Acquisition and Generalization of Pitch Probability Profiles
Collett, J. Meghan
MetadataShow full item record
Krumhansl (1990) has proposed that our sense of tonality is based, in part, on the perception and internal representation of the hierarchies of pitch class salience in music. It has further been proposed that regularities in pitch patterns may be acquired through statistical learning. To further explore this proposal, we conducted two experiments in which musically untrained participants were exposed to tone sequences generated from one of two pitch profiles: Lydian or Hypophrygian. Tone sequences were randomly generated from event frequency profiles computed by Huron and Veltman (2006), with frequencies converted to probability of occurrence. Exposure trials consisted of 100 sequences generated from one mode for half the participants and from the other mode for the remaining participants. Sequences generated from the unexposed mode appeared in test trials only. Following the exposure trials, testing involved pairing exposed and unexposed tone sequences at each of three levels of distinctiveness. Versions of the tone sequences were constructed to be more or less distinctive following an algorithm described by Smith & Schmuckler (2004). In Experiment 1, participants were asked to record which pair member they preferred and in Experiment 2, participants were asked to record which pair member was more familiar. In both experiments, both groups received the same test pairs. Results of Experiment 1 indicated no preference for any tone sequence type. However, results of Experiment 2 revealed participants had acquired knowledge of the exposed pitch distribution, and were able to generalize to the more distinctive level. The findings support those of Loui, Wessel, and Hudson Kam (2010) in terms of a dissociation between recognition and preference. We suggest this may be due to methodology, stimulus-type and participant strategy. The findings also support Krumhansl (1990), as salient pitches appear to be important in the recognition of pitch probability profiles.