Reflexive Game Analysis

Origin. Vladimir Lefebvre's reflexive control theory applied to game-theoretic decision-making; developed in Soviet military-strategic research (1960s-1970s); extended by Novikov and Chkhartishvili.

Mechanism. In adversarial decisions, the opponent's future move depends on their model of your strategy, which depends on their model of your model of their strategy, recursively. Classical game theory assumes common knowledge of rationality; reflexive game analysis drops this assumption and models what each player actually believes about the other. If you know your opponent's model of you is inaccurate at level N, their plan based on that model is exploitable.

Procedure. Construct the decision tree from both perspectives. Identify the levels of modeling: your strategy, their model of your strategy, your model of their model. For each level, assess whether the opponent's model is accurate. Where the opponent's model diverges from your actual strategy, their expected payoffs are miscalculated. Compute your optimal response against their actual (mistaken) strategy. If you can maintain their inaccuracy through information control, the advantage persists.

Applies to. Competitive strategy, security, negotiation, and any interaction with incomplete information where the opponent's beliefs about you are partially observable.

Limitations. Requires accurate knowledge of the opponent's model, which is precisely what is uncertain. The method is valuable when you have evidence of the opponent's beliefs (via signals, past behavior, or intelligence), not when you are guessing. Also: exploiting the opponent's model error is itself information leakage; repeated exploitation teaches the opponent that their model is wrong.