We propose a novel technique for identifying evoked components by using prior information of the stimulus time course as a guiding factor, allowing for modeling of trial variability, in a regularized optimization framework.