Gradient symbolic structure in cognition
Dr. Matt Goldrick
Department of Linguistics
Northwestern University
Abstract: Across many domains, human behavior exhibits rich variation, suggesting that gradient, stochastic process play a key role in cognition. But to what extent does gradience extend to the building blocks of cognition--representations themselves? I'll review recent experimental data from speech errors that suggest gradience pervades all aspects of mental representations. To account for such data, I'll introduce the Gradient Symbolic Computation (GSC) framework (developed in collaboration with Paul Smolensky and colleagues at Johns Hopkins). This framework integrates the structured, systematic and combinatorial representational structures that form the backbone of symbolic theories with gradient, dynamic computation. I'll discuss how GSC not only provides insight into gradient experimental data but also potential new perspectives on the structure of linguistic knowledge in monolingual and bilingual speakers.
Research Background: Research in our group uses a variety of methods to examine the cognitive processes underlying our ability to produce and comprehend speech. In particular, we focus on (a) the production of speech and (b) our knowledge of sound structure. Methodologies include: On-line language processing, Psycholinguistics, Cognitive neuropsychology, Acoustic phonetics, Functional neuroimaging and electrophysiology, Mathematical and computational models of language processing, Generative phonology and Optimality Theory and Harmonic
