Continuum neural network based on aggregate field of local nonlinear coupling

Neural networks are usually considered as arrays of interconnected nonlinear gain elements with linear couplings. Some newer concepts indicate that biological neural networks may be considered to be linear gain units interconnected with nonlinear couplings. A neural network model proposed by B. Baird (1989) and having these characteristics is extended to a neural network behaving as a continuum or as a neural medium. This type of neural network has characteristics that make it ideal for implementation in a coherent optical system. The behavior of this neural network may be reproduced with an optical resonator containing a nonlinear optical crystal. Information can be made to correspond the resonator's modes. Depending on various parameters, it can exhibit winner-take-all dynamics or chaos.<>