Using Neuronal Populations to Probe Perceptual Decisions

描述（由申请人提供）：系统神经科学的一个主要焦点是理解感官信息是如何编码并用于指导行为的。感知决策，就像几乎所有正常的行为过程和神经系统紊乱一样，被认为涉及大量神经元的活动。然而，技术上的限制迫使大多数生理学研究集中在单个神经元上。这些研究提供了许多重要的见解，但它们必然错过了关于感觉神经元群与决策之间关系的关键信息。例如，单个神经元的反应不能告诉我们该神经元的活动是如何相互作用的，以及如何与皮质区域内或区域之间的其他神经元相结合的。此外，我之前的工作表明，即使是训练有素的受试者的认知状态也会随着时间的推移而波动很大，这对感知任务的表现产生了显著的影响。因此，将不同时间记录的单个神经元的信息结合起来，就会产生许多行为状态的平均值。在这个提议的实验中，当动物执行视觉辨别任务时，使用同时记录的许多神经元的活动来跟踪每个感知决策

DESCRIPTION (provided by applicant): A main focus of systems neuroscience is to understand how sensory information is encoded and used to guide behavior. Perceptual decision-making, like nearly all normal behavioral processes and disorders of the nervous system, is thought to involve the activity of large groups of neurons. Technical limitations, however, have forced most physiological studies to focus on single neurons. These studies have provided many important insights, but they necessarily miss key information about the relationship between groups of sensory neurons and decisions. For example, single neuron responses cannot tell us how that neuron's activity interacts and is combined with that of other neurons within or between cortical areas. Furthermore, my prior work showed that the cognitive state of even a well-trained subject fluctuates greatly from moment to moment, with striking consequences on performance on perceptual tasks. Therefore, combining information from single neurons recorded at different times produces an average over many behavioral states. The experiments in this proposal use the activity of many simultaneously recorded neurons while animals perform a visual discrimination task to track each perceptual decision while it is in progress. Our goal is to understand how the responses of neurons with different physiological, visual and cognitive properties who come from different cortical areas and interact with nearby neurons in different ways are combined to drive decisions. In Specific Aim 1, we will ask how the responses of neurons with different functional properties are combined to guide decisions by assessing the relationship between the animal's choices and the activity of neurons that differ in their ability to encode the relevant information or are modulated differently by cognitive factors such as visual attention. In Aim 2, we will assess the validity of current models of decision-making by determining the effect of shared (or correlated) response variability on perceptual performance. The responses from a few dozen simultaneously recorded neurons will provide a means of assessing the correlation structure of a large population and isolating the effects of correlated variability from accompanying changes in firing rates. In Aim 3, we will examine the role of neurons in different visual areas in decisions, ask whether there is complementary choice-related activity in different areas, and determine whether the role of each area is flexible We will record simultaneously from groups of neurons in V4, a ventral stream area, and MT, a dorsal stream area, to determine whether we can better predict the animal's decisions from groups of neurons in the two areas together or from one area alone. We will also adjust the visual stimulus to favor the tuning properties of neurons in each of the two areas to see whether the role of each area in decisions depends on the suitability of its neurons for the particular perceptual task. We hope that extending the study of decision-making to populations of neurons will improve our understanding of the underlying neuronal mechanisms and guide future experimental and theoretical work.