Neural and Computational Architecture for Complex Navigation and Subjective Self-Location

项目摘要 候选人和职业目标：我打算成为一家研究第一的学术机构的独立科学家， 跨越描述层次(即，从计算到神经元)的桥梁，并加深我们对 大脑如何推断我们周围的世界，以及我们自己在其中。我接受了认知科学家的培训(获胜 世界认知科学最佳论文奖)，重点是理解我们的感觉 自我定位；“我”在空间中的位置。我现在正在接受系统神经科学方面的培训，开发专业知识 在大规模啮齿动物神经生理学中，专注于自我定位的动态方面；空间 导航。这些体验是相辅相成的；从行为计算到单位计算，以及从 静态到动态自我定位。环境和职业发展计划：我的导师是多拉·安吉拉基博士 (纽约大学，航海专业知识)，大卫·施奈德博士共同指导(纽约大学，啮齿动物自我生成行动) 和克里斯蒂娜·萨文(纽约大学数据科学专业)。此外，我是国际大脑实验室的科学家成员，允许 我有机会利用在啮齿动物神经生理学方面的世界级专业知识(22个实验室)。我在大赛期间的训练 K99将专注于对连续复杂的自然主义过程中的行为和神经元进行基于模型的分析 任务，以及实验室管理技能(即人员、拨款申请、沟通)。研究计划：空间 导航是适应行为的核心，是我们权衡利用当前位置的能力的基础 随着对新奇事物的探索。精美的作品详细介绍了许多空间代码(例如，地点和格网 细胞)，然而我们(1)缺乏一个规范的框架来解释 自然导航，(2)不理解来自海马体结构的空间代码如何与皮质相互作用， 以及(3)专注于理解我们如何建立我们周围世界的内部模型，而忽略了 它的起点是我们自己。在奖项的K99阶段，我将开发一个自然导航任务 要求啮齿动物消除复杂信号歧义的虚拟现实。这些动物将被训练成 按顺序将通过视网膜的运动(即，光流)得到的速度信号积分到位置估计中 将路径整合到潜在目标的位置。然后，他们将在一种新的情况下进行测试，这种情况下 流动可能是由自身和/或目标运动引起的。我希望动物的行为符合贝叶斯因果关系 推断(BCI)-一种典型的计算，其中在小但不大的信号中会出现估计偏差 差异(即观察员在不正确的内部模式下运作时)。此外，我将广泛地映射神经 通过利用新的大规模神经生理学技术，在脑机接口期间在啮齿类动物的大脑中进行活动。 在颁奖的R00阶段，我将直接操纵自我定位的主观感觉--最初的 航行条件-通过K99期间开发的任务在啮齿类动物中测量这一现象。 除了将BCI建立为指导自然航海的基本计算外，我还期待拟议的 为自闭症和精神分裂症等推理障碍的下一代疗法提供信息的项目。

Project Summary Candidate and Career Goals: I intend to become an independent scientist at a research-first academic institution, bridging across levels of description (i.e., from computations to neurons) and furthering our understanding of how the brain infers the world around us, as well as ourselves within it. I trained as a cognitive scientist (winning the Glushko Prize for best dissertation in Cognitive Science, worldwide), with a focus on understanding our sense of self-location; where am “I” located in space. I am now training in systems neuroscience, developing expertise in large-scale rodent neurophysiology and with a focus on the dynamic aspects of self-location; spatial navigation. These experiences complement each other; from behavioral computations to single-units, and from static to dynamic self-location. Environment and Career Development Plan: I am mentored by Dr. Dora Angelaki (NYU, expertise in navigation) and co-mentored by Drs. David Schneider (NYU, rodent self-generated actions) and Cristina Savin (NYU, data science). Further, I am a scientist member of the International Brain Lab, allowing me the opportunity to leverage world-class expertise (22 labs) in rodent neurophysiology. My training during the K99 will focus on model-based analyses of behavior and neurons during continuous and complex naturalistic tasks, as well as lab management skills (i.e., personnel, grant-writing, communication). Research Plan: Spatial navigation is central to adaptive behavior, underlying our ability to trade-off the exploitation of our current location with the exploration of novel ones. Beautiful work has detailed a number of spatial codes (e.g., place and grid cells) in the hippocampal formation, yet we (1) lack a normative framework accounting for the complexities of natural navigation, (2) do not understand how spatial codes from the hippocampal formation interact with cortex, and (3) have focused on understanding how we build internal models of the world around us, while neglecting its starting point – ourselves. During the K99 phase of the award, I will develop a naturalistic navigation task in virtual reality where rodents will be required to disambiguate complex signals. These animals will be trained to integrate velocity signals derived from motion across their retina (i.e., optic flow) into a position estimate, in order to path integrate to the location of a latent target. Then, they will be tested in a novel situation, one where optic flow may be caused by self- and/or target-motion. I expect animals to behave in line with Bayesian Causal Inference (BCI) – a canonical computation wherein estimation biases emerge during small, but not large, signal disparities (i.e., when observers operate under the incorrect internal model). Further, I will broadly map neural activity throughout the rodent’s brain during BCI by leveraging novel large-scale neurophysiology techniques. During the R00 phase of the award, I will directly manipulate the subjective sense of self-location – the initial condition for navigation - and measure this phenomenology in rodents via the task developed during the K99. Beyond establishing BCI as a fundamental computation guiding naturalistic navigation, I expect the proposed project to inform next-generation therapeutics for disorders of inference, such as Autism and Schizophrenia.