Research interests
I study human cognitive processes, especially language processing, using a combination of computational modeling and experimentation. Most of my research can be characterized as applying the paradigm of rational analysis to problems in cognition. In rational analysis, one precisely articulates the goals and the cognitive and physical constraints relevant to a problem and develops an explicit computational model of optimal behavior under those conditions. These optimal models typically involve methods from machine learning, statistics, artificial intelligence, natural language processing, and information theory. I then use a variety of behavioral methodologies to test the predictions of these models on human performance. To the extent that human behavior matches the behavior of the optimal model, such modeling provides us with a way of understanding human behavior as an optimal response to the structure of the problem. Recent areas I've worked on with this paradigm include eye movements in reading, difficulty in human sentence processing, and language learning.