In the theory of psychological games it is assumed that players preferences on material consequences depend on endogenous beliefs. Most of the applications of this theoretical framework assume that the psychological utility functions representing such preferences are common knowledge. But this is often unrealistic. In particular, it cannot be true in experimental games where players are subjects drawn at random from a population. Therefore an incomplete-information methodology is called for. We take a Örst step in this direction, focusing on models of guilt aversion in the Trust Game. We consider two alternative modeling assumptions: guilt aversion depends on the role played in the game, because only the ìtrusteeîcan feel guilt for letting the co-player down, guilt aversion is independent of the role played in the game. We show how the set of Bayesian equilibria changes as the upper bound on guilt sensitivity varies, and we compare this with the complete-information case. Our analysis illustrates the incomplete-information approach to psychological games and can help organize experimental results in the Trust Game