Saberi Moghadam Sohrab, Samsami Khodadad Farid, Khazaeinezhad Vahid
Faculty of Engineering Modern Technologies, Amol University of Special Modern Technologies, Amol, Iran.
Department of Physiology, Faculty of Biology and Medicine, University of Lausanne, Lausanne, Switzerland.
Basic Clin Neurosci. 2019 Sep-Oct;10(5):443-449. doi: 10.32598/bcn.9.10.395. Epub 2019 Sep 1.
One of the interesting topics in neuroscience is problem solving and decision-making. In this area, everything gets more complicated when events occur sequentially. One of the practical methods for handling the complexity of brain function is to create an empirical model. Model Predictive Control (MPC) is known as a powerful mathematical-based tool often used in industrial environments. We proposed an MPC and its algorithm as a part of the functionalities of the brain to improve the performance of the decision-making process.
We used a hybrid methodology whereby combining a powerful nonlinear control system tools and a modular fashion approach in computer science. Our hybrid approach employed the MPC and the Object-Oriented Modeling (OOM) respectively. Therefore, we could model the interaction between most important regions within the brain to simulate the decision-making process.
The employed methodology provided the capability to design an algorithm based on the cognitive functionalities of the PFC and Hippocampus. The developed algorithm applied for modulation of neural circuits between cortex and sub-cortex during a decision making process.
It is well known that the decision-making process results from communication between the prefrontal cortex (working memory) and hippocampus (long-term memory). However, there are other regions of the brain that play essential roles in making decisions, but their exact mechanisms of action still are unknown. In this study, we modeled those mechanisms with MPC. We showed that MPC controls the stream of data between prefrontal cortex and hippocampus in a closed-loop system to correct actions.
神经科学中一个有趣的话题是问题解决和决策。在这个领域，当事件按顺序发生时，一切都会变得更加复杂。处理大脑功能复杂性的一种实用方法是创建一个经验模型。模型预测控制（MPC）是一种强大的基于数学的工具，常用于工业环境。我们提出了一种MPC及其算法，作为大脑功能的一部分，以提高决策过程的性能。
我们使用了一种混合方法，将强大的非线性控制系统工具与计算机科学中的模块化方法相结合。我们的混合方法分别采用了MPC和面向对象建模（OOM）。因此，我们可以对大脑中最重要区域之间的相互作用进行建模，以模拟决策过程。
所采用的方法提供了基于前额叶皮质（PFC）和海马体认知功能设计算法的能力。所开发的算法应用于决策过程中皮质与皮质下神经回路的调制。
众所周知，决策过程源于前额叶皮质（工作记忆）和海马体（长期记忆）之间的通信。然而，大脑中还有其他区域在决策中起着重要作用，但其确切的作用机制仍然未知。在本研究中，我们用MPC对这些机制进行了建模。我们表明，MPC在闭环系统中控制前额叶皮质和海马体之间的数据流，以纠正行为。
