By extending the bidirectional associative memory neural network model, a mathematical model of multidirectional associative memory(MAM) neural networks with constant time delays is proposed. By using Brouwer fixed point theorem and the upper right Dini derivative, a sufficient condition for the existence and the global exponential stability of an equilibrium point is obtained. And for a special MAM neural network which connection weights is positive, a sufficient and necessary condition for the existence and the global exponential stability of an equilibrium point is obtained. The results are...
