Cognition with and without Cortex: Implications from Mouse Navigation in a Novel Reconfigurable Maze

Harvard RL and Brain Seminar

This talk introduces the Manhattan Maze as a reconfigurable navigation task for studying rapid learning and flexible adaptation. Wildtype mice rapidly learn their first map, retain overnight memory, and accelerate learning in new configurations. Acortical mice, despite lacking hippocampus and most of neocortex, can learn multiple maps and retain solutions after long gaps, suggesting cortex is not strictly required for solving the Manhattan Maze.

Presented for the Harvard RL and Brain Seminar on November 14, 2024. The talk connects mouse navigation, cortex, flexible decision-making, long-term memory, and the broader question of how brains organize cognition.