(A) study on complexity measure of rhythmic pattern and human preference in evolutionary art

Abstract

In visual art domain in evolutionary art(EA) field, imageries are created by complexity measures that can induce human preference. However, in temporal art domain i,e. music, few researches have been studied about complexity measure that is relevant to human preference. Instead, only a study on relationship between human’s cognitive ability and preference is existed. Therefore we measure the cognitive complexity values for generating human-preferred novel rhythmic patterns. We focus on measuring complexity of the the patterns with three criteria based on a ’cognitive complexity’ theory by Thul and Toussaint: perceptual complexity, performance complexity, and structural complexity. In order to do this we first extracted preferred rhythmic patterns from each group by using interactive genetic algorithm(IGA). Next, we calculated each complexity of the selected patterns. With this measurements we use the average complexity scores of the pattern to generate novel 8 rhythms. For assessing these values are effective for creation, we compared preference score between ’novel’ and the other rhythmic pattern sets - novel(ps), novel(lhl), novel(met), novel(odd), and existed - by conducting t-test. In addition, in order to find out which complexity factors should be considered for inducing public preference, we conduct comparative analysis between complexity values of preferred patterns from two groups: experts and non-experts. The result first shows that novel pattern sets has a higher preference score than that of the rests. However, there is no significant difference in preference for novel and each novel(lhl), novel(met), and existed. Moreover, there are significant differences in lhl-value and metrical value for patterns preferred by experts and non-experts group. Thus, the primary implication of this is that the quantitative measurement of performance complexity and metrical complexity are important factors in human preference and can be useful to generate novel rhythmic pattern.