Inverse engineering preferences in simple games

摘要

A method for inverse engineering decision-makers’ preferences based on observable behaviour is designed. This technique allows analysts to narrow down the list of potential preference rankings of possible states in a conflict for each decision-maker using probabilities and expected values. During the inverse engineering procedure, the list of all possible preference rankings is narrowed as decision-makers move and counter-move. Accurate preference information is key to building quality conflict and game models; however, preference rankings for decision-makers are often difficult to obtain directly. A simple two decision-maker, four-state game is used to demonstrate the applicability of the method and to illustrate the insights it provides.