Modeling and Measuring Flows between Cognitive and Neural Processes

认知神经科学主要关注与感知、注意和记忆等点状过程有关的区域和网络的定位。这种静态的观点忽略了这样一个事实，即这些认知过程随着时间的推移与正常的心理操作密切相关。认知过程的神经基础也是动态互动的。我们提出了一种更深思熟虑的方法，使用经验神经成像数据的多尺度分析和虚拟大脑（TVB: thevirtualbrain.org）的大规模大脑模拟，以更好地表征认知过程和支持这些流动的功能大脑结构之间的流动。我们将把复杂的大脑动力学分解成概率功能模式。这些模式在数学上可操作化为流形，轨迹随着嵌入在低维空间中的动力学展开而演变（流形上的结构化流[SFM]）。神经网络中可用的功能模式集合构成了它的功能库，这些功能库共同实例化了一套完整的潜在认知功能和显性行为。******将大脑网络动态与认知联系起来的困难在于，大多数认知的行为测量都是单点的，例如反应时间或反应的准确性。我们将评估实时行为测量来构建流的使用，并将这些流与类似地由脑磁图（MEG）测量的神经生理学得出的流联系起来。简单地说，我们将构建与相应的大脑SFM相关的认知SFM。在一个系列中，当人们扫描场景时，眼球运动轨迹将被测量，扫描模式可能与注意力和记忆过程有关。在第二个系列中，参与者将记录对音乐片段的判断。行为轨迹（流）将被组合成流形，流形上特定主题的结构化流（SFM）。然后，行为和大脑SFM可以分析地结合起来，以确定一个流形上的特定试验流是如何被另一个流形上的特定试验流预测的。******将使用经验数据作为TVB中单个大规模脑网络模型的约束，进一步了解SFM之间的联系。TVB产生生物约束的脑网络动态，其输出可以实例化为神经生理信号，如局部场电位、MEG和BOLD-fMRI数据。每个模型的参数将适合于个人独特的大脑SFM。然后将重建单个流，以确定预测每次试验变化的模型参数。大脑和行为SFM之间的联系使得跨时空尺度的大脑动态推断能够支持认知过程的流动

Cognitive neuroscience has focused mainly on localizing regions and networks that are engaged by punctate processes such as perception, attention and memory. Such a static perspective misses the fact that these cognitive processes are intimately intertwined over time for normal mental operations. So too, the neural bases for the cognitive processes interact dynamically. We propose a more deliberate approach using multiscale analyses of empirical neuroimaging data and large-scale brain simulations with TheVirtualBrain (TVB: thevirtualbrain.org) to better characterize the flow between cognitive processes and the functional brain architectures that support these flows. We will decompose complex brain dynamics into probabilistic functional modes. These modes are mathematically operationalized as manifolds, along which trajectories evolve as the dynamics unfold embedded in a low-dimensional space (structured flows on manifolds [SFM]). The collection of functional modes available in a neural network constitutes its functional repertoire, which together instantiates a complete set of potential cognitive functions and overt behaviors. ******The difficulty with relating the brain network dynamics to cognition is that most behavioural measures of cognition are single points, such as reaction time or accuracy of responses. We will evaluate the use of moment-by-moment behavioural measures to construct flows and relate these to flows that are similarly derived from neurophysiology measured with magnetoencephalography (MEG). Simply stated, we will construct cognitive SFM that will relate to the corresponding brain SFM. In one series, eye-movement trajectories will be measured as people scan scenes, where the scan patterns can be related to attention and memory processes. In a second series, participants will register judgments of music clips as they evolve. The behavioural trajectories (flows) will be combined to create manifolds, subject-specific structured flows on manifolds (SFM). The behavioral and brain SFM can then be analytically combined to ascertain how the trial-specific flow on one manifold is predicted by the trial-specific flow on the other.******Further insights into the links between SFM's will be gathered using the empirical data as constraints for individual large-scale brain network models in TVB. TVB generates biologically-constrained brain network dynamics, and its outputs can be instantiated as neurophysiological signals, such as local-field potentials, MEG and BOLD-fMRI data. Parameters for each model will be fit to the person's unique brain SFM. The individual flows will then be reconstructed to identify the model parameters that predict trial-by-trial variation. The link of the brain and behaviour SFM enables inferences of the brain dynamics across spatiotemporal scales that support flow of cognitive processes