Localization of the cognitive activity in the brain is one of the major problems in neuroscience. Current techniques for neuro-imaging are based on functional Magnetic Resonance Imaging (fMRI), Positron Emission Tomography (PET), and Event Related Potential (ERP) recordings. The highest temporal resolution is achieved by ERP, which is crucial for temporal localization of activities. However, the spatial resolution of scalp topography for ERP is low. There is a severe limitation for the parametric inverse solution algorithms that they can only perform well for the temporally uncorrelated sources. In this study, a spatial decomposition method is proposed to separate the temporally correlated sources using their topographies prior to their localization.