The neural architecture of pragmatic processing

项目摘要 功能性磁共振成像（fMRI）对于阐明大脑的功能性 架构这对于认知能力尤其重要，因为动物模型的效用有限，例如 语言然而，人类认知神经科学领域一直在努力，因为许多发现都是 不可复制，存在统计缺陷，或由于使用不同的方法而难以在研究之间进行比较 分析方法。许多人现在已经认识到需要更强大、可复制和有意义的科学。 我们在此建议开发和传播一种强有力的工具，以提高该领域的能力， 建立一个强大的和累积的研究企业。利用我们实验室过去收集的数据 十年（>800名神经典型参与者，> 1，200次扫描），我们建议开发和 为四个对高水平至关重要的大脑网络提供公开的概率功能图谱， 认知：语言选择网络（支持语言处理; Fedorenko等人，2010, 2011年），域一般多需求（MD）网络（支持执行功能，如认知 控制;邓肯，2010年），默认模式网络（DMN）（支持内部导向的认知和 情境模型的构建; Buckner & DiNicola，2019），以及心理理论网络（支持 一般社会推理; Saxe & Kanwisher，2003）。这些地图集将根据大量的 针对这些目标的成熟和广泛验证的“定位器”任务的单独激活图 网络（前三个网络有700多个参与者，ToM网络有约150个参与者）， 用于估计公共大脑空间中任何给定位置属于特定位置的概率。 功能网络。在目标1中，我们将开发这些概率地图集。为此，我们将汇总所有 每个定位器任务的相关数据，通过跨两个最重要的 通常使用的软件包（SPM，Friston，1997;和FreeSurfer，Dale等人，1999年），并覆盖 相关体积和表面空间中的单独激活图。我们还将提取一组密钥 个体水平的神经标记物，以便它们的分布可以被规范地用于与其他神经标记物的比较。 人口。在目标2中，我们将制作地图集（以及组成个体激活图和神经标记） 公开可用。为此，我们将建立一个强大的互动网络平台， 地图集拟议的项目是连接两个根本不同的，目前 功能性脑成像中不相交的分析传统.组平均法和功能性 通过提供共同的参照系在个体大脑中定位：基于概率功能图谱 关于四个高级大脑网络的完善和广泛使用的定位器。能够更 直接比较不同研究结果必然会导致更严格和透明的 科学，从而提高我们对人类沟通和相关能力的理解。

PROJECT SUMMARY Functional magnetic resonance imaging (fMRI) has been invaluable for illuminating the brain’s functional architecture. It has been especially important for cognitive abilities where animal models have limited utility, like language. However, the field of human cognitive neuroscience has been struggling in that many findings are not replicable, suffer from statistical flaws, or are difficult to compare across studies due to the use of divergent analytic approaches. Many have now recognized the need for more robust, replicable, and meaningful science. We here propose the development and dissemination of a powerful tool that can improve the field’s ability to establish a robust and cumulative research enterprise. Leveraging the data collected in our lab over the last ten years (>800 neurotypical participants across >1,200 scanning sessions), we propose to develop and make publicly available probabilistic functional atlases for four brain networks critical for high-level cognition: the language-selective network (which supports language processing; Fedorenko et al., 2010, 2011), the domain-general Multiple Demand (MD) network (which supports executive functions like cognitive control; Duncan, 2010), the Default Mode network (DMN) (which supports internally-directed cognition and construction of situation models; Buckner & DiNicola, 2019), and the Theory of Mind network (which supports general social inference; Saxe & Kanwisher, 2003). These atlases will be created based on large numbers of individual activation maps for well-established and extensively validated ‘localizer’ tasks targeting these networks (700+ participants for the first three networks, and ~150 participants for the ToM network) and can be used to estimate the probability that any given location in the common brain space belongs to a particular functional network. In Aim 1, we will develop these probabilistic atlases. To do so, we will aggregate all the relevant data for each of the localizer tasks, preprocess it through a uniform pipeline across two most commonly used software packages (SPM, Friston, 1997; and FreeSurfer, Dale et al., 1999), and overlay the individual activation maps in the relevant volume and surface spaces. We will additionally extract a set of key individual-level neural markers, so that their distributions can be used normatively for comparisons with other populations. In Aim 2, we will make the atlases (and constituent individual activation maps and neural markers) publicly available. To do so, we will create a robust and interactive web-based platform for the dissemination of the atlases. The proposed project is a critical step to bridge two fundamentally different and currently disjoint analytic traditions in functional brain imaging—group-averaging approaches and functional localization in individual brains—by providing common reference frames: probabilistic functional atlases based on well-established and widely used localizers for four high-level brain networks. The ability to more straightforwardly compare findings across diverse studies is bound to lead to more rigorous and transparent science thus improving our understanding of human communication and related abilities.