Natural categorization through multiple feature learning in pigeons

Recently (Troje, Huber, Loidolt, Aust, & Fieder 1999), we found that pigeon s discriminated between large sets of photorealistic frontal images of huma n faces on the basis of sex. This ability was predominantly based on inform ation contained in the visual texture of those images rather than in their configural properties. The pigeons could learn the distinction even when di fferences of shape and average intensity were completely removed. Here, we proved more specifically the pigeons' flexibility and efficiency to utilize the class-distinguishing information contained in complex natural classes. First, we used principal component as well as discriminant function analys is in order to determine which aspects of the male and female images could support successful categorization. We then conducted various tests involvin g systematic transformations and reduction of the feature content to examin e whether or not the pigeons' categorization behaviour comes under the cont rol of category-level feature dimensions-that is, those stimulus aspects th at most accurately divide the stimulus classes into the experimenter-define d categories of "Male'' and "Female''. Enhanced classification ability in t he presence of impoverished test faces that varied only along one of the fi rst three principal components provided evidence that the pigeons used thes e class-distinguishing stimulus aspects as a basis for generalization to ne w instances.