Collaborative Research: Flexibility and Robustness of attack and evasion: reverse-engineering the mechanisms of behavioral control

动物如何在新的环境中产生有效的行为是行为科学长期以来的谜团之一。例如，网球运动员每次回发球时，条件与运动员过去经历的条件不同：球以稍微不同的速度，或以不同的角度，或以不同的旋转行进。然而，神经系统能够在这些新的条件下产生有效的反应。这怎么可能？该项目将通过将新的数学方法与新的实验系统相结合来解决这个问题，该系统使用实时计算机视觉来跟踪动物解决顺序决策任务。作为这项研究更广泛影响的一部分，研究人员将与本科教育计划合作，旨在为传统上代表性不足的STEM学生提供研究机会。跨学科的学生团队将在开发和执行与较大项目的主题相关的小型研究项目时得到指导，其目标是为基础科学专业以及工程和数学专业提供一个进入尖端跨学科科学的门户。此外，研究人员将在指导佛罗里达大学长期惠特尼实验室REU计划（31年）的参与者时纳入这项研究。这项工作的结果将揭示动物如何产生如此多样化的行为以在新的情况下生存，并将导致新的问题和方法，可用于理解脊椎动物神经系统的功能和动物承担的一些最重要任务中行为控制的起源。与长期以来作为动物决策模型的二元选择任务不同，更复杂的感觉-运动行为，例如躲避捕食者或捕获猎物，通常涉及响应于感觉刺激的动态流而做出的一系列决定。这种行为的一个关键要求是它们是健壮的。例如，逃离捕食者所需的确切决策链在不同的环境中会有所不同，但动物必须产生一系列独特的适合手头情况的反应。这突出了一个关于动物行为的长期问题：在一个环境中学习或进化的行为如何推广到动物可能遇到的大量可能情况？该项目使用数学建模和高分辨率闭环实验的组合来解决这个问题，使用虹鳟鱼（Oncorhynchus mykiss）的猎物攻击和捕食者逃避行为作为研究动物如何产生灵活，强大的行为序列的模型。特别是，研究人员将测试新兴的假设，这些强大的，更高层次的行为反应是通过一种称为行为增益控制的机制实现的。研究将探索行为增益控制如何参与在正确的时间启动行为序列，平衡多个竞争目标，协调控制沿着多个行为维度（例如，加速、转弯），并在不断变化的环境条件下保持性能。这项工作有可能揭示复杂的行为是如何在生态和进化相关的任务的背景下产生的新的光。这个奖项反映了NSF的法定使命，并已被认为是值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估的支持。

How animals generate effective behavior in new situations is one of the longstanding mysteries of behavioral science. For example, every time a tennis player returns a serve, conditions are different from those the player has experienced in the past: the ball is travelling at a slightly different speed, or at a different angle, or with a different rotation. Yet, the nervous system is able to generate a response that is effective under these novel conditions. How is this possible? This project will address this question by combining new mathematical methods with a novel experimental system that uses real-time computer vision to track animals as they solve sequential decision tasks. As part of the broader impacts of the study, the investigators will partner with an undergraduate education program intended to introduce research opportunities to traditionally underrepresented STEM students. Interdisciplinary teams of students will be mentored in developing and executing small research projects related to the theme of the larger project with the goal providing a gateway for both basic science majors, and also engineering and mathematics majors into cutting-edge, interdisciplinary science. In addition, the investigators will incorporate this research when mentoring participants in the University of Florida's longstanding Whitney Lab REU program (31 years). The results of this work will shed light on how animals generate such a diverse range of behaviors to survive in novel situations, and will lead to new questions and approaches that can be used to understand the function of the vertebrate nervous system and the origins of behavioral control in some of the most important tasks animals undertake.Unlike binary choice tasks, which have long served as the model for animal decision-making, more complex sensory-motor behaviors such as avoiding predators or capturing prey often involve sequences of decisions made in response to dynamic streams of sensory stimuli. A key requirement of such behaviors is that they be robust. For example, the exact chain of decisions required to escape a predator will differ from one setting to another, yet an animal must generate a sequence of responses uniquely suited to the situation at hand. This highlights a perennial question about animal behavior: how can behaviors that are learned or evolved in one context generalize to the enormous set of possible situations an animal might encounter? This project attacks this question using a combination of mathematical modeling and high-resolution, closed loop experiments using the prey attack and predator evasion behaviors of rainbow trout (Oncorhynchus mykiss) as a model for studying how animals generate flexible, robust behavioral sequences. In particular, the investigators will test the emerging hypothesis that these robust, higher-level behavioral responses are achieved through a mechanism called behavioral gain control. Research will explore how behavioral gain control is involved in initiating behavioral sequences at the right time, balancing multiple competing objectives, coordinating control along multiple behavioral dimensions (e.g., acceleration, turning), and maintaining performance across changing environmental conditions. This work has the potential to shed new light on how complex behaviors are generated in the context of ecologically and evolutionarily relevant tasks.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.