ABSTRACT
This chapter focuses on the problem of quantifying and representing the information in human faces. It presents a quantifiable theory of the perceptual information in faces and proposes a simple statistical/neural network model to simulate the learning of this information. The chapter outlines a sample of approaches that have been used for specifying the information in faces. It provides a brief definition of the auto associative neural network model and shows that how faces are described as a weighted sum of the eigenvectors extracted from the auto associative matrix. The chapter demonstrates that the computational model is capable of solving some useful face processing tasks. It reviews some studies suggesting its potential psychological relevance and discusses the relationship of a perceptual learning representation to the approaches discussed in the representational issues. The chapter also discusses the relationship between model and human recognition performance considering the application of the model to psychological issues.
