Feedback stimuli are fundamental components in Brain-Computer Interfaces. It is known that the presentation of feedback stimuli elicits certain brain potentials that can be measured and classified. As stimuli can be given through different sensory modalities, it is important to understand the effects of different types of feedback on brain responses and their impact on classification. This paper presents a protocol used to obtain brain potentials elicited by visual, auditive or vibrotactile feedback stimuli. Experiments were carried out with five different subjects for each modality. Four different single-trial classification strategies were compared, according to the information used to train the classifier, achieving a classification rate of approximately 80% for each modality.