Semantic Surprise in Music Listening: Validating a Short EEG Test of the N400 ERP
Abstract
The N400 is a well-established phenomenon in EEG research that reflects semantic (i.e. meaning) violations, thus, this Event Related Potential (ERP) is measured when the progression or pairing of stimuli do not match expectations. A most famous example of a sentence that elicits an N400 is “I take my coffee with cream and dog” [1]. Previous research suggests that the N400 also exists in music listening [2], but this conclusion is based mostly on mixed/cross-module long-lasting paradigms, e.g. a combination of linguistic and music stimuli with a duration of around 2 hours [3]. Due to its link to semantic memory, the N400 has potential to be used as a measure of memory, and even intervention effectiveness, in memory patient groups such as Alzheimer patients. However, the extended length of previous N400 music-paradigms limits their use in these patient groups. The question thus arises whether similar results can be obtained during a condensed music-only paradigm.
To shorten the duration while still introducing enough violations to reliably measure the N400 in a participant, the newly-designed paradigm tested in this project consisted of simple musical stimuli with multiple in-key violations. In-key violations are notes that alter the melody of the stimuli, but that do not feel wrong in a harmonic sense because they match the key of the song. 40 participants were recruited using advertising and snowballing methods. During the study, the participants listened to 34 short piano stimuli while an EEG recording was made. 17 songs were melodies of known folk songs, and 17 were new compositions. The 17 folksongs were selected in an earlier survey where Austrian people were asked whether they knew a certain folk song or not. The new songs were composed in such a way that they mimicked the harmonic tendencies of folk songs. Each of the 34 stimuli incorporated multiple in-key violations of musical expectations in terms of familiarity for the known songs and melodic expectations for the unknown songs. The total duration of the experiment was only 20 minutes. At the time of writing, data collection is just ending, so the final results will not be available yet at the conference. Eventually, the results of this study will contribute to the development of less time intensive research tools that are particularly well-suited for clinical settings.
References
[1] M. Kutas and S. A. Hillyard, “Brain potentials during reading reflect word expectancy and semantic association,” Nature, vol. 307, no. 5947, pp. 161–163, Jan. 1984, doi: https://doi.org/10.1038/307161a0.
[2] R. A. Miranda and M. T. Ullman, “Double dissociation between rules and memory in music: An event-related potential study,” NeuroImage, vol. 38, no. 2, pp. 331–345, Nov. 2007. doi:10.1016/j.neuroimage.2007.07.034
[3] N. Calma-Roddin and J. E. Drury, “Music, language, and the N400: ERP interference patterns across cognitive domains,” Scientific Reports, vol. 10, no. 1, Jul. 2020. doi:10.1038/s41598-020-66732-0