What would a computer look like if we took our current design principles -- separation of processing and memory / small (64-bit) registers / deterministic algorithms -- and did the opposite? What would a philosophy of mind look like if we reversed our fundamental assumptions about rules and representations, types, and categories? In this talk I will present a novel approach to computing and cognition, called Vector Symbolic Architecture (VSA), that takes this contrarian approach. In developing VSA over the past decade, my colleagues and I have have been struck by the extent to which it is also the kind of model of mental activity that we arrive at if we take seriously Ludwig Wittgenstein’s critiques of the philosophy of language and mind. (We note, for example, that the uniform nature of representation in VSA eliminates the sort of problems that early Wittgenstein criticized in Russell’s theory of types and related formalisms.) In this talk I will present several tasks that VSA can handle in a better or more biologically plausible way than traditional computational systems, including analogical reasoning, rule induction, and behavior-based robotics. The talk should be of interest to students and researchers in computer science, engineering, philosophy, neuroscience, and mathematics.
Simon D. Levy is an Associate Professor of Computer Science at Washington and Lee University, where he has also served as head of the Computer Science Department. He holds a bachelor's degree in linguistics from Yale University, a master's degree in linguistics from the University of Connecticut, and a Ph.D. in Computer Science from Brandeis University. His current work includes the material he will present in this talk, as well as the development of open-source software for robotics, parallel computing, geology, and other fields, and most recently, the use of miniature aerial vehicles for indoor surveillance and mapping.