Cognitive control mechanisms in human prefrontal cortex

 描述（由申请人提供）：人类大脑具有灵活分配认知资源以满足任务需求的卓越能力。认知控制机制必须保持对当前情境（任务需求）、先前选择的结果以及解决冲突和适应即将到来的行为选择的控制选项的清晰表示。这些手术中的任何一种缺陷都可能导致神经精神功能障碍。 迄今为止的研究确定了一个区域网络，包括dACC，对认知控制至关重要。然而，认知控制的关键方面，包括网络组件如何相互作用，以及它们如何表示和管理冲突，仍然存在争议。在技术层面上，人类和猴子研究中使用的方法之间的巨大差距引起了许多不确定性，包括认知控制机制的物种间差异的可能性。我们确定并瞄准了两个主要的进展障碍：1）缺乏一个强大的，但具体的概念框架，可以整合不同的实证研究结果; 2）缺乏高质量的人类数据，从单个神经元尖峰到群体活动。 我们开始与一个综合的概念模型。最近，控制期望值（EVC）理论被提出来解释dACC在认知控制中的作用。根据该模型，dACC权衡成功完成控制要求高的任务的预期收益与这样做所需的成本。基于此计算，它生成一个信号，指定要执行哪个（如果有的话）任务，以及执行多少 控件分配给任务。具体来说，EVC模型预测dACC应该执行开头提到的关键功能：当前上下文监控、结果监控和控制信号规范。我们测试这些功能的方法，有效地弥合了猴子单个神经元活动和人类非侵入性群体测量之间的差距。 我们提出了一系列的实验，使用功能成像和颅内电生理方法在人类。受试者为计划植入颅内电极监测癫痫发作的难治性癫痫患者。在电极植入之前，他们将接受高分辨率的功能磁共振成像。当它们被植入时， 执行行为任务，旨在测试关于EVC理论预测的dACC作用的假设。我们将同时收集单单位和LFP记录从dACC和LFPs从横向PFC。我们的广泛目标是验证一个全面的认知控制理论，使用多模态人类记录。我们通过测试与上述3个建议的dACC功能一致的3个特定目标来实现。在具体目标1中，我们阐明了dACC的当前上下文监控功能。我们测试的假设，dACC编码纯粹的冲突信号，这些神经元相位锁定θ节律，以协调dACC与其他控制相关区域的通信。在具体目标2中，我们定义了dACC的结果监测功能。我们测试假设，人类dACC神经元编码奖励预测错误（RPE），这些神经元夹带θ振荡产生错误和反馈相关的θ信号，可以观察到中额头皮脑电图。使用信息传递分析，我们进一步假设，RPE信息从dACC到横向PFC使用θ范围振荡。在具体目标3中，我们为dACC的控制规范功能建立了证据。我们测试假设调查控制信号的目标和强度规格。 许多神经精神障碍可归因于无法将适当的价值归因于情境刺激，关注相关特征而忽略不相关的特征，以及基于反馈依次增加或减少行为的奖励偶然性。该过程的障碍的实例包括情绪/焦虑障碍（强迫症、抑郁症、慢性疼痛的情感方面）、成瘾、注意力缺陷障碍和精神病。更清楚地了解人类认知控制的神经生理学对于成功理解和治疗这些行为障碍至关重要。

 DESCRIPTION (provided by applicant): The human brain has a remarkable ability to flexibly allocate cognitive resources to meet task demands. Cognitive control machinery must maintain clear representation of current context (task demands), outcomes of prior choices, and control options for resolving conflict and adapting upcoming behavioral choices. Deficiencies in any of these operations can contribute to neuropsychiatric dysfunction. Research to date identifies a network of areas, including the dACC, as essential to cognitive control. However, key aspects of cognitive control, including how network components interact, and how they represent and manage conflict, remain controversial. On a technical level, the wide gap between methods used in human and monkey studies raises a host of uncertainties, including the possibility of substantial interspecies differences in cognitive control machinery. We identify and target two major impediments to progress: 1) lack of a robust, yet specific conceptual framework that can integrate disparate empirical findings; and 2) lack of high quality human data spanning the scale from single neuron spikes to population activity. We begin with an integrative conceptual model. The expected value of control (EVC) theory has recently been advanced to explain the role of the dACC in cognitive control. According to this model, the dACC weighs the expected benefit of successfully completing a control-demanding task against the cost required to do so. Based on this calculation, it generates a signal that specifies which (if any) task to perform, and how much control to allocate to the task. Specifically, the EVC model predicts that the dACC should perform the key functions mentioned at the outset: current context monitoring, outcome monitoring, and control signal specification. We test these functions with methods that effectively bridge the gap between single neuron activity in monkeys and noninvasive population measures in humans. We propose a series of experiments using functional imaging and intracranial electrophysiology methods in humans. Subjects will be patients with medically intractable epilepsy scheduled for intracranial electrode implantation for seizure monitoring. Prior to electrode implantation, they will undergo high-resolution fMRI. While implanted, they will perform behavioral tasks designed to test hypotheses regarding the role of the dACC as predicted by EVC theory. We will collect simultaneous single-unit and LFP recordings from dACC and LFPs from lateral PFC. Our broad goal is to validate a comprehensive theory of cognitive control using multi-modality human recordings. We do so by testing 3 Specific Aims aligned with the 3 proposed dACC functions mentioned above. In Specific Aim 1, we clarify the current context monitoring function of the dACC. We test hypotheses that the dACC encodes pure conflict signals, and that these neurons phase-lock to theta rhythms to coordinate dACC communication with other control-related regions. In Specific Aim 2, we define the outcome monitoring function of the dACC. We test hypotheses that human dACC neurons encode reward prediction errors (RPEs) and that these neurons entrain to theta oscillations to generate error- and feedback-related theta signals that can be observed on mid-frontal scalp EEG. Using information transfer analyses, we further hypothesize that RPE information travels from dACC to lateral PFC using theta-range oscillations. In Specific Aim 3, we establish evidence for a control specification function of the dACC. We test hypotheses investigating control signal targeting and intensity specification. Many neuropsychiatric disorders are attributable to the inability to ascribe appropriate value to contextual stimuli, attend to relevant features while ignoring irrelevant ones, and sequentially increase or decrease reward contingencies of actions based on feedback. Examples of disorders of this process include mood/anxiety disorders (OCD, depression, affective aspects of chronic pain), addiction, attention deficit disorders, and psychoses. A clearer appreciation of the neurophysiology of human cognitive control will be essential for the successful understanding and treatment of these behavioral disorders.