哺乳动物大脑中空间定位与路径整合的神经机制

Neural Mechanisms of Spatial Localization and Path Integration in Mammalian Brain

近一百年来,对大脑功能的研究不断探索着神经系统编码与解码的原则,感知、识别、定位、运动控制、抉择等方面的编码原则逐层被解读。自1971年位置细胞被发现以来,以学者John O'Keefe为代表的神经科学家们在空间定位与路径整合方面,进行了大量的代表性研究。不同类型的神经细胞陆续被发现,其编码的神经网络机制也日渐清晰。同时,由于近年无人系统的蓬勃发展,工程领域对良好的空间算法也产生了强烈的渴求。在本综述中,我们梳理了当前神经科学领域在空间定位方面的研究进展,并探讨了应用人工神经网络所取得的最新相关理论成果。在可预见的未来,神经科学实验研究与人工神经网络模型研究将会互相促进、迭代进步,而这种循环发展模式也必将为神经科学及工程领域做出更大的贡献。

In the past decades,the neuroscientists explored a plenty of encoding and decoding principles of the central nervous system.The underlying neural mechanisms of perception,recognition,motion control and decision making had been discovered gradually.The place cell was first recorded in hippocampus in 1971,and since then John O'Keefe and others have led a great many pioneering studies on spatial localization and path integration.Neurons with unique firing patterns were studied,and the underlying neural network mechanisms are becoming clearer.Mean while,due to the vigorous development of unmanned systems in recent years,there exists a strong demand for efficient spatial algorithms in the engineering field.Here we reviewed current research progresses on spatial localization in neuroscience,and then discuss the latest relative theoretical achievements obtained by artificial neural networks.In the foreseeable future,experimental researches in neuroscience and theoretical studies of artificial neural networks will promote each other and make iterative progress,and this cycle development mode will be sure to make great contributions to both neuroscience and engineering.