Keynotes | K&Ts | GACs | Talks | Posters | Search

Poster C in Poster Session C: Wednesday, August 5, 9:30 – 11:15 am, Kimmel Center, Shorin & Rosenthal Rooms

Mechanistic Model of Salience-Gated Memory Consolidation via the Anteromedial Thalamus

Stephan Kim¹, Leslie Sibener², Sarthak Chandra³, Priya Rajasethupathy⁴, Ila R Fiete¹; ¹Massachusetts Institute of Technology, ²Columbia University, ³International Centre for Theoretical Sciences, ⁴Rockefeller University

Presenter: Stephan Kim

New memories are formed in the hippocampus and later stabilized in cortex through memory consolidation, but the mechanism that determines which memories are selected remains unknown. Experimental evidence identifies the anteromedial thalamus (AMT) as a key gating structure, despite its small size and predominantly feedforward connectivity. Here, we present a mechanistic model, in which the AMT filters salient memories through interactions with a high-capacity entorhinal–hippocampal memory system and cortical attractor dynamics. Reverse replay exposes outcome-related signals early, enabling the AMT to generate a sustained modulatory signal that gates cortical plasticity through a three-factor rule, thereby selectively consolidating only salient memories. The model reproduces the full spectrum of experimental findings, including region-specific perturbation effects and the bidirectional control of consolidation via AMT–cortex pathway modulation. These results show that a small feedforward circuit can implement scalable and selective consolidation across a large memory space, while yielding concrete functional implications and experimental predictions.

Topic Area: Memory, Learning & Knowledge Structures