.

• […]