Position: Hippocampal Explicit Memory Is the Cornerstone for AGI

A research position paper argues that integrating explicit memory systems into Large Language Models is essential for achieving Artificial General Intelligence. The paper contends that current LLMs rely on implicit statistical learning analogous to human implicit memory, but AGI requires higher-order cognitive functions like strategic planning and symbolic reasoning that depend on hippocampal explicit memory mechanisms.

This academic position paper addresses a fundamental architectural limitation in contemporary Large Language Models by drawing parallels between human neuroscience and artificial intelligence design. The research identifies a critical gap: while LLMs excel at pattern recognition through implicit learning mechanisms, they lack the explicit memory systems necessary for advanced cognitive tasks that characterize human general intelligence.

The argument builds on established neuroscience literature distinguishing between implicit and explicit memory systems in human cognition. Implicit memory handles automatic processing and pattern learning, while explicit memory enables conscious recall, episodic reasoning, and long-term planning. Current LLMs function predominantly through implicit mechanisms, excelling at language prediction and pattern completion but struggling with tasks requiring true strategic planning, metacognitive reflection, or symbolic reasoning chains.

For the AI development community, this analysis suggests that scaling parameters alone cannot bridge the gap to AGI. The computational requirements for artificial explicit memory systems present both theoretical and practical challenges that researchers must address through novel architectural innovations rather than incremental improvements to existing transformer-based designs. This perspective could redirect research priorities toward hybrid systems combining implicit and explicit memory mechanisms.

Looking forward, the field will likely see increased exploration of memory-augmented architectures and neuroscience-inspired AI designs. Success in integrating explicit memory systems could represent a pivotal milestone toward more capable AI systems, though significant engineering challenges remain in efficiently implementing such systems at scale without degrading inference performance.

• →Current LLMs rely on implicit learning mechanisms similar to human implicit memory, limiting their capacity for true AGI-level reasoning
• →Explicit memory systems modeled on hippocampal function are identified as necessary for strategic planning, metacognition, and symbolic reasoning
• →The paper draws neuroscience findings to propose computational requirements for implementing artificial explicit memory in AI systems
• →Parameter scaling alone cannot achieve AGI without architectural innovations addressing memory system integration
• →Hybrid architectures combining implicit and explicit memory may represent the next generation of AI development