Feature learning in pigeons

Categorization is certainly one of the fundamental achievements in the evolution of cognition. It can be seen as the nexus between perception and cognition, because it is the process by which we recognize objects and events as meaningful entitites of our visual world. In this project we want further explore the abilites of pigeons to categorize the visual world. Specifically, we will attempt to collect further empirical support for our recently proposed Modified Feature Theory of Categorization (Huber & Aust, 2003). This model has been developed on the basis of a decade`s research in our pigeon lab in the course of which we have found cumulative evidence for the pigeon`s quick and flexible use of category-specific features in the solution of complex, natural categorization tasks. Indirect support for this feature-based categorization comes from cognitive neuroscience in monkey and man, as well as from computational modelling of perception and cognition. Here, we want to take our attempts further to investigate whether feature learning, in contrast to the mere memorization of exemplars or the formation of abstract representations, is also involved in ultra-fast categorization, in view-point-invariant recognition, in the categorization of prototypical classes, and in 2D-3D transfer. Furthermore, we want to explore the role of class experience in a people/non-people categorization. The latter task was used almost 40 years ago as the hallmark of visual categorization in pigeons. Long-term experience gathered during a pigeon`s life has received only scant attention in the psychological laboratory. In this project we will not only investigate the role of long-term experience in an indirect manner by using artificial picture classes (fish/non- fish), as Herrnstein did many years ago, but also in a direct manner by controlling the pigeon`s visual experience with the targets (humans) of a later presented categorization task. The performance of those "constrained" pigeons on this task can then be compared with that of pigeons that had normal visual contact with humans, namely animal caretakers and experimenters.

In this project we further investigated the abilities of pigeons to categorize objects of their visual environment. Categorization is certainly one of the fundamental achievements in the evolution of cognition. It can be seen as the nexus between perception and cognition, because it is the process by which we recognize objects and events as meaningful entities of our visual world. In the course of this project we were able to find further empirical support in favor of our recently proposed `Modified Feature Theory` of visual categorization in pigeons (Huber & Aust, 2006). This model has been developed on the basis of more than a dozen years of research in our pigeon lab, in the course of which we have found cumulative evidence for the pigeons` quick and flexible use of category-specific features while solving complex, natural categorization tasks under laboratory conditions. With our present knowledge, it seems that pigeons even make their quick and flexible use of these features in a way that they actually recognize a "human" when being presented with a two-dimensional digitized picture showing a human person (i.e., a three-dimensional object) on the screen; they seem to be capable of some kind of representational insight. This project contributed also to the enhancement of our methodological tools. It allowed (i) developing an integrative, flexible computer program for running and managing experiments in the visual cognition laboratory ("CognitionLab"), (ii) a novel feeding device for pigeons ("GrainLifter"), and (iii) a special, worldwide unique aviary (itself now representing a new methodological device) that enables us to selectively control for the pigeons` visual experience. It is now possible to compare the performance of "ontogenetically deprived" pigeons with that of normally housed pigeons. Continuation of our lines of research will enable us to determine the role of experience in visual categorization.