Modeling And Simulation Of Associative Memory Dysfunction Of Hippocampal CA3 Region With Hopfield-like Network

Objective:This paper aims to study how synaptic deletion and synaptic compensation phenomenon caused by some nervous system diseases(such as Alzheimer's disease) effects autoassociative memory function and heteroassociative memory function of Hippocampal CA3 region. The study probes into the functional neurons of Hippocampal CA3 region which is one of key structures of memory, expecting to reflect how synaptic deletion and synaptic compensation with different levels effect both memory functions of the region through the changement of the properties of the model, neural ensembles firing patterns and average firing rate of neurons.Methods:1. Establish Hopfield-like neural network model, simulate the properties and the autoassociative and heteroassociative memory functions of the model, and calculate the firing rates of neurons. According to the anatomical and internal synaptic connection characteristics of Hippocampal CA3, we establish a three-layered Hopfield-like neural network model, where the neurons is presented by Izhekvich model and connected each other by Hebbian rule. The model is composed of 260 neurons and these neurons are divided into three groups (1-80, 81-180,181-260) which denote respectively CA3a, CA3b and CA3c region. All the three regions have recurrent axons, the CA3b connects both CA3a and CA3c with synapses, but there is no synaptic connection between CA3a and CA3c. Memory can be represented by the patterns of neurons firing sequences, and different memories can be expressed by different patterns.2. Synaptic deletion can be realized by decreasing weight value among neurons, simulate the properties and the autoassociative and heteroassociative memory functions of the model, and calculate the firing rates of neurons under synaptic deletion.3. Synaptic compensation is realized by multiplying compensation factor to weight value, simulate the properties and the autoassociative and heteroassociative memory functions of the model, and calculate the firing rates of neurons under synaptic deletion.4. Compare the properties,the associative memory function and the neurons firing rate of the model respectively under the above three situations, and get the following results and conclusions.Results:1. The Hopfield-like neural network proposed here has autoassociative meory function as well as heteroassociative memory function.2. With increasing of synaptic deletion level, the properties,both associative memory functions and neurons firing rates of the model were all gradually impaired.3. With increasing of synaptic compensation level, the properties,both associative memory functions and neurons firing rates of the model were all gradually increased.Conclusions:1. The autoassociative and heteroassociative memory functions of hippocampal CA3 region can be effectively simulated by the Hopfield-like neural network.2. The phenomenons of synaptic deletion in some neural system diseases can effect the autoassociative memory function as well as heteroassociative memory function of hippocampal CA3 region, which lead to the dysfunction of hippocampal CA3 memory function.3. Compensation strategy can reparation the hippocampal CA3 memory dysfunction leaded by synaptic deletion in certain extent.