Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session A: Tuesday, August 12, 1:30 – 4:30 pm, de Brug & E‑Hall

An Algorithmic Model of Working Memory Based on Sparse Variational Gaussian Processes

Dongyu Gong¹, Mario Belledonne¹, Ilker Yildirim¹; ¹Yale University

Presenter: Dongyu Gong

Working memory (WM) involves dynamically manipulating information to support perception, decision-making, and other higher-order cognitive processes. Despite extensive interests in modeling WM, the algorithmic basis of how WM encodes and manipulates information and does so in a goal-driven manner remains unclear. Here, we propose a novel algorithmic model of WM that combines sparse variational Gaussian processes with an adaptive computation algorithm. The model recapitulates a wide range of WM phenomena, including capacity limitations, attraction-repulsion dynamics, and retrocue benefits.

Topic Area: Memory, Spatial Cognition & Skill Learning

Extended Abstract: Full Text PDF