What Is Backward Induction?

Backward induction is a classic algorithm for solving dynamic games. It is a backward programming method of dynamic programming, which is derived from the last item one by one. The discussion of the cognitive mechanism is mostly based on a static cognitive model.

Backward induction

Forward induction is exactly the opposite of backward induction. Forward induction refers to participants inferring some information from other participants' early and past decisions to help solve the problem.

In dynamic games, the characterization of players' rational choices is often carried out based on a static cognitive model. In this type of cognitive model, not only the knowledge (or beliefs) of players on the path of rational decision-making need to be described, but also that when an action that is not on the path of rational decision-making is selected by an opponent, each player's original The initial knowledge (or belief) of the player, in this case the player's modification of the original knowledge (or belief), and the opponent's modified knowledge (or belief) of the player. For example, in a dynamic game, player 2 initially knows (or believes) that rational player 1 should choose to end the game immediately. However, he also needs to know (or believe) that if player 1 allows the game to continue, give the opportunity. When letting him make a choice, what is the knowledge (or belief) on which Player 1 is based to guide Player 2 and make a rational choice in this situation. Therefore, such models will necessarily involve complex conditional knowledge (or belief) systems or hierarchical systems and the problem of belief modification. At the same time, based on this type of model, players' initial rational public knowledge cannot be the result of backward induction. ^[3]