Magnetoencephalography and functional MRI: a quantitative study of speech perception

Abstract

Magnetoencephalography (MEG), sensitive to evoked neuronal activity, and functional magnetic resonance imaging (fMRI), sensitive to hemodynamic consequences of cortical activation, were compared in normal subjects, presented with simple speech elements (vowels). The subjects were instructed either to listen passively or to perform phonetic or pitch discrimination tasks. Sources of activation detected using the two technologies were found to co-localize in the superior temporal gyrus. Furthermore quantification of the evoked responses, using both techniques, revealed a task-induced bilateral increase in activation, with the greatest increases observed in the left hemisphere, during performance of the phonetic discrimination task. It is concluded that not only do fMRI and MEG indicate similar underlying cortical activation areas, but that they offer quantitative indices of hemispheric lateralization induced by performance of a linguistic task. Their co-variance with respect to hemisphere and task performance suggests that despite their different neurophysiological sensitivities (hemodynamic changes vs. neuronal electrical activity), both techniques offer quantitative approaches to the study of speech perception, and by extension, to the processing of language.