Reflexive Control

Origin. Vladimir Lefebvre, Soviet military-strategic research, 1960s-1970s. Formalized in "Algebra of Conscience" and reflexive game theory. Developed to model decision-making in adversarial contexts where each party reasons about the other's reasoning.

Mechanism. Decision-making is modeled by explicitly representing each agent's model of the other agents' models, nested to arbitrary depth. The controller influences not the opponent's situation but the opponent's perception of the situation and their model of your model. Information becomes a means of control: intervention operates by feeding the opponent's decision algorithm rather than by changing objective facts or concealing information.

Procedure. Analyze and intervene in adversarial decision-making through reflexive modeling: (1) Map the decision structure — what do you know? What do they know? What do they know you know? What do you know they know you know? Build the recursive model explicitly. (2) Identify the recursion break — at what level is their model of your model inaccurate or absent? This is where intervention is possible. (3) Design the information insertion — what information, inserted at the identified level, would propagate through their decision chain to produce the desired action? (4) Trace the propagation — follow the information through their model: how does it change their perception of the situation, their model of your intentions, their evaluation of options? (5) Execute and observe — insert the information and monitor whether the predicted decision occurs. Discrepancies reveal errors in your model of their model. (6) Iterate — refine the reflexive model based on observed behavior.

Applies to. Negotiation. Competitive strategy. Security design. Disinformation analysis. Any situation where influencing another's decision is the objective and their reasoning process is partially observable.

Limitations. The depth of recursion required increases with opponent sophistication; beyond depth three, models become empirically indistinguishable. The method assumes the opponent reasons; purely reactive or random opponents cannot be reflexively controlled. Lefebvre's ethical formalization (two types of conscience) is culturally specific and does not generalize. The method is descriptive of adversarial cognition but does not itself provide ethical guidance.