Understanding the mechanisms by which the human cerebral cortex produces human cognition is one of the grand challenges facing neuroscience. We are approaching this question in two different ways, both of which are heavily computational in nature.
First, we are attempting to extract the large-scale structure of the major functional cortical cognitive circuits by applying neuroinformatic methods to a very large database of functional neuroimaging data. We view this problem as similar to the decoding problem faced in genomic analyses and are using bioinformatic machine learning algorithms to extract the structure of functional brain circuits from the raw imaging data.
We are also studying the properties of very large-scale competitive learning networks to model the specific properties of specific areas of the primate cerebral cortex. These models appear to provide insights into the basic computational operations of the primate cortex in that their activity spontaneously parallels many previously reported experimental results.