Seminar: 10/26 - Matt Goldrick, Northwestern University
11:00 to 12:30 PM at: 5101 Tolman Hall
Gradient symbol processing: Integrating continuous and symbolic computation in cognitionSymbolic theories explain the highly structured, systematic, and productive nature of cognition by appealing to discrete, combinatorial mental representations. A large body of work has argued that such theories have difficulty capturing gradient aspects of cognition. I'll discuss recent work in my lab that has used acoustic variation in speech production to reveal the gradient structure of mental representations. I'll then discuss a theoretical framework, Gradient Symbol Processing, that integrates combinatorial symbolic computation with the gradient, continuous aspects of neural computation. A case study will examine how this framework captures gradient patterns in the phonetics of speech errors.