Recently proposed approach to recognitize facial expressions have been proposed Jager with the so called Echo State Neural Network (ESNN). The ESSN approach assumes a sort of "block box" operability of the network and clients a broad applicability to several different problems using same principle, here we proposes a simplified version of ESNN which we call a simple echo state network which exhibits good results in memory capacity and facial matching and recognition which allows a better understating of the capability and restriction of ESNN. ESSN gives promising result when the input are distorted. Simulation results show that a proposed system (ESNN) achieves a excellent performance with high training and recognition speed. © Medwell Journals, 2008. © 2013 Elsevier B.V., All rights reserved.