Large-scale recordings in Primate Prefrontal Cortex: Mechanisms of Value and Attention

前额叶皮层（PFC）对一系列高级认知功能至关重要，如注意力和决策- 制作。研究这些过程是困难的，因为它们是隐蔽的，动态的，并在控制之下。 而不是实验者。它们的测量传统上依赖于通过以下方式推断它们的存在： 行为最近的技术进步，特别是神经元数量的增加， 同时记录，提出了直接从神经活动解码认知过程的可能性。 在拟议的项目中，我们将利用最近开发的高密度，高沟道数，硅 探针（神经像素），可以产生超过20倍的神经元记录产量增加， 常规方法。我们将使用这些新的探针和解码算法来识别细胞和电路- 注意力控制和决策的灵长类PFC水平的机制。我们将在 实时地并且使用输出来控制刺激扰动或电微刺激的应用， 这将"闭合回路"并确定解码信号对于认知的必要性。最后，我们将研究 通过将解码器限制在特定的神经元亚群的贡献。 第一个目标是了解注意力控制是如何通过外侧前额叶皮层（LPFC）实现的。 我们将同时记录额叶眼野和脑皮层46区的大量神经元。 猴子执行选择性注意任务。神经元的特征在于层状位置， 感觉、运动和记忆相关的特性，并作为推定的锥体或中间神经元。的贡献 注意力控制的特定亚群将通过构建一个解码器来确定， 使用来自感兴趣的亚群的神经活动的当前注意力轨迹。反过来，解码器将被用于 以控制刺激扰动来评估解码信号对注意力的行为影响。在 第二个目标，我们将研究眶额皮质（OFC）神经元亚群对价值的贡献， 基于决策，包括不同皮层，细胞类型和OFC子区域的贡献。 另外，解码器的输出将被用来控制电微刺激的应用，以进行检查 选择行为是否会偏向于特定的选择方案。在最后一个目标中，我们将建立在 前两者的工作旨在确定价值和注意力如何相互作用。虽然有证据表明 注意力优先于高价值的刺激，其他证据表明，事实恰恰相反。结果 将用于优化OFC中的解码器，该解码器将输出当前在 考虑，以及LPFC中的解码器，其将输出内隐空间注意的当前位置。我们将 然后使两个解码器输出信号互相关以确定一个信号是否可靠地滞后或领先于另一个信号。 拟议的工作利用了三个PI的专业知识：摩尔是LPFC在注意力中作用的专家， 沃利斯认为OFC在决策中起作用，而谢诺伊则认为OFC在实时解码和闭环控制中起作用。

Prefrontal cortex (PFC) is critical for a range of high-level cognitive functions, such as attention and decision- making. Studying these processes is difficult, since they are covert, dynamic and under the control of the subject, rather than the experimenter. Their measurement traditionally relies on inferring their presence via behavior. Recent technical advances, particularly the increase in the number of neurons that can be simultaneously recorded, has raised the possibility of decoding cognitive processes directly from neural activity. In the proposed project, we will capitalize on the recent development of high-density, high-channel count, silicon probes (Neuropixels) that can produce a more than 20-fold increase in neuronal-recording yield over conventional methods. We will use these new probes and decoding algorithms to identify the cellular and circuit- level mechanisms of attentional control and decision-making in primate PFC. We will perform this decoding in real-time and use the output to control the application of stimulus perturbations or electrical microstimulation, which will ‘close the loop’ and determine the necessity of the decoded signals for cognition. Finally, we will study the contribution of specific subpopulations of neurons by restricting the decoder to those populations. The first aim focuses on understanding how attentional control is achieved by lateral prefrontal cortex (LPFC). We will record simultaneously from large populations of neurons within the frontal eye field and area 46 of monkeys performing a selective attention task. Neurons will be characterized by the laminar location, by their sensory, motor and memory-related properties, and as putative pyramidal or interneurons. The contribution of specific subpopulations to attentional control will be determined by building a decoder to report the animal’s current attentional locus using neural activity from the subpopulation of interest. In turn, the decoder will be used to control stimulus perturbations to assess the behavioral effect of the decoded signals on attention. In the second aim, we will examine the contribution of orbitofrontal cortex (OFC) neuronal subpopulations to value- based decision-making, including the contribution of different cortical layers, cell types, and OFC subregions. In addition, the decoder output will be used to control the application of electrical microstimulation to examine whether choice behavior can be biased towards a specific choice alternative. In the last Aim, we will build on the work from the first two aims to determine how value and attention interact. While some evidence suggests that attention is prioritized to stimuli with high value, other evidence suggests that the opposite is true. Results from Aims 1 and 2 will be used to optimize a decoder in OFC that will output the value of objects currently under consideration, and a decoder in LPFC that will output the current location of covert spatial attention. We will then cross-correlate both decoder output signals to determine whether one reliably lags or leads the other. The proposed work leverages the expertise of the three PIs: Moore is an expert in the role of LPFC in attention, Wallis in the role of OFC in decision-making, and Shenoy in real-time decoding and closed-loop control.