Bottom-up Neurogenic-inspired Computational Model

In this study, we explore computationally the role of the dentate gyrus (DG) in the hippocampus, particularly its significant contribution to the neurogenic process in the adult brain and its role in learning and memory. We introduce a novel computational model, a feedforward multi-layer perceptron oscillatory neural network (MLP-ONN), that incorporates biologically accurate features of the hippocampal-entorhinal cortical network (i.e., DG, CA3, EC). This model leverages empirical data gathered from large-scale simultaneous recordings using a high-density 4096-microelectrode sensing array, providing exceptional spatiotemporal resolution. Unlike previous models, ours considers a dynamically varying DG network size, sparse non-linear oscillatory input patterns, and a dynamic supervised-Hebbian learning rule. We suggest that the incorporation of increasing firing units within the DG layer, with specific levels of sparsity, enhances the well-posedness of our model. This work marks a significant leap in understanding the implications of newly generated neurons in the adult DG network and their broad functional and computational impacts.