Singular-continuous nowhere-differentiable attractors in neural systems

We present a neural model for a singular-continuous nowhere-differentiable (SCND) attractors. This model shows various characteristics originated in attractor's nowhere-differentiability, in spite of a differentiable dynamical system. SCND attractors are still unfamiliar in the neural network studies and have not yet been observed in both artificial and biological neural systems. With numerical calculations of various kinds of statistical quantities in artificial neural network, dynamical characters of SCND attractors are strongly suggested to be observed also in neural systems experiments. We also present possible information processings with these attractors. (C) 1998 Elsevier Science Ltd. All rights reserved.