Semantic memories are not generated from multiple episodic memories

Fact knowledge – known as semantic memories – is proposed to be derived from individual memories of events during a memory consolidation period after learning. But so far, this has never been shown. Here, we tracked gene expression in a 2-week semantic-like memory paradigm in mice across the whole-brain spatial memory network. Absolute gene expression was driven by first experiences and later learning led to smaller changes, highlighting that multiple episodic memories are not encoded and then consolidated to create semantic-memories, instead already by the second experience semantic-memories present with unique encoding mechanisms in the cortex. Network and graph-theory analysis revealed the retrosplenial cortex to be the central hub in all networks connecting to memory-type unique nodes. Semantic-like memory was characterized by prelimbic-centered network, while 1-Trial and Interference learning presented with a Ca1 and DG-centered network respectively.