Neurogenesis paradoxically decreases both pattern separation and memory interference

摘要: The hippocampus has been the focus of memory research for decades. While functional role this structure is not fully understood, it widely recognized as being vital rapid yet accurate encoding and retrieval associative memories. Since discovery adult hippocampal neurogenesis in dentate gyrus by Altman Das 1960's, many theories models have put forward to explain plays learning memory. These postulate different ways which new neurons are introduced into their importance Few if any previous incorporated unique properties young adult-born granule cells developmental trajectory. In paper, we propose a novel computational model that incorporates trajectory cells, including changes synaptic plasticity, connectivity, excitability lateral inhibition, using modified version Restricted Boltzmann machine. Our results show superior performance on reconstruction tasks both recent distally learned items, when characteristics taken account. Even though hyperexcitability generates more overlapping neural codes, reducing pattern separation, nonetheless contribute retroactive proactive interference, at short long time scales. sparse connectivity particularly important generating distinct traces highly patterns within same context.