Neuro-Robotic Models of Learning and Addiction

描述：(申请人提供) 这项研究计划的总体目标是提供一个现实的计算模型，说明神经调节系统如何作用于人脑的皮质和皮质下网络，以及它们的行为如何影响现实世界中的正常行为模式和成瘾行为模式。为了(A)发现和实施神经调节功能的解剖学和生理学原理，(B)研究它们与其他大脑区域和过程的整合，特别是那些潜在的学习和记忆，以及(C)研究(内部)神经事件和(外部)行为之间的相互作用，将寻求一种综合的系统级计算方法。在日益复杂的阶段，将在自主机器人中设计和实现由感觉和运动皮质区域、皮质下回路和神经调制核组成的详细神经网络模型。该模型将结合现实的解剖学和生理学特性，并能够根据实际的感觉体验和行为在连接性方面进行可塑性变化。在第一阶段，我们计划实施一个神经调节系统，具有类似于中脑多巴胺系统的特性，产生与奖励和奖励预测刺激相关的神经反应。除了这个奖励系统，我们还将实施一个单独的系统，对厌恶刺激做出反应，并调查它们之间可能的功能相互作用模式。为了调查成瘾行为的发展与记忆和学习相关的过程之间的假设联系，我们将扩展该模型，以包括额外的皮质和皮质下网络。我们的建模研究将使我们能够分析神经结构的不同组成部分、药理和生理特性、学习和记忆以及实际行为在从正常和受控的行为模式向成瘾转变过程中所起的因果作用。一个全面和详细的具体化(机器人)模型有可能成为一种独特的解释性和预测性工具，有助于未来对药物滥用和成瘾的实证研究。

DESCRIPTION: (provided by applicant) The overall aim of this research program is to provide a realistic computational model of how neuromodulatory systems act upon cortical and subcortical networks of the human brain, and of how their actions influence normal and addictive modes of behavior in the real world. An integrated systems-level computational approach will be pursued in order to (a) discover and implement anatomical and physiological principles underlying neuromodulatory functions, (b) study their integration with other brain areas and processes, especially those underlying learning and memory, and (c) study the interactions between (internal) neural events and (external) behaviors. In stages of increasing complexity, a detailed neuronal network model of sensory and motor cortical areas, subcortical circuits and neuromodulatory nuclei will be designed and implemented in an autonomous robot. The model will incorporate realistic anatomical and physiological properties and be capable of plastic changes in connectivity depending upon actual sensory experience and behavior. In a first stage, we plan to implement a neuromodulatory system with properties similar to a midbrain dopamine system, producing neural responses that are related to reward and reward predicting stimuli. In addition to this reward system we will implement a separate system responsive to aversive stimuli and investigate possible modes of functional interaction between them. In order to investigate the hypothesized connection between the development of addictive behavior and processes related to memory and learning we will expand the model to include additional cortical and subcortical networks. Our modeling studies will allow us to provide an analysis of the causal roles played by different components of the neural architecture, of pharmacological and physiological properties, of learning and memory and of actual behavior in the switch from normal and controlled modes of behavior to addiction. A comprehensive and detailed embodied (robot) model has the potential of serving as a unique explanatory and predictive tool aiding in future empirical research on drug abuse and addiction.