What Suppresses Nash Equilibrium Play in Large Language Models? Mechanistic Evidence and Causal Control

ArXi:2604.27167v1 Announce Type: cross LLM agents are known to deviate from Nash equilibria in strategic interactions, but nobody has looked inside the model to understand why, or asked whether the deviation can be reversed. We do both. Working with four open-source models (Llama-3 and Qwen2.5, 8B to 72B parameters) playing four canonical two-player games, we establish the behavioral picture through self-play and cross-play experiments, then open up the 32-layer __TECH_PRESERVE_0TECH_PRESERVE_6__ and examine what actually happens during a strategic decision.