College of Liberal Arts & Sciences
Math Biology Seminar
Abstract:
Humans encounter many different stimuli from the environment; the ability to categorize similar stimuli is necessary for survival. In particular, language acquisition requires a form of auditory categorization. For example, speech sounds combine features which are grouped into categories to be generalized between speakers. We propose a neural network framework that combines plausible biological mechanisms and the theory of dynamic neural fields to model the categorization process. The model will simulate a task in which a listener is presented with pairs of tones belonging to 4 categories. In this talk we will provide a description of the auditory task as well as a detailed discussion of the equations governing the model. We will then examine preliminary implementations of the predictive model. Finally, we will discuss current challenges with the model as well as future directions to consider.