Converting Value into Action: Computations in Corticostriatal Circuits for Flexible Decision Making

项目摘要 为了灵活地执行行为，选择是基于先前的结果，将最大化奖励。 至关重要的是，学习每个行动的价值以获得奖励被认为是推动这一决策过程的动力。 在基于价值的决策框架中，首先计算这些价值，然后使用这些价值来选择和 执行动作。这种决策过程的功能障碍在许多神经精神疾病中是明显的 包括成瘾和额叶皮质损伤的患者，他们表现出无法灵活地调整或适应 他们的行为皮质纹状体通路，从内侧前额叶皮质（mPFC）到其下游靶点， 背内侧纹状体（DMS）是参与促进灵活行为的候选回路。神经相关物 在不同物种的mPFC中发现了有价值的，传统上认为DMS参与了行动， 选择，其中已经在DMS神经元中发现了动作值的表示。但目前尚不清楚 价值的表达是如何转化为适应行为的实际动作的。 同样，许多描述这两个地区价值表征的研究都是使用 行为的离散二进制度量，并且仅将神经活动与动作的端点相关联（例如，发出 舔左或右喷口，或对目标进行扫视）。本项目将研究 通过研究mPFC和DMS神经活动如何代表持续过程中的价值， 运动（目的1），并将揭示特定的细胞类型和皮质纹状体通路内的途径， 是价值表现和行动选择所必需的（目标2）。在这两个目标中，来自mPFC的动作电位 当口渴的头部固定小鼠在需要操纵杆的动态觅食任务中执行时，DMS将被记录下来 移动以搜索高概率奖励，其中高概率奖励的位置改变 随着时间基于价值的决策变量，如行动价值和相对价值，将通过以下方式量化： 将动物的操纵杆运动行为与这两个区域的神经活动关联起来（目标1）。这将揭示 mPFC和DMS如何在正在进行的运动中计算决策变量。目标2将揭示具体的 mPFC和DMS之间的神经回路; mPFC第5层锥体束神经元的扰动将揭示 基于价值的决策变量如何在皮质纹状体通路中表现，以及通过使用转基因 与逆向刺激配对的小鼠系，将鉴定DMS中的直接和间接通路神经元， 确定每种细胞类型在代表价值和将价值转化为行动方面的作用。该项目总体将 更好地了解决策过程中价值如何归因于行动，以及 将确定皮质纹状体通路中的特定细胞如何计算和转换有关价值的信息， 做出最佳选择。

PROJECT SUMMARY To flexibly execute behavior, choices are made based on previous outcomes that will maximize reward. Crucially, learning the value of each action to obtain a reward is thought to drive this decision making process. In a value-based decision making framework, these values are first computed and then used to select and execute actions. Dysfunction in this decision making process is evident in many neuropsychiatric disorders including addiction and in patients with frontal cortical damage who show an inability to flexibly adjust or adapt their behavior. The corticostriatal pathway, from the medial prefrontal cortex (mPFC) to its downstream target, dorsomedial striatum (DMS), is a candidate circuit implicated in promoting flexible behavior. Neural correlates of value have been found in mPFC across species, and DMS is traditionally thought to be involved in action selection, where representations of action values have been found in DMS neurons. However, it is unclear exactly how representations of value are then transformed into actual movements for adaptive behavior. Similarly, many studies that have described representations of value in either region have done so using discrete binary measures of behavior and correlate neural activity only to the action's endpoint (e.g. emitting a lick at a left or right spout, or making a saccade to a target). This project will investigate the transformation of value into actions in mice by investigating how mPFC and DMS neural activity represent value during ongoing movements (Aim 1), and will reveal the specific cell types and pathways within the corticostriatal pathway that are necessary for value representation and action selection (Aim 2). In both aims, action potentials from mPFC and DMS will be recorded as thirsty head-fixed mice perform in a dynamic foraging task that requires a joystick movement to search for the high-probability reward, where the location of the high-probability reward changes over time. Value-based decision variables, such as action values and relative value, will be quantified by correlating the animal's joystick movement behavior to neural activity in both areas (Aim 1). This will reveal how mPFC and DMS compute decision variables during ongoing movements. Aim 2 will reveal the specific neural circuitry between mPFC and DMS; perturbations of mPFC layer 5 pyramidal tract neurons will reveal how value-based decision variables are represented in the corticostriatal pathway, and by using transgenic mouse lines paired with antidromic stimulation, direct and indirect pathway neurons in DMS will be identified to determine each cell type's role in representing value and transforming value into action. This project overall will provide a better understanding of how values are attributed to actions during the decision making process, and will identify how specific cells in the corticostriatal pathway compute and transform information about value to make optimal choices.