Pitch and space maps of skilled cellists: accuracy, variability, and error correction.

Based on a newly developed method that combines finger position tracking and spectral analysis of the concurrent acoustic record, we studied the accuracy and variability of pitch performance in eight skilled cellists and the role of acoustic feedback in their performance. The tasks required shifting movements between pairs of notes and separated by various distances (pitch intervals) on a single string at the rate of 1 note/s. The same tasks were performed either using the bow, providing acoustic feedback, or without the bow. Overall, our subjects exhibited a high degree of accuracy in executing tasks when using the bow. When using the bow, two types of variability were observed: (1) trial-to-trial variability: in most subjects the mean fundamental frequency of a single nominal note was significantly different from trial to trial; and (2) within-trial variability. The within-trial variability includes two sub-types: (a) the pitch of a given note changed between notes within a 50-note trial; and (b) within a single note there were positional changes that we hypothesize are attempts by the performer to adjust the fundamental pitch within the note. When acoustic feedback was absent, note distributions were shifted, multimodal, and had large variability; error-correction movements within a single note also significantly decreased, indicating that the stability and precision of the motor map depends on constant re-calibration and updating by acoustic information. Our results suggest that a performer's intonation should not be viewed as a fixed entity implied by the score but as a sample from a statistical distribution.