This paper introduces a Cognitive Personalized Search (CoPS) model that integrates Large Language Models (LLMs) with an efficient memory mechanism to enhance user modeling and improve personalized search results. Traditional search engines provide identical results for all users, but personalized search re-ranks results based on user preferences. However, deep learning-based methods rely heavily on training data, making them vulnerable to data sparsity. CoPS addresses this by leveraging LLMs, which excel in zero-shot scenarios, and incorporates a cognitive memory mechanism inspired by human cognition, consisting of sensory memory, working memory, and long-term memory. The model handles new queries through three steps: identifying re-finding behaviors, constructing user profiles with historical information, and ranking documents based on personalized intent. Experiments show that CoPS outperforms baseline models in zero-shot scenarios. The model's contributions include an LLM-empowered personalized search model, efficient access to extensive user histories via external memory units, and a memory architecture mimicking the human brain's memory mechanism. The paper also discusses related work, experimental settings, and results, demonstrating that CoPS achieves superior performance in personalized search, particularly in handling repeated queries and improving efficiency. The model's design addresses privacy concerns and ensures scalability, making it suitable for real-world applications.This paper introduces a Cognitive Personalized Search (CoPS) model that integrates Large Language Models (LLMs) with an efficient memory mechanism to enhance user modeling and improve personalized search results. Traditional search engines provide identical results for all users, but personalized search re-ranks results based on user preferences. However, deep learning-based methods rely heavily on training data, making them vulnerable to data sparsity. CoPS addresses this by leveraging LLMs, which excel in zero-shot scenarios, and incorporates a cognitive memory mechanism inspired by human cognition, consisting of sensory memory, working memory, and long-term memory. The model handles new queries through three steps: identifying re-finding behaviors, constructing user profiles with historical information, and ranking documents based on personalized intent. Experiments show that CoPS outperforms baseline models in zero-shot scenarios. The model's contributions include an LLM-empowered personalized search model, efficient access to extensive user histories via external memory units, and a memory architecture mimicking the human brain's memory mechanism. The paper also discusses related work, experimental settings, and results, demonstrating that CoPS achieves superior performance in personalized search, particularly in handling repeated queries and improving efficiency. The model's design addresses privacy concerns and ensures scalability, making it suitable for real-world applications.