Visual Categorization in Pigeons

In general the main focus of this project was the investigation of categorization abilities in pigeons (Columba livia) and to improve our understanding of complex visual stimulus control in these birds. We took our investigations in two directions. Experiments of the sensoric/physiological approach were devoted to the investigation of the role and saliency of motion as an important descriptor of natural classes or objects. In particular we were interested in finding out whether or not motion has a facilitating effect on shape perception. an aspect, which has been completely neglected in our previous experiments on classification of complex static stimuli, The results suggest that pigeons - in contrast to humans - have difficulty to use the extra structural information provided by the rigid transformational motion of the object (a human head). It seems likely that under the specific circumstances of this experiment they saw the movies in a stroboscopic manner and failed to integrate the multiple views into the representation of a 3D-object. Additionally we presented real 3D objects (models of the human heads). By using pointlight-displays of moving subjects (humans and pigeons) we also started to examine the role of biological motion. In cooperation with the Ruhr-University Bochum we investigated asymmetric processing of sensoric information in the pigeon brain. The second line of investigations that we followed was the cognitive/associative approach These experiments were devoted to the study of the associative mechanisms of pigeon categorization. They focused on the role of the category structure on the pigeons classification strategy. With checkerboard patterns as stimuli we tested the hypothesis that the ratio of inter/intra-class distance between classes could be an important dimension along which classification behavior is controlled. A series of experiments with synthetic human faces as stimuli showed that pigeons are able to abstract both item- and category-specific information about a stimulus. Another experiment with the same kind of stimuli showed that the judgment of the sex of a human face does not show the typical characteristics of categorical perception but is determined by perceptual similarity alone, involving shading and brightness information. Furthermore we developed and tested a novel multi-stimulus, multiple-matching learning paradigm for pigeons, which, we believe, not only simulates their foraging behavior in natural environments, but also more realistically tests their categorization, including a touch screen panel placed in a learning chamber. After having solved some initial technical problems concerning the touch screen technique the multiple matching method proved to be superior to the conventional successive go-nogo procedure concerning the speed of learning as well as the discrimination performance.