Abstract

Scientists argue for their claims using many kinds of representations: text, mathematical formulas, and figures such as diagrams, graphs, charts, electron micrographs, etc. The prevalence of visual representations suggests that they play some important role in science that is different from that of linguistic representations. But the foundational questions about the presence of visual representations in scientific arguments have not yet been addressed. What do figures represent? Can they bear truth? How do they support inferences? My dissertation develops answers to these questions, and uses them to explain how figures contribute to the expression and justification of scientific knowledge. ;Since philosophers think of arguments as consisting of statements or mathematical representations, the first task is to show that figures are symbols that can participate in arguments. Goodman's analysis of symbol systems provides semantic and syntactic criteria to compare and contrast different types of representations---visual as well as serial representations like text and mathematical expressions. The visual symbol systems used in science are characterized by systematic form-content relationships, which can be used to define Tarski-style concepts of truth. This is evidence that visual representations can bear truth, supporting my contention that they are genuine components of arguments. ;Analysis of figures in terms their form and symbol system properties shows how visual representations play particular roles in arguments, such as providing evidential support. Figures are also used to express conclusions: models of biological structures and mechanisms often appear in the form of diagrams. I show why the visual format is especially effective in representing the kind of explanation Cummins identified as functional analysis, and why linguistic formats are better for other types of explanations. ;While analysis in terms of symbol systems is essential to understanding the function of visual representations in scientific arguments, this method does not resolve every issue about the use of figures in science. As a final step, I relate current models of visual perception and cognition to their epistemic roles. The result is a coherent picture of the roles figures play in establishing scientific knowledge claims, and how those roles are adapted to human cognitive abilities