Scholarlink

<p>This paper introduces a novel cognitive architecture that integrates Global Workspace Theory with hierarchical memory systems to model human-like cognitive processes. We present a comprehensive framework that implements Baddeley’s working memory model, episodic and semantic memory systems, and attention mechanisms within a unified architecture. Our experimental evaluation on working memory, episodic memory, attention allocation, and memory consolidation tasks demonstrates that the architecture achieves performance profiles closely matching human cognitive characteristics, with working memory efficiency of 0.86±0.07, episodic recall accuracy of 89.3%, and attention precision of 0.873. These results represent substantial improvements over existing cognitive architectures that do not integrate Global Workspace Theory with hierarchical memory systems, particularly in tasks requiring coordination between attention, working memory, and long-term storage. The architecture shows promise for applications in artificial consciousness, human-AI interaction, and cognitive science research.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A cognitive architecture for modeling human memory and consciousness

  • 期刊论文

  • 发表日期：

    2026年3月23日

查看全文
  • Nicolás Torres

摘要

This paper introduces a novel cognitive architecture that integrates Global Workspace Theory with hierarchical memory systems to model human-like cognitive processes. We present a comprehensive framework that implements Baddeley’s working memory model, episodic and semantic memory systems, and attention mechanisms within a unified architecture. Our experimental evaluation on working memory, episodic memory, attention allocation, and memory consolidation tasks demonstrates that the architecture achieves performance profiles closely matching human cognitive characteristics, with working memory efficiency of 0.86±0.07, episodic recall accuracy of 89.3%, and attention precision of 0.873. These results represent substantial improvements over existing cognitive architectures that do not integrate Global Workspace Theory with hierarchical memory systems, particularly in tasks requiring coordination between attention, working memory, and long-term storage. The architecture shows promise for applications in artificial consciousness, human-AI interaction, and cognitive science research.