Investigating PCA Based Techniques for Objectively Measuring the Impact of the Feldenkrais Method on Pianists' Coordination Characteristics

Abstract

This thesis addresses the need for an objective means of measuring pianists' unique coordination characteristics that can examine movement relationships from many variables distributed throughout the body simultaneously during the performance of complex, bimanual piano playing tasks. Developing such a method would permit objective study of the effect of somatic methods, such as the Feldenkrais Method® (FM), on pianists' movement in future research. Lack of objective measurements in this field is related in part to the difficulty of measuring whole-body coordination during complex movements, like piano playing. Measuring coordination characteristics during complex movements is a critical challenge facing movement researchers. This thesis addresses this problem by exploring PCA-based (principal component analysis) approaches for identifying task-specific and pianist-specific coordination characteristics in the motion capture data of six advanced pianists performing a battery of twelve contrasting pianistic tasks at three motion capture sessions separated by one-week intervals. Each article of the thesis addresses a different aspect of this research problem. Articles one and two contextualize the problem by offering an analysis of the limitations of existing approaches for measuring posture and movement for assessing Feldenkrais outcomes relating to pianists' coordination. Article three assesses the limitations of standard PCA approaches for studying pianists' coordination characteristics. Article four proposes a novel framework for categorizing the different sources of task-determined and participant-determined variation layered in motion capture data to aid in the development of PCA procedures that target variation related to individual's coordination choices. Article five builds on the findings of the previous studies, presenting a new, PCA-based approach for identifying pianist-specific coordination characteristics that we call Functional Subspace Identification. Functional subspace identification compares inter-PC angles between pairs of PCs (principal components) from separate PCAs on subsets of the motion capture data to identify invariant, one-dimensional PC subspaces. The results show that functional subspace identification is an effective means of identifying task-specific variation characteristics that are common to all pianists, as well as variation characteristics unique to individual pianists. The findings of this thesis contribute to the study of coordination characteristics during complex movements in both musicians and non-musicians.