Long-term memory reorganization of navigational episodes

During navigation, the brain builds representations of self-motion and of environmental information for future action. The classic view suggests that these representations consolidate and eventually stabilize. However, there is no data on their fate at extended memory delays. Here, we investigated memory of navigational episodes from a real-world setting across memory delays of up to three decades. We show that memory of navigational episodes does not achieve a stable state, but rather continues to transform for many years. Our data suggest that at any given point in time, memory of navigational episodes is a changing combination of episode-independent schematic information with several interacting spatial representations that directly relate to a navigational episode and that show distinct decay rates. Consistent with recent accounts of memory reorganization, we further show that neither current theories of systems consolidation nor classic models of forgetting fully explain spatial memory performance at extended memory delays.