Neural Networks for Adaptive Behavior

9309691奇尔为了生存和繁殖，动物必须追求特定的目标，同时不断改变自己的行为，以应对不断变化的内部和环境条件。在几个小时或几天的时间里，动物必须在不同的相互排斥的行为中做出选择，比如进食或交配。当它们在几分钟或几小时内从事一种特定的行为时，它们必须在不同的行为反应中进行选择，以便继续执行该行为(例如，拒绝一块不能吃的食物以继续进食)。它们还必须在几秒钟内迅速改变自己的行为反应，以应对环境的变化(例如，对难咬的食物咬得更厉害)。动物产生适应行为的能力背后的神经机制是什么？为了解决这个问题，首席研究员建议研究海洋软体动物海兔的行为开关。初步结果表明，随着环境条件的变化，海兔能够在不同的行为反应之间进行切换。具体地说，如果动物试图摄取不可食用的物质，它可能会从摄取反应切换到排斥反应。研究人员建议使用体内和体外神经细胞的细胞内记录(1)识别和表征导致这些运动神经元活动时间转换的运动前神经元和中间神经元的特性，以及(2)识别和表征触发这些不同活动模式的感觉神经元的特性。利用这些数据，他将(3)模拟和分析负责行为切换的神经网络。这三个特定的目标将使他能够测试关于行为开关的神经组织的不同假设：这是由于专用神经电路、分布式神经电路还是具有一些共享和一些不同神经元件的电路？它还将使他能够基于动力系统理论的概念，开发出对行为切换的更抽象的描述。

9309691 Chiel In order to survive and reproduce, animals must pursue specific goals while continuously modifying their behavior in response to changing internal and environmental conditions. Over periods of hours or days, animals must choose among different mutually exclusive behaviors, such as feeding or mating. As they engage in a particular behavior over minutes or hours, they must select among different behavioral responses in order to continue to perform that behavior (e.g., rejecting an inedible piece of food in order to continue feeding). They must also rapidly alter their behavioral responses within seconds to cope with changes in the environment (e.g., biting harder on tough food). What are the neural mechanisms underlying the ability of animals to generate adaptive behavior? In order to address this question, the principal investigator proposes to study a behavioral switch in the marine mollusc Aplysia californica. Preliminary results indicate that Aplysia is capable of switching among different behavioral responses as environmental conditions change. Specifically, the animal can switch from ingestive to rejection responses if it attempts to ingest inedible material. The investigator proposes to use intracellular recording from nerve cells in vivo and in vitro (1) to identify and characterize the properties of premotor neurons and interneurons that cause the switch in the timing of the activity of these motor neurons, as well as (2) to identify and characterize the properties of sensory neurons that trigger these different patterns of activity. Using these data, he will (3) simulate and analyze the neural network responsible for the behavioral switch. These three specific aims will allow him to test different hypotheses about the neural organization of a behavioral switch: is it due to dedicated neural circuits, to a distributed neural circuit, or to circuits that have some shared and some distinct neural elements? It will also allow him to develop a more abstract description of behavioral switching, based on concepts from dynamical systems theory.***