TRAINING IN METHODS OF COMPUTATIONAL NEUROSCIENCE

海洋生物实验室计算神经科学方法课程的目标是在计算神经科学的基本知识问题上每年培养22名高级学生和博士后，并利用分析和数值工具来协助他们的研究。这门课程具有独特的特点，即使不是不可能，也很难在大学环境中复制。首先，它继续吸引和教育来自影响神经科学的广泛学科的学生，包括生物学、认知科学、计算机科学、工程学、数学和物理学。其次，为了在不断发展的计算神经科学领域提供连贯和相对完整的视角，学生除了接受主任、驻校学者和课程助理的指导和指导外，还可以接触到大量的客座教授。在大学环境中，不太可能召集如此多的教员来开设类似的课程。最后，本课程为学生提供了从详细的细胞到更抽象的系统级模拟神经系统的最先进的计算技术和计算机硬件基础。本课程安排为为期四周的密集讲座和系列实验。讲座的进展是随着计算复杂度的增加。他们最初专注于单个神经元和突触的电生理学和化学动力学。接下来是关于局部神经元回路和小神经系统的动力学和可塑性的讲座，这些主题说明了单细胞特性如何通过小网络形成计算。最后的系列讲座涉及大型神经元系统（特别是皮层）中涉及感觉和知觉问题的计算，以及抽象模型的新兴应用，为理解这些系统提供了框架。该实验室强调成功研究计算神经科学所必需的各种工具。一方面，该课程提供通用仿真软件的培训，以及针对单细胞动力学的仿真软件，或针对单细胞和大型网络的仿真软件。另一方面，学生在应用数学领域的训练，如概率论和矩阵代数，这是至关重要的表述和解决问题的领域。

The goal of the Methods in Computational Neuroscience Course at the Marine Biological Laboratory is to train 22 advanced students and post-doctoral fellows a year in the fundamental intellectual issues in computational neuroscience and to avail them with analytical and numerical tools to assist in their studies. The course has unique features that would be hard if not impossible to replicate in a university setting. First, it continues to attract and educate students from a broad range of established disciplines that impact upon neuroscience, including biology, cognitive science, computer science, engineering, mathematics and physics. Secondly, as a means to provide a coherent and relatively complete perspective on the growing field of computational neuroscience, students are exposed to a large visiting faculty in addition to receiving instruction and guidance from directors, scholars-in-residence and course assistants. It is unlikely that such a combination of faculty could be convened for an analogous course in a university setting. Lastly, the course provides students with access to state-of-the-art computational techniques and computer hardware fundamental to simulating neural systems from the detailed cellular through more abstracts systems level. The course is organized as an intensive four week lecture and laboratory series. The lectures progress along an increasing scale of computational complexity. They initially focus on the electrophysiology and chemical dynamics of individual neurons and synapses. This is followed by lectures on the dynamics and plasticity of local neuronal circuits and small nervous systems, topics that illustrate how single-cell properties shape computations by small networks. A final series of lectures addresses computations that involve sensory and perceptual issues in large neuronal systems, particularly cortex, and the emerging application of abstract models to provide a framework for understand these systems. The laboratory emphasizes the spectrum of tools necessary for successful research in computational neuroscience. On the one hand, the course provides training with general purpose simulation software as well as with simulation software that is specific to single cell dynamics, or to both single cells and large networks. On the other hand, students are trained in areas of applied mathematics, such as probability theory and matrix algebra, that are essential to formulating and solving problems in the field.