DRAFT, revised December 2008
For Games, Groups and the Global Good, ed. Simon Levin, 2009
Abstract: A social norm is embedded in a network of supporting norms, which call on other parties to confer punishments or rewards depending on the actor’s compliance with the original norm. One puzzle is how to avoid an infinite hierarchy. A repeated game model of apologizing shows how norms can be arranged in loops of mutual support. Apologies are “all-purpose” supporting norms since the prospect of having to apologize helps deter a range of violations. A second puzzle is why an apology has so many facets – it acknowledges that one committed the offense, that it was wrong and that it caused risk or harm. An apology also expresses remorse and promises that there will be no repetition. Sometimes the actor is ready to perform only some of these speech acts, but recipients typically want full apologies and there are no single words for the partial subsets of functions. A possible explanation is that the elements are synergistic. A game model hypothetically reduces an apology to just a promise not to repeat the offense, but those apologizer-types who are less scrupulous about keeping their promise would be more ready to make one, so promises would not be believed and in the end none are made. Adding a requirement to show remorse gives the words credibility and produces an equilibrium with promising. (219 words).