Learning in the Presence of Unawareness

: The standard assumption in game theory and decision theory is that the game/decision problem is, in a sense, completely understood. For example, decision problems are typically described in terms of states and outcomes, where acts are taken to be functions from states to outcomes. It is typically assumed that a decision maker (DM) knows the state space, the outcome space, and the set of feasible acts. But this is far from clear in practice. In a complex decision problem, agents may be unaware of many relevant features, and thus unaware of possible states, outcomes, and feasible acts.