Comparative analysis for category goodness-rating vs. discrimination sensitivity of musical rhythm patterns using Signal Detection Theory

George Papadelis

Aristotle University of Thessaloniki, Greece

A basic concept within signal detection theory (Green & Swets, 1966; Macmillan & Creelman, 1991) is the metaphor of using discrimination statistic measures as a form of ‘perceptual distance’ metrics. That approach has been broadly applied to the categorical perception research of various sensory continua, especially to those related to speech and phoneme perception. The majority of experimental works to the perception of musical - rhythm related continua, which have been concentrated to the problem of the categorical vs. non-categorical nature of perceiving musical rhythm patterns, have adopted the classical experimental approach of correlating identification and discrimination performance in terms of percentage correct responses. In addition, a few of them have further been concentrated to a more elaborate description of the structural aspects for certain types of rhythm categories, using various kinds of analyses and data transformations to map the corresponding ‘rhythm spaces’ at a perceptual level. The study described here is mainly focused on various methodological issues for measuring ‘perceptual distances’ for musical rhythm patterns using detection theory. Our investigation is based on a subsequent analysis of previously collected data as a part of an ongoing research which aims at a development of a prototypical experimental design to study certain aspects of rhythm categorization behavior and meter induction at a cross-cultural basis. Calculations of ‘perceptual distances’ along a continuum of micro temporal transformations of a rhythm pattern are presented, both in terms of category goodness - ratings and discrimination sensitivity.