海马记忆功能的神经网络模型

综合神经心理学，神经生理学、解剖学与神经网络研究的成果，提出一个海马记忆功能的神经网络模型。模型由三个神经网络所组成；海马的ＣＡ１和ＣＡ３网络和大脑皮层联合区，ＣＡ３的功能是将不同感觉输入联合起来，而ＣＡ１的作用是将它们结成一个单一的记忆。而大脑皮层则是长期记忆的部位。在ＶＡＸ１１／７５０上进行计算机仿真，仿真证明模型有近期及长期记忆功能，破坏模拟海马的部分，模型显示出与顺行性遗忘症相似的特性。在

NEURAL NETWORK MODEL OF HIPPOCAMPAL MEMORY FUNCTION

Integrating the results of neuropsychological. neurophysiological. anatomical and neural network researches. in this paper, a new neural network model of hippocampal memory function is proposed. The model consists of three neural networks, i. e. the hippocampus CA1, CA3and the association areas of neocotex. The role of CA3 is to associate different sensory input sig-nals. that of CA1 is for conjugating them to form a single memory and the neocortex is the site for long term memory. The simulations were performed on VAX 11/750 computer, and showed that the model have temporal and long term memory. When the part that simulated the hippocampus, was damaged the model behaved just like anterograde amnesia. Under the frame-work of the model, a new hypothesis is proposed for memory consolidation. This hypothesis is more concrete than that has been published in other literatures, and easy to be implemented by using neural networks.