A Brain-Like Computational Model Based on a Shared Memory

Recently, scientists have found great similarities between human brains and those of primates. We propose a brain-like computational model after significantly studying the problem solving systems of the behavioral control in the primate brain. Instead of working on the specific roles of tens of billions of neurons and how they might be connected with each other, we mainly focus on the accumulative and overall performance of the brain subsystems, each of which is modeled by a computational system or a certain algorithm. We propose a four-level computational system, where the four levels of the model are independent with each other, at the same time, one level is also connected to the others. The connectivity among levels has been studied as well. The four parallel subsystems are different in many ways, such as the sensory inputs, computational power, model complexity, and so on. However, all systems share a same working memory which stores past experience in all states and helps to make better decision at a near future. The model has been simulated in a challenging foraging problem. The experimental results show that the model works well on various environments with spares or dense obstacles.