The ability to learn during sleep is a popular topic since the second half of the nineteen-century. Although various studies claimed that learning during sleep was possible, a systematic and detailed inspection of these studies shows that experiments with positive results were actually biased and lacked a controlled condition. A recent study however brought more compelling evidence that sleeping subjects can learn new basic associations between sounds and odours (more precisely between sounds and the olfactory respiratory response), and that this association persists until the next morning. Whether learning information from pure auditory stimulation and stimulus-stimulus type is possible remains an open question. The aim of this project is to test the possibility of creating new complex associations between stimuli using magnetoencephalography (MEG) as an online marker of learning.