That may be so, for the present, but in the meantime it is both valid and constructive to discuss the possibility of an alternative approach to perceptual modeling, even if we are not yet ready to define the system more explicitly. Even in its currently vague form, this model offers solutions to certain fundamental problems that are not available with alternative approaches. Of the conventional view, specifically the models of Marr and Biederman, it can also be said that without a simulation we just don't know whether their proposed idea works. And yet those models serve a useful purpose of expressing a general concept of visual processing, which I challenge with an alternative general concept.