The paper aims at recognizing the various human facial expressions. Every countenance is marked by changes in the feature points of the face. These feature points are located in various regions of the face. There are two phases in the facial expression recognition technique described here. The first phase is image processing and the second phase is setting up and training of the neural network. The image processing phase or the pre-processing phase involves a number of steps. In the first step the image of the human face is normalized. The normalized image is subjected to a grayscale transformation in the second step. The third step is marked by partitioning of the transformed image in two portions: the upper half and the lower half. A frequency analysis of the normalized image is performed in the upper half partition as the fourth step. This tracks the position of the eyes and the eyebrows. Contouring is performed to trace out the shape of the eyes and the eyebrows. A further frequency analysis in the lower half reveals information about the nose, the mouth and the cheeks as the fifth step of the first phase. The contours are vectored to obtain a feature vector. In the next phase this feature vector is used for training the Hopfield neural network.