ABSTRACT
I think it is evident that we don’t all mean quite the same thing by Prägnanz. It can mean something quite different from the old classical Gestalt brain-field idea. Let me suggest an alternative formulation. On one hand, I believe in mental images. I believe that the system in which they exist may legitimately be called an analogue system. Roger Shepard presents some evidence to that effect in Chapter 10. I also believe that the nervous system describes the images—that is, that people who talk in terms of propositional representation are essentially right as well. In fact, I would like to hypothesize a Prägnanz principle that is based on a two-way communication between an analogue image system and a languagelike or propositional system to allow hill climbing toward simple representations. Suppose that what the system likes is short descriptions and that the 418image is progressively changed, within the constraints of the input, until its description is minimized. This way of looking at the matter, which is considerably different from the classical Gestalt point of view, has the advantage of taking into account not only intrinsic stimulus properties—that is, redundancy, uniformity, or homogeneity in the stimulus itself—but also schemata corresponding to familiar objects. If an input can be brought into conformity with a well-formed schema that is frequently used and to which a short symbol has been assigned, it might be described quite as economically as if it were intrinsically simple.
