We pose the problem of modeling an agent's cognitive model subject to bounded rationality and high-cognitive load. Cognition is considered as a game of complexity between nature and the agent, both operating on a cognitive graph. While the first player tries to maximize the complexity by maximizing the entropy, the second player tends to minimize it. We establish the game theoretical equilibrium and provide the family of dominant strategies that minimize the cognition complexity. Particularly, we prove that adopting a power law distribution while exploring the cognitive graph tends to minimize the cognition complexity.