Director, Brain and Machine Laboratory
Ph.D., University of Illinois, Urbana-Champaign
Ruth L. Kirschstein Postdoctoral Fellow, Carnegie Mellon University
Area: Cognitive and Brain Sciences
Office: Science IV, Room 104
Lab: Brain and Machine Lab
Curriculum vitae (212kb)
Member of the Society for Psychophysiological Research, the Cognitive Neuroscience Society, The Cognitive Science Society, and Women in Cognitive Science
Computational modeling and human electrophysiology, language comprehension, reading development
In the Brain and Machine Laboratory, we use human electrophysiology and computational modeling to study how reading is learned by children and how it works in adults. We interweave our physiological and computational work in a symbiotic fashion, such that we interpret our electrophysiological data in the context of our computational models, while our models are constrained by the reality of neurophysiology. For example, our computational modeling of the processes which link print with meaning takes into account several important computational properties of neurons--such as the separation of excitation and inhibition, and the differing time constants of different populations of inhibitory neurons-- in service of producing simulations which continuously mirror the dynamics of the neural systems implementing reading. This work contrasts with more traditional reading models in that it strives to simulate reading processes as the occur, not after they have concluded-- as is the case with models that simulate only behavioral reaction time or accuracy. Meanwhile, we use the Event-Related Potential (ERP) technique to collect and analyze massive data sets appropriate for computational modeling, such as the single-item ERP corpus, which includes stable ERPs representing the neural response to single words.
Philosophy of Graduate Training:
Because of the small magnitude of the signals involved, human electrophysiology is a precise science. Graduate students in my lab are given the training to approach psychophysiology with best practices, and to understand those practices at both a pragmatic and theoretical level. In contrast, computational modeling often seems to be more art then science, and graduate students with computational interests are encouraged to be intimately familiar with the fundamentals of computational psychology so that they have a foundation from which to work creatively with their models.
Students are assigned to projects which, as much as possible, are pertinent to their interests while still having the maximum likelihood of successful publication. This process is meant to help students understand that selecting projects that are likely to succeed is beneficial both to their confidence and reputation as scientists as well as their success in exiting graduate school in a timely fashion.