The proposed research aims to identify sources of reading-related sentence comprehension difficulties that are most subject to individual differences, and to study their cognitive and neural underpinnings. The novelty of this project is to bring together the knowledge base on reading differences and advanced techniques for studying on-line sentence processing, and to target a largely neglected population in their late teens and early twenties who represent a wide range of reading abilities. Earlier work has shown that deficient phonological processing skills lead to slow or inaccurate recognition of the individual words, which is the hallmark of reading disability. Behavioral studies and supporting functional neuroimaging studies by the Haskins group have contributed toward understanding the cognitive and neural bases of word reading and their dysfunction in reading-disabled individuals. We hypothesize that insufficient decoding skill at the word level creates a bottleneck that prevents working memory from being used efficiently to support integrative processes in reading phrases and sentences. Alternatively, reading difficulties may stem from a learning deficit that broadly impairs the learning and retention of linguistic sequences. Predictions generated by each hypothesis will be tested in five coordinated experiments that employ a three-pronged methodology: 1) eye movement recording that tracks on-line sentence processing; 2) event-related functional neuroimaging that reveals brain activity in word reading and sentence comprehension; and 3) computational modeling that predicts behavioral and neural patterns. The experiments will address four classes of phenomena that have been found to affect the ease or difficulty of understanding connected material: detection of anomaly, resolution of ambiguity, memory demands on complex sentences, and learning of new word sequences and set phrases (idioms). Multiple analyses will be conducted to explore relationships between reading behavior and brain activation patterns
Showing the most recent 10 out of 26 publications
