Decoding Single Pitch Imagery from the Human EEG

Abstract

Pitch is one of the main attributes of sound, an ingredient in musical melodies, a major cue to distinguish between male and female or child and adult voices, and one of the important factors of tonal languages like Mandarin Chinese. However, the perception of pitch is not the same in all people, due to factors such as music education as well as innate factors. Nevertheless, most people have an inherent ability to imagine pitch. Will the difference in perception depending on the effect of musical training have a correlate in neural activity? Is it possible to decode a pitch using the neural signals that are created when people try to imagine it? To examine this issue, participants wore EEG equipment and listened to or imagined seven notes (C4 - B4) in the key of C major. The group that was musically trained could more accurately discriminate between different pitches, and their ERP patterns showed significant pitch-to-pitch gaps in the bilateral frontal and temporal areas compared to the non-trained group. This group also showed the best decoding accuracy (47.86%), which is well above the level of chance (14.28%), on a 7-class pitch classification task. The significance of this study is to propose a neural representation of a music training effect on perception and to demonstrate the possibility of decoding pitch imagery from human EEG signals.