The computational importance of cerebellar processing

摘要 小脑处理与一系列 行为，从感觉运动转换到执行控制。考虑到这么宽 范围，在组织中的模式和物种之间有显著的一致性 小脑微电路和小脑区的闭合环路 连接到其他大脑区域。这种一致性表明了一种常见的计算 角色，我们假设它通常被描述为自适应时间滤波器。至 检验这一假设，我们将研究简单运动可塑性中的小脑功能， 学习注视稳定性。在该设置中，作为自适应滤波器的处理应为 实现为小脑中的一种能力，以交替地充当成比例的、整合的、 或差分式增益元件。在目标1中，将使用以下方法评估小脑滤波 斑马鱼幼体的双光子钙成像测量两个输入端的活动 颗粒和输出浦肯野细胞群。将确定过滤器的适应情况 通过测量作为注视的输入和输出神经元之间关系的变化 被训练成更稳定或更不稳定。在目标2中，计算模型 小脑将被构建成产生实验测量的信号 并对小脑过滤的机制做出预测。这些 预测将通过局部刺激颗粒细胞和测量 由此产生的浦肯野神经元反应。这些数据结合在一起，有望产生最大的 到目前为止对小脑在计算上的重要性的完全理解 行为。

ABSTRACT Cerebellar processing is associated with the accurate performance of a range of behaviors, from sensorimotor transformations to executive control. Given this wide range, there is remarkable consistency across modality and species in the organization of cerebellar microcircuitry and the closed-loop manner with which cerebellar regions are connected to other brain areas. This consistency suggests a common computational role, which we hypothesize is most generally described as an adaptive temporal filter. To test this hypothesis, we will investigate cerebellar function in a simple motor plasticity, the learning of fixation stability. In this setting, processing as an adaptive filter should be realized as a capacity in the cerebellum to alternatively act as a proportional, integrating, or differentiating gain element. In Aim 1, cerebellar filtering will be assessed by using two-photon calcium imaging in the larval zebrafish to measure activity at both input granule and output Purkinje cells populations. Adaptation of the filter will be determined by measuring changes in the relationship between input and output neurons as fixations are trained toward greater or lesser stability. In Aim 2, computational models of the cerebellum will be constructed that generate the experimentally measured signal transformation and make predictions about the mechanisms of cerebellar filtering. These predictions will be tested by focal stimulation of granule cells and measurement of resultant Purkinje neuron responses. Together these data promise to generate the most complete understanding to date of the cerebellum's computational importance in behavior.