GENESIS: A Generative Model of Episodic-Semantic Interaction
New computational framework uses a 'Cortical-VAE' and 'Hippocampal-VAE' to explain how human memory works.
Deep Dive
Researchers Marco D'Alessandro and team introduced GENESIS, a unified computational model of human memory. It formalizes memory as the interaction between two limited-capacity generative systems: a Cortical-VAE for semantic learning and a Hippocampal-VAE for episodic encoding within a RAG architecture. The model reproduces key behavioral findings like generalization, recall effects, and constructive simulation, providing a principled account of memory as an active, constructive process.
Why It Matters
Offers a unified framework that could lead to more human-like memory systems in AI, improving reasoning and learning.