A computational approach to network control in the brain

结构如何与功能相关的问题是神经科学中最丰富、最持久的主题之一。大脑不可避免地是一个错综复杂的网络，由各个层次的许多相互作用的组件组成，从细胞内的分子相互作用，到由突触连接的神经元，再到由白质束连接的宏观区域。在过去的几年里，这些网络的更详细的地图已经出现。然而，很少有研究能够提供结构、功能和行为之间的定量联系，特别是在人脑中。健康大脑的连接必须有助于控制：神经系统能够持续控制其电路状态，并保持其输入(外部刺激)和输出(行为反应)之间的功能控制。越来越多的证据，包括我自己对线虫的研究，正在证明网络控制有望实现这一难以实现的目标，并做出可实验验证的预测。线虫是一种长约1毫米的蛔虫，仍然是在神经元水平上完整绘制神经系统的仅有的例子之一。在这个网络中，节点是通过电和化学突触连接的神经元。通过这一提议，我们计划使用网络神经科学和网络控制理论的工具来开发对线虫神经网络的结构和功能的真正跨学科描述。我们的目标是确定一个神经系统内的相互作用如何依赖于来自多个感觉系统的输入来实现和维持控制的理论和算法理解。具体地说，我们试图了解：(I)在网络层面上，大脑如何促进单个行为和整个行为序列；(Ii)大脑状态转换的能量原理，以及为什么在某些情况下选择特定的行为；(Iii)如何恢复受损网络的功能。这些发现将对我们更广泛地理解生物体大脑网络中的结构-功能图谱产生影响。

The question of how structure relates to function is one of the most rich and enduring themes in neuroscience. The brain is inescapably an intricate and complex network of many interacting components at all levels, from the molecular interactions within the cell, to neurons connected by synapses, to macroscopic regions connected by white matter tracts. Over the past years, ever more detailed maps of these networks have been emerging. Yet, few studies are able to provide quantitative links between structure, function, and behaviour, especially in the human brain. The wiring of the healthy brain must facilitate control: a nervous system is able to continuously control the state of its circuitry, and maintain functional control between its inputs (external stimuli) and outputs (behavioural responses). Mounting evidence, including my own research in C. elegans, is demonstrating the promise of network control to achieve this elusive goal and make experimentally testable predictions. C. elegans, a ~1mm long roundworm, remains one of the only examples of a completely mapped nervous system at the neuronal level. In this network, nodes are neurons which are connected by electrical and chemical synapses. With this proposal, we plan to use tools from network neuroscience and network control theory to develop a truly transdisciplinary description of the structure and function of the C. elegans neuronal network. Our goal is to ascertain a theoretical and algorithmic understanding of how the interactions within a neural system, relying on inputs from multiple sensory systems, achieve and maintain control. Specifically, we seek to understand: (i) How, at the network level, the brain facilitates single behaviours, and whole repertoires of behaviours; (ii) The energetic principles of state transitions in the brain and why specific behaviours are selected in certain circumstances; (iii) How to restore function to a damaged network. These findings will have implications for our understanding of the structure-function mapping in the brain networks of organisms more generally.