A multi-electrode study of cortical attractor networks in spatial working memory

描述(由申请人提供)： 我们将利用背外侧前额叶皮质(DLPFC)的多个单一单位记录来探索空间工作记忆的电路，以研究神经元活动如何在长时间延迟时变化，并利用这些信息更好地理解工作记忆和空间计算的电路。这个项目将利用多个单单元记录来测试关于空间工作记忆神经结构的几个假说。工作记忆在许多认知功能中起着至关重要的作用，因此成为心理学、生理学和计算研究的重要课题。工作记忆受损是阿尔茨海默氏症和精神分裂症等许多疾病的主要因素。了解工作记忆涉及的神经机制和电路，对于理解依赖于工作记忆的认知功能以及理解相关疾病的病理生理都是至关重要的。空间工作记忆会随着时间的推移而衰退。在单个神经元水平上测量的这种衰退的神经元相关性，将有助于阐明空间信息是如何编码的，以及空间工作记忆如何在单个神经元和小电路水平上发挥作用。DLPFC中的单个细胞具有限定的助记场，类似于感受野(Goldman-Rakic 1987；Funahashi，Bruce和Goldman-Rakick，1989)。我们将研究这些场是否以及如何随着记忆的衰退而改变。我们在吸引子网络的计算环境中考虑这一点，吸引子网络是空间工作记忆的主要模型(Comte等人。2000)。目的1：测试DLPFC中单个神经元的助记场如何随着记忆衰退而变化，并在单细胞水平上将这些变化与记忆随时间衰退的行为分析联系起来。我们将获得单个细胞助记场在时间和空间上的调谐曲线。通过同时从多个细胞记录，我们将获得关于这些曲线如何随时间变化的绝对和相对信息，以及这些变化如何与空间工作记忆内容的逐次尝试变化相关。根据这些数据，我们将确定由人口形成的活动凸起是否在较长的延迟期内漂移、幅度下降、扩大或变薄。我们还将确定这些变化中的哪些与反映在行为测量中的记忆衰退的变化相关。目的2：通过比较存储单个空间位置与存储多个位置的神经元关联，检验目标1中揭示的空间工作记忆机制的普遍性。我们将构建猕猴在工作记忆中持有两个空间位置时的调谐曲线，并研究这些调谐曲线与当两个相同位置中的每一个单独保持时生成的调谐曲线的比较。我们将考虑和测试单个空间工作记忆网络与多个独立网络的假设，以及替代策略的假设，例如保持一组位置之间的相对关系(例如，记住一行而不是两个位置)。

DESCRIPTION (provided by applicant): We will probe the circuitry of spatial working memory using multi single-unit recording in the dorsolateral prefrontal cortex (DLPFC) to study how neuronal activity changes over long delays, and use this information to better understand the circuitry underlying working memory and spatial computations. This project will make use of multi single-unit recording to test several hypotheses regarding the spatial working memory neural architecture. Working memory plays a crucial role in many cognitive functions and as a result has become an important topic of psycophysical, physiological, and computational study. Impairments in working memory are major factors of many disorders such as Alzheimer's Disease and schizophrenia. Understanding the neural mechanisms and circuitry involved in working memory is vital for understanding cognitive functions that depend on working memory as well as for understanding the pathophysiology of related disorders. Spatial working memory decays over time. The neuronal correlates of this decay, measured at the single neuron level, will help elucidate how spatial information is encoded and how spatial working memory functions at the singe neuron and small circuit level. Single cells in DLPFC have circumscribed mnemonic fields, analogous to receptive fields (Goldman-Rakic 1987; Funahashi, Bruce, and Goldman-Rakick, 1989). We will study whether and how these fields change as memory decays. We consider this in the computational context of an attractor network, the premier model for spatial working memory (Compte et al. 2000). Aim 1: Test how the mnemonic fields of individual neurons in DLPFC change as memory decays, and relate these changes at the single cell level to a behavioral assay of memory decay over time. We will obtain tuning curves, in time and space, for the mnemonic fields of individual cells. By recording from multiple cells simultaneously, we will have both absolute and relative information about how these curves change over time, and how these changes relate to trial-by-trial changes in spatial working memory content. From these data, we will determine whether the activity bump formed by the population drifts, drops in amplitude, broadens or thins out over a long delay period. We will also determine which of these changes are correlated with changes in memory decay as reflected in behavioral measures. Aim 2: Test the generality of the spatial working memory mechanisms revealed in Aim 1 by comparing the neuronal correlates of storing a single spatial location with multiple locations. We will construct tuning curves when a macaque is holding two spatial locations in working memory and examine how these tuning curves compare to tuning curves generated when each of the same two locations is maintained individually. We will consider and test hypotheses of a single spatial working memory network vs. multiple independent networks, as well as hypotheses of alternative strategies such as maintaining relative relationships between a set of locations (e.g. remembering a line instead of two locations).