Project 1: Source localization
Source localization in EEG/ MEG is an ill-posed problem due to the unknown number of sources and finite number of detectors. There are other issues such as volume conduction effects which attenuate the signal. So, most of the algorithms assume some sort of relationships among source networks either implicitly or explicitly, e. g., minimum correlation, maximum coherence, statistical independence. We are working on novel algorithms which has more biophysically motivated constraints for source localization. This project is being conducted using neurodynamical models of EEG generation in conjunction with actual EEG during steady-state auditory response. We plan to validate these tools with MEG in parallel.
Project 2: Cognitive Subtraction
The functional neuroanatomy of cognitive processes is generally derived by subtraction of a control task from an experimental task. This imposes the somewhat unrealistic constraint that neuronal networks related to different cognitive components are independent of each other. We are attempting to develop novel design-analysis techniques for fMRI to address this issue by taking into cognizance the segregation and integration of information among candidate brain networks.
Our attempt is to determine the brain networks involved in the integration of multiple senses to form complex percepts. Our hypothesis is such integration of senses happen over a network of simultaneously "active" brain areas and how they evolve over time at millisecond resolution. We plan to conduct behavioral, EEG, fMRI and MEG on different multisensory paradigms to challenge our hypothesis. Our current project takes the McGurk effect as an window to study the integration of auditory and visual information. We are studying how various psychophysical variables induce change in perception and how such is correlated with underlying brain network dynamics.