This paper investigates the role of unusualness ratings in predicting music preference. In addition, the variables complexity and familiarity were rated for five music stimuli covering a range of styles. Ninety-four participants were exposed to each stimulus ten times over a three-week period. The three variables were tested as predictors of preference using linear and quadratic curve-fitting procedures. A linear increasing relationship was observed for familiarity, and inverted-U relationships were observed for unusualness and complexity. These results are consistent with Berlyne’s inverted-U model, or a segment of the inverted-U in the case of familiarity. Unusualness was a good indicator of music preference, and explained more variance than complexity or familiarity. Furthermore, the two stimuli that scored highest in unusualness produced consistently low ratings of preference independent of exposure, which appears to be a hallmark of “extreme” music stimuli.
Improved Determination of<tex>$Q$</tex>-Factor and Resonant Frequency by a Quadratic Curve-Fitting Method