Neural Network Models of Cognitive Maps

摘要: Neural networks - parallel systems consisting of highly interconnected simple neuron-like processing elements have become a subject intense interest to scientists spanning broad range disciplines including psychology, physics, mathematics, computer science, biology and neurobiology. A variety connectionist or neural network models been proposed, ranging from simplified more realistic learning, associative memory, sensori-motor development. hold great promise in the field spatial cognition since they are capable (in principle) approximating any real valued function mapping, used solve complex problems allied fields such as visual pattern analysis robotic control. In addition, biological relevance, problem that has plagued symbol Artificial Intelligence (AI) systems. simulated based on known physiological anatomical properties brain may reveal process by which groups neurons interacting according some local rales undergo self-organization. This paper will examine specific where produced plausible behavior. helped understanding types computations might be performed place cells hippocampus when animal moves about constructs an internal representation. Other recognizing places locating goals further demonstrate success this domain. model route learning is proposed can learn different routes environment locate goal given information.