Source: https://arxiv.org/abs/1701.07646
作者:Pranav G. Reddy, Marcelo G. Mattar, Andrew C. Murphy , Nicholas F. Wymbs, Scott T. Grafton, Theodore D. Satterthwaite, Danielle S. Bassett
学习需要遍历固有不同的认知状态来产生行为适应。然而明确测量状态的非侵入式成像工具 , 以及在学习过程中评估其动态的工具 , 仍然有限。本文描述一种基于图论的新颖方法,其中时间点由网络节点表示,并且两个不同时间点之间的脑状态的相似性被表示为网络边缘。我们使用基于图形的聚类技术来识别代表规范大脑状态的时间点团簇,并评估大脑学习过程中从一个状态转移到另一个状态的方式。我们观察到两个主要状态的存在,以感觉运动皮质中的高激活或额叶皮质下系统中的高活化为特点。随着学习进展,这些主要状态和其他较不常见状态之间的柔性切换变得越来越频繁,并且与学习率的个体差异成反比关系。这些结果与更大自由的从其他过程使用认知资源的自动发展(the development of automaticity )理念是一致的。综合起来,我们的工作提供了对早期学习中大脑运动约束,低维度特征的的新见解,在之后的学习中形成较少受限的高维度运动。
延展性,适应性和可塑性(Malleability, adaptability, and plasticity )通常在描述一个系统的结构或功能的定量统计中表现为一个变量。在人脑中,这种统计可能指神经生理学噪声的测量或休息状态功能连接模式的变化。最近,使用社区检测算法来确定大脑功能网络的动态变化已经被用来描述网络的灵活性,其在个体之间是不同的,相应于个人在学习,认知灵活性和执行功能的差异。从同一参与者在跨越学习过程的多个时间点收集数据的动态网络方法非常有用。但动态网络重构度量方法无法评估大脑动力学活动模式的特征变化,因为它们需要长时间地计算来估量功能连接性。
为了克服上述弊端,利用网络科学作者发展出另外的方法以识别时变激活模式并评估其灵活性。原始方法资料可以参考引用文献。在图形信号处理文献类似方法在独立发展。利用这种方法我们能追问”大脑的激活模式如何随着学习而变化?“ 作者利用人类和动物模型中研究典型学习过程的方法--明确获得一种新的机动视觉技能--研究这个问题。当参与者练习任务时,假设大脑横穿不同地标准状态,这种穿越的特征能预示学习中的个体差异,而且学习过程的早晚期规范状态本身本质上是不同的。
实验过程:20名健康的成年人参与者练习一组十个运动序列。以按钮按钮的形式将视觉刺激转换为运动反应(参见图2)。 任务执行期间每隔两周分开4次采集BOLD数据(图1a-b);在两次扫描之间,参与者在家里做十次训练。为了评估行为变化,将运动时间(MT)定义为给定运动序列的第一个按钮按下和最后一个按钮按下之间时间差,学习率由拟合MT数据的两项指数函数(two-term exponential function)的指数下降参数量化。为了评估与行为变化相关的大脑活动变化,将大脑分为112个皮层和皮质下区域,并计算了各区域BOLD时间序列(图1c)。作者将每个时间点的大脑状态定义为跨区域的BOLD强度。然后,用所有状态对之间的秩相关度测量(a rank correlation measure between all pairs of states ),来量化不同时间脑状态的相似度,从而构建每个试验的相关性的对称矩阵(图1d)。在每次试验中,使用基于网络的聚类算法来查找独立于其时间顺序的反复出现的脑状态(图1e-f)。
Figure 1: Schematic Depicting Construction of Adjacency Matrices.
(a) Blood-oxygen-leveldependent (BOLD) signal from functional magnetic resonance imaging (fMRI) data was acquired from healthy adult subjects.
(b) We calculated the mean BOLD magnitudes in each of 112 cortical and subcortical regions as a function of time.
(c) The regional time series is represented in matrix format, and
(d) the correlation between matrix columns (TRs) is used to create a time-by-time adjacency matrix. The ijth element of this matrix measures the similarity between the regional pattern of BOLD magnitude between TR i and TR j. Adjacency matrices representing time-by-time networks form the fundamental data structure on which community detection algorithms function. We maximize a modularity quality function informed by these matrices to extract network communities: groups of TRs that show similar regional patterns of BOLD magnitudes.
(e) Due to the near-degeneracy of the modularity landscape, this procedure is repeated 100 times per matrix.
(f) Across these 100 partitions of TRs (nodes) into groups (communities), we construct a representative or “consensus” partition
(g) that summarizes the significant structure in the original matrix.
作者发现,大脑状态的3-5个社群有两个最高频次出现的”反相关“状态。随着练习深入,大脑的”状态灵活性“增加,这主要是由传统上认为与任务学习与记忆的大脑区域驱动。此外,具有较高状态灵活性的人比在脑状态之间切换较少的人学到更快。 这些结果表明,脑活动的全局模式为支持适应行为的神经生理动力学提供了重要的视野,强调了用全脑状态动力学估计来理解高阶认知功能(如学习)。
实验和数据获得:
fMRI 成像
网络构建和分析
Time by time network analysis identifies frontal and motor states
State flexibility increases with task practice
Regional contributions to state flexibility vary by function
state flexibility:following [80], we specified state flexibility (F ) to be the number of state transitions (T ) observed relative to the number of states (S), or F = TS . Intuitively, state flexibility is a measure of the volatility versus rigidity in brain dynamics, directly representing the frequency of dynamic state changes.
如何计算单个不同区域对state flexibility的影响?利用破坏缺失的办法。即在状态时间序列矩阵中抽调该区域的时间序列,计算前后的state flexibility的变化。作者发现:Regions in the motor and visual cortex tend to have negative contributions to state flexibility while regions in the frontal lobe tend to have positive contributions to state flexibility.
State flexibility is correlated with learning rate
learning rate: To assess behavioral change, we defined movement time (MT) to be the time between the first button press and the last button press for any given sequence, and learning rate was quantified by the exponential drop-off parameter of a two-term exponential function fit to the MT data.
Figure 7: Individual Differences in Learning Rate Correlated with State Flexibility. State flexibility difference refers to the difference in state flexibility between subsequent scans. State flexibility differences for all regions were computed and were found to be significantly positively correlated with individual differences in learning rate (p = 1.163 × 10^−7), suggesting that the observed increase in flexibility is associated with the learning rate of subjects.