Variation as a neural code

点击翻译按钮获取中文摘要

Visually guided reaching is a centrally important behavior in both human and non-human primates. Reaching is a highly stereotyped behavior, with similar movement paths and velocity profiles observed across a wide range of conditions, across subjects, and even across species. Yet another hallmark of visually guided reaching is trial by trial variability in the details of the movement and in the movement endpoint. The goal of this project is to characterize the nature of this variability and its neural origins and to test two hypotheses about the role that variability plays in such sensorimotor circuits. There are three specific aims. 1) We will characterize the relationship between variability in visually guided reaching and neural activity in Macaque ventral premotor cortex (PMv) and primary motor cortex (M1). We will perform a detailed analysis of the trial by trial movement variability, using a number of novel analysis techniques. In addition to yielding new insight into the nature of movement variability, this analysis will provide a rich set of behavioral variables that can be compared to neural activity. We will record from PMv and M1 in Macaque monkeys while they perform a reaching task. The trial by trial neural activity will be decomposed into variability that can be explained by the behavior and a residual variability that unrelated to behavioral variability. 2) We will test our hypothesis that reach variability is under central control by attempting to alter variability using surreptitious manipulations of movement feedback. With human subjects, we will either reshape the mapping from reach endpoint to reward feedback or introduce time-varying changes in the visual feedback of hand position in order to reshape the variability of reaching. We will then employ these behavioral techniques with Macaque while we record from PMv and M1, and we will correlate the changes in movement variability along each spatial axis with changes in the explained and residual variability of neural activity. 3) We will test our hypothesis about the relationship between neural variability and learning. Specifically, we propose that explained variability in a neural circuit is aids learning in that circuit, since it provides a richer sampling of the input-output space, while residual variability is detrimental for learning, since it degrades the quality of the error signal. We will test this hypothesis by studying a particular form of sensorimotor learning: the rapid adaptation that follows exposure to shifted visual feedback. By comparing the learning rate of shift adaptation to the level of behavioral and neural variability, we will determine whether either type of variability affects sensorimotor learning.