论文标题
多元信息理论发现人类大脑皮层的协同子系统
Multivariate Information Theory Uncovers Synergistic Subsystems of the Human Cerebral Cortex
论文作者
论文摘要
功能连接网络是将大脑建模为复杂系统的最公认的工具之一，该网络研究了相互作用的大脑区域对之间的相关性。虽然功能强大，但网络模型受到仅可见成对依赖性且遗漏潜在的高阶结构的限制。在这项工作中，我们探讨了多元信息理论如何揭示人脑中的高阶，协同依赖性。使用O信息，一种衡量系统结构是冗余或协同主导的量度，我们表明协同子系统在人脑中很普遍。我们对O信息提供了数学分析，以将其定位在更大的多元复杂度度量的分类法中。我们还表明，O信息与先前建立的度量，Tononi-Sporns-Edelman复杂性有关，可以理解为系统量表之间集成的预期差异。高度协同的子系统通常位于规范功能网络之间，并且可以用于集成这些网络。然后，我们使用模拟退火来找到最大的协同子系统，发现此类系统通常构成$ \ $ \ $ \ $ 10的大脑区域，也从多个规范的大脑系统中招募。 Though ubiquitous, highly synergistic subsystems are invisible when considering pairwise functional connectivity, suggesting that higher-order dependencies form a kind of ``shadow structure" that has been unrecognized by established network-based analyses. We assert that higher-order interactions in the brain represent a vast and under-explored space that, made accessible with tools of multivariate information theory, may offer novel scientific insights.
One of the most well-established tools for modeling the brain as a complex system is the functional connectivity network, which examines the correlations between pairs of interacting brain regions. While powerful, the network model is limited by the restriction that only pairwise dependencies are visible and potentially higher-order structures are missed. In this work, we explore how multivariate information theory can reveal higher-order, synergistic dependencies in the human brain. Using the O-information, a measure of whether the structure of a system is redundancy- or synergy-dominated, we show that synergistic subsystems are widespread in the human brain. We provide a mathematical analysis of the O-information to locate it within a larger taxonomy of multivariate complexity measures. We also show the O-information is related to a previously established measure, the Tononi-Sporns-Edelman complexity, and can be understood as an expected difference in integration between system scales. Highly synergistic subsystems typically sit between canonical functional networks, and may serve to integrate those networks. We then use simulated annealing to find maximally synergistic subsystems, finding that such systems typically comprise $\approx$10 brain regions, also recruited from multiple canonical brain systems. Though ubiquitous, highly synergistic subsystems are invisible when considering pairwise functional connectivity, suggesting that higher-order dependencies form a kind of ``shadow structure" that has been unrecognized by established network-based analyses. We assert that higher-order interactions in the brain represent a vast and under-explored space that, made accessible with tools of multivariate information theory, may offer novel scientific insights.
