I am a graduate student in the Computer Science Dept. and the Center for the Neural Basis of Cognition at CMU. I come from Portugal and, after a long and tortuous path, ended up in Pittsburgh, where I spend most of my time reading in cafes and hoping for a bright idea for a thesis proposal. Needless to say, I still don't have one... My research aims at bridging the fine grain knowledge about neurophysiology and high level cognitive models, using data from functional Magnetic Resonance Imaging (fMRI) studies of live, functioning humans. We try to say more than just where activity is located, which is the more traditional use for this data. Examples would be predicting what a subject will do next in a task, identifying causal links between different brain regions, or even dissecting exactly what the subject is thinking about. Ultimately, this should allow us to build better cognitive models, perhaps semi-automatically. In addition, I have an interest in the question of how large populations of neurons, considered as stochastic and non-linear devices, can be used to encode and process information. The brain obviously succeeds at using populations to perform functions reliably, or do massively parallel computations such as extraction of features in a visual scene. I'd like to understand the behaviours that emerge from a population of neurons, and what algorithms can be implemented given such behaviours and operations.