Backward Induction & Subgame Perfect Nash Equilibrium Explained

Added:

The Paradox of Backward Induction and Bounded Rationality

While backward induction and Subgame Perfect Nash Equilibrium (SPNE) are mathematically elegant, they face severe criticism from both epistemic game theory and behavioral economics. The core issue lies in the assumption of 'common knowledge of rationality.' In sequential games like the Centipede Game, backward induction predicts that the first player should defect immediately, resulting in minimal payoffs. However, empirical studies show that real human players routinely cooperate much longer, driven by trust, fairness, or bounded rationality. Furthermore, epistemic critics argue that if a player makes an 'irrational' move off the equilibrium path, the assumption of future rational play becomes logically contradictory. Why should a player assume their opponent will act perfectly rationally in the future if they have already deviated? Alternative frameworks, such as Quantal Response Equilibrium (QRE) and Behavioral Game Theory, offer more realistic predictions by incorporating cognitive limits, noisy decision-making, and social preferences.

Understanding of extensive form games (game trees), including decision nodes, branches, information sets, and payoffs.
The fundamental concept of simultaneous vs. sequential-move games.
Familiarity with standard Nash Equilibrium (NE) and how players identify best responses.
The basic assumption of rationality, where players always seek to maximize their own payoffs.
Analyzing games of imperfect or incomplete information, moving toward Perfect Bayesian Equilibrium (PBE).
Studying repeated games (both finitely and infinitely repeated) and the Folk Theorem.
Applying SPNE to classic economic models, such as Stackelberg competition and Rubinstein's sequential bargaining.
Exploring the limitations and paradoxes of backward induction, such as in the Centipede Game or finitely repeated Prisoner's Dilemma.
68.3K views911likes19:14@selcukozyurtOriginal Release: 2020-10-29

Backward induction is a solution method for extensive form games with perfect information where we start solving from the last players and work backwards, assuming sequential rationality (players choose optimally at every subgame). This process identifies the Subgame Perfect Nash Equilibrium (SPNE), which is a Nash equilibrium that remains optimal in every subgame. To verify SPNE is indeed a Nash equilibrium, we must confirm that no player can unilaterally deviate from their equilibrium strategy and achieve a higher payoff, considering all potential deviations at each decision point.