The Role of Featural and Configural Information in Face Classification A Simulation of the Expertise Hypothesis

Face recognition in adults is the product of a unique mechanism in the brain and it is based on years of experience. The goal of this paper is to analyze the role of configural and featural information for face classification and to compare the performance of Bayesian network classifiers with the human performance in three experiments: similarity matching, gender, and race classification. Our results show that despite the fact that the machine classification results are worse than the one of the humans, they are consistent with human classification results