Sparse Firing and Connectivity in Neurogenesis Improve Pattern Separation and Memory in a Simulated Dentate Gyrus

Adult neurogenesis, the birth of new neurons continuing into adulthood, is believed to play a role in pattern separation during the encoding of episodic memories, particularly in the dentate gyrus of the hippocampus. However, computational models of the dentate gyrus that incorporate neurogenesis have shown inconsistent results and often omit learning and memory processes. We incorporated the dynamics of aging dentate granule cells into a spiking restricted Boltzmann machine model of the dentate gyrus to study the effects of neurogenesis on pattern separation and memory interference. Our simulations show that including neurogenesis and modeling the detailed changes over the granule cell life cycle leads to a general improvement in pattern separation. At the same time, both proactive and retroactive interference are greatly reduced. These results support the view that sparse firing in mature granule cells and sparse connectivity in young granule cells are critical factors in establishing a functional role for neurogenesis in the dentate gyrus.