This paper aims to design an artificial neural network to discover the impression by recognizing the expression of the human face. To achieve this goal, the artificial neural network was analyzed and to create patterns of the database containing a set of images with different expressions. The learning process of the network was also conducted through patterns training. The extent to which patterns of online training were recognized was compared to the true values of expressions. The grid was trained in 200 patterns and the anomalies were removed. Then re-learned the network again and analyzed the network performance by comparing the real expression with the expected expression and outputting the error for the network appearing. Impression recognition in the grid applied a three-layer back propagation model, with an average error of 0.321. The performance of the artificial neural network in the recognition of impressions was 80%