We present a method for distinguishing two subtly different mental states, on the basis of the underlying brain activation measured with fMRI. The method uses a classifier to learn to distinguish between brain activation in a set of selected voxels (volume elements) during the processing of two types of sentences, namely ambiguous versus unambiguous sentences. The classifier is then used to distinguish the two states in untrained instances. The method can be generalized to accomplish knowledge discovery in cases where the contrasting brain activation profiles are not known a priori.