Functional magnetic resonance imaging (fMRI) is important modality in neurosciences. It is possible to (indirectly) measure neuronal activity and detect and localize activity in the brain. A typical fMRI acquisition consists of a series of data volumes acquired while the subject was exposed to a number of controlled stimuli. The use of modern classifiers such as support vector machines (SVM) in the analysis of fMRI data is becoming increasingly important. The rationale behind these multi-variate methods is to recognize trends in the data which are "invisible" when looking at timecourses of single voxels. Some successful examples have been shown already to demonstrate the retonotopic organization in the visual cortex. The aim of this project is explore the use of SVM and to apply them to experimental data that we have available. A prototype software should be developed using Matlab. This project will be in collaboration with the Center for Biomedical Imaging (CIBM) and the University Hospital of Geneva.