Iterated elimination of strictly dominated strategies is an order dependent procedure. It can also generate spurious Nash equilibria, fail to converge in countable steps, or converge to empty strategy sets. If best replies are well–defined, then spurious Nash equilibria cannot appear; if strategy spaces are compact and payoff functions are uppersemicontinuous in own strategies, then order does not matter; if strategy sets are compact and payoff functions are continuous in all strategies, then a unique and nonempty maximal reduction exists. These positive results extend neither to the better–reply secure games for which Reny has established the existence of a Nash equilibrium, nor to games in which (under iterated eliminations) any dominated strategy has an undominated dominator.

We examine experimentally the impact of communication on trust and cooperation. Our design admits observation of promises, lies, and beliefs. The evidence is consistent with people striving to live up to others' expectations so as to avoid guilt, as can be modeled using psychological game theory. When players exhibit such guilt aversion, communication may influence motivation and behavior by influencing beliefs about beliefs. Promises may enhance trustworthy behavior, which is what we observe. We argue that guilt aversion may be relevant for understanding strategic interaction in a variety of settings, and that it may shed light on the role of language, discussions, agreements, and social norms in these contexts.

We show experimentally that whether and how communication achieves beneficial social outcomes in a hidden-information context depends crucially on whether low-talent agents can participate in a Pareto-improving outcome. Communication is effective (and pat terns of lies and truth quite systematic) when this is feasible, but otherwise completely ineffective. We examine the data in light of two potentially relevant behavioral models: cost-of-lying and guilt-from-blame.

“A clear conscience is a good pillow.” Why does this old proverb contain an insight? The emotion of guilt holds a key. Psychologists report that “the prototypical cause of guilt would be the infliction of harm, loss, or distress on a relationship partner” (Roy Baumeister, Arlene M. Stillwell, and Todd F. Heatherton 1994, 245; June Price Tangney 1995). Moreover, guilt is unpleasant and may affect behavior to render the associated pangs counterfactual. Baumeister, Stillwell, and Heatherton state, “If people feel guilt for hurting their partners ... and for failing to live up to their expectations, they will alter their behavior (to avoid guilt) in ways that seem likely to maintain and strengthen the relationship.” Avoided guilt is the down of the sound sleeper’s bolster. How can guilt be modeled? How are human interaction and economic outcomes influenced? We offer a formal approach for providing answers. Start with an extensive game form which associates a monetary outcome with each end node. Say that player i lets player j down if as a result of i’s choice of strategy, j gets a lower monetary payoff than j expected to get before play started. Player i’s guilt may depend on how much he lets j down. Player i’s guilt may also depend on how much j believes i believes he lets j down. We develop techniques to analyze equilibria when players are motivated, in part, by a desire to avoid guilt. The intellectual home for our exercise is what has been called psychological game theory. This framework—originally developed by John Geanakoplos, David Pearce, and Ennio Stacchetti (1989) and recently extended by Battigalli and Dufwenberg (2005) (henceforth B&D)—allows players’ utilities to depend on beliefs (about choices, states of nature,