AUTHOR=Miles Scott A. , Rosen David S. , Grzywacz Norberto M. 

TITLE=A Statistical Analysis of the Relationship between Harmonic Surprise and Preference in Popular Music

JOURNAL=Frontiers in Human Neuroscience

VOLUME=Volume 11 - 2017

YEAR=2017

URL=https://www.frontiersin.org/journals/human-neuroscience/articles/10.3389/fnhum.2017.00263

DOI=10.3389/fnhum.2017.00263

ISSN=1662-5161

ABSTRACT=Introduction

	Although music has been said to have no apparent direct survival value (Pinker, 1997), it can elicit pleasurable reward responses in the human brain like those associated with food or sex (Blood & Zatorre, 1999; Blood & Zatorre, 2001; Salimpoor, et al., 2011; Salimpoor et al., 2013). Recent neuroimaging work has revealed many of the mechanisms recruited by the brain during the perception of reward from listening to music that is preferred (Pereira, et al., 2011; Salimpoor, et al., 2013). Less is known, however, about what structural aspects of music are required to drive these mechanisms. A prevailing theory of this process states that adherence to or deviation from listeners’ expectations leads to emotional reward (Meyer, 1956; Huron, 2006). One challenge in testing this theory of how music structure leads to reward is to quantify expectations. Another challenge is to test the theory in an ecologically valid way, that is, avoiding the artificial stimuli that are often used in laboratory tests, which can feature isolated chords (e.g., Koelsch, et al., 2001) or a novel musical system (Loui & Wessel, 2008). 

	The framework of information theory has proved to be useful in quantifying expectations and deviations from them in naturalistic music. Through information theory, one can investigate expectations based on statistical measures of representative corpora. Such an approach was proposed by Leonard Meyer (1957), and has been expanded upon by other researchers in subsequent years (e.g., Knopoff & Hutchinson, 1981; Temperley, 2007; Agres, et al., 2013; for a review, see Rohrmeier & Koelsch, 2012). While he was cautious about overstating how much statistical analysis could reveal, Meyer claimed that knowledge about deviation from expectations, examined through probability, could help reveal their relationship to emotion, a primary source of meaning in music (Meyer, 1957). So far, however, information-theory-driven corpus analyses have primarily been aimed at describing aspects of music, with less emphasis on investigating any relationship these aspects might have with how the music is perceived. We set out to extend such analyses to determine the capacity of aspects within popular music to evoke a reward response. In a first step