Individualized Cognitive Simulation in Large Language Models: Evaluating Different Cognitive Representation Methods