Abstract

Scientific inquiry makes use of visual models to represent empirical systems. Many philosophers claim that models function only as analogies, and that their role is limited to a didactic or heuristic role. I analyze four families of visual models: maps, graphical images from the study of turbulence, fractal images, and strange attractors. I show that these models are projected from data; that they are dynamic pictures, which can be manipulated by the researcher; that they are necessary for the theories in which they function; and finally that they interact at many levels with the theory. I call models possessing these qualities generative, because they have contributed to the elaboration and formulation of theories. Strange attractors, visual models produced in the study of dynamical systems, are the best example of generative models. ;I show the limitation of conceiving of models as analogies, and restrict the scope of concepts such as similarity, resemblance, correspondence, and analogy. The study of generative visual models shows that visualization takes us beyond the perceptively visible, but that the representational function of the model is retained. Visual models represent the world, even if they do not resemble or look like the world. The objects of scientific inquiry are represented through our constructed worldview. This position sides with a constructivist account of reality