Sparse memory ensembles set brain-wide network states to sustain learned associations

The ability to recall spatial memories and adapt behavior to changing conditions is essential for efficient navigation of environments. These processes are thought to rely on sparse populations of neurons, embedded within distributed brain networks. However, how the activity of these neurons impacts large-scale network connectivity to sustain behavior remains poorly understood. To address this, we tagged neurons active during peak performance in an appetitive spatial memory T-maze task in male and female TetTag-hM3Dq transgenic mice and chemogenetically reactivated these ensembles during functional magnetic imaging (fMRI) of brain activity. Reactivation of these ensembles triggered widespread network reorganization, optimizing the brain’s modular specialization while maintaining integration across the network via key hubs and gateways. We identified two distinct clusters dominated by the ventral and dorsal hippocampus, respectively, that exhibit distinct connectivity with cortical and subcortical areas. Control mice showed effective extinction learning (EL) in the T-maze when the reward incentive was absent. Reactivating the tagged memory ensembles during EL significantly prevented this process. These findings reveal the brain-wide network hubs that support effective spatial memory acquisition, but also indicate that decay of this network connectivity is likely to be an essential facet of effective EL of spatial experience.