Some robust higher-level percepts for music.

We explored a range of higher-level percepts in music. Participants were asked to make two-alternative forced-choice judgments of extracts of instrumental music on various dipole categories, such as happy/sad or male/female. The consistency with which each stimulus was judged on a response category across listeners provides an indication of the extent to which the musical percept can be mapped reliably onto that dimension. High consistency would suggest that the response category is related to one of the natural perceptual dimensions for music. We found very high consistency (90% +) for various response categories normally used as descriptions of people (such as male/female and happy/sad). Other types of response category gave much lower consistency. Perhaps our participants are experts in making fine distinctions in person-related categories for almost any stimulus. We tested this with a control experiment where foodstuffs replaced the musical stimuli. We did not find high agreement for person-related categories. The differences between responses to music and food were statistically highly significant.