FROM GEONS TO STRUCTURE - A NOTE ON OBJECT REPRESENTATION

Two models of object perception are compared: recognition by component s (RBC), proposed by Biederman, and structural information theory (SIT ), initially proposed by Leeuwenberg. According to RBC a complex objec t is decomposed into predefined elementary objects, called geons. Acco rding to SIT, the decomposition is guided by regularities in the objec t. It is assumed that the simplest of all possible interpretations of any object is perceptually preferred. The comparison deals with two as pects of the models. One is the representation of simple objects-vario us defintions of object axes are considered. It is shown that the more these definitions account for object regularity and thus the more the y agree with SIT, the better the object representations predict object classification. Another topic concerns assumptions underlying the mod els: the identification of geons is mediated by cues which are suppose d to be invariant under varying viewpoints of objects. It is argued th at such cues are not based on this invariance but on the regularity of actual objects. The latter conclusion is in line with SIT. An advanta ge of RBC, however, is that it deals with the perceptual process from stimulus to interpretation, whereas SIT merely concerns the outcome of the process, not the process itself.