Physiologically constrained computational modeling of visual word recognition

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): The research outlined in this proposal aims to use a methodologically novel combination of computational modeling and electrophysiology to produce a model of reading that simulates not just the behavioral signifiers that comprehension of text has taken place -- such as lexical decision -- but the actual brain processes that instantiate that comprehension itself -- such as the N400 ERP component or activation of the anterior temporal lobe. While many excellent models of reading already exist, not one of them is able to simulate any physiological results, in first part because it is difficult to reduce functional neuroimaging data sets down to parameters that are appropriate for computational simulation, and in second part because creating a model which incorporates neuroimaging data requires interdisciplinary expertise in both computational modeling and neuroimaging. The proposed research plan provides traction on both of these difficulties; the first by outlining a procedure for describing rich electrophysiological data in a manner tractable for computational simulation (e.g., mean amplitude over a time window), the second by bringing together Carnegie Mellon's excellence in computational modeling with the University of Illinois' noted strength in psychophysiological research. Reading is such an automatic and effortless skill in fluent readers that its complexity, and the serious consequences of its impairment, are often only noticed in dysfunction. Dyslexia can be acquired through brain damage or developed during education, and despite its prevalence s still not well understood. This research aims at developing a model of reading based on recordings from reading brains-a model which can be used as the basis for targeted treatments of both developed and acquired dyslexia.