|
Abstract:
|
An important problem in neuroscience is characterizing how various brain regions represent information in their networks of neurons. These representations are studied using multielectrode arrays, electroencephalography, magnetoencephalography, and a range of imaging modalities, including functional magnetic resonance imaging. The statistical analysis of these high-dimensional and highly dynamic time series present interesting challenges for solving dynamic inverse problems. In this talk, we will summarize our recent work on the design and application of Bayesian signal processing algorithms to characterize the dynamics of brain activity. We will illustrate these methods using multielectrode recordings from neurons in the rat hippocampus while the animal executes learning tasks and EEG, MEG, and fMRI recordings from human subjects executing somatosensory and motor tasks.
|
