Modeling cognition with a human memory inspired advanced neural controller

Abstract

In this paper is studied how the imitation of the structures and the processes of memory can possibly makes cognition arise in a computational model. More precisely, the combination of a perceptron and an associative memory leads to build a scalable behavioral controller expected to reveal intelligent behaviors. This approach differs from traditional behavioral animation hybrid architectures[1], in which the agent knowledge is a collection of modeller-defined symbolic objects or frarnes[2] and its behavior a set of scripts or automatons[3]. To our concern, this prevents the agent from adaptiveness in dynamic environments. The advanced neural network proposed below is the association of two networks. The vertical "procedural" network is a traditional perceptron that binds perception to action for achieving the agent's action selection. Classical gradient methods such as the delta rule can be used to train this network. The horizontal "semantic" layer is an associative network that builds inner representations from assembling perceptions together into patterns. The unsupervised Hebbian rule[4] is used for that purpose. Simple examples can reveal how the network takes advantage of inner representations to outperform the behaviors traditionally obtained with classical perceptrons. © Springer-Verlag Berlin Heidelberg 2006.