Data Reduction and Analysis

A low pass Butterworth filter was applied to the motion of each sensor with a cutoff of 10 Hz. Each movement was visually analyzed, using a display of index tip X, Y, and Z position and velocity vs. time to mark the index tip sensor’s initial and final positions. This process was done visually instead of using a velocity criterion for movement onset and termination due to small tremors associated with holding the arm in starting and ending positions. The starting position was marked a few milliseconds before movement was initiated and the movement was marked as complete when the hand position was constant. Movements that were not fully captured within the 5 second recording period were removed from the analysis. These missing data occurred in three subjects and ranged from 1 to 5 trials missing from among 14 trials for a particular target/condition.

Yaw, elevation and roll of the arm (humeral) and forearm segments (radius and ulna) were calculated as a series of ordered rotations about axes fixed to the segment from a standard upper limb configuration in which the arm was pointing straight forward from the subject. The plane of the upper limb and orientations of the arm (humerus) and forearm segments were defined from the XYZ positions of the acromion, humerus lateral epicondyle and styloid process of the ulna recorded from the electromagnetic sensors (Figure ​(Figure1D).1D). Referring to Figure ​Figure1D,1D, the yaw angle η was defined as a rotation of the segment about the vertical axis with 0° being directly forward from the subject and 90° being directly to the left of the subject (positive Y axis). The second rotation was about a horizontal axis (medial to lateral in the standard configuration) and defined elevation θ of the segment, which was measured as the angle of the segment from the vertical Z axis with 0° indicating that the arm segment is pointing straight down and 90° corresponding to a horizontal segment. The third rotation, at the shoulder only, was about the long axis of the humerus to define arm roll ζ (internal/external rotation). The angle of inclination of the plane of the upper limb ν defines the configuration of the arm in three dimensions. This angle is derived from the vector p, the cross product of the arm and forearm vectors, which is perpendicular to the plane of the arm. Once this vector is calculated, its angle relative to the XY plane (Figure ​(Figure1D)1D) is computed by the equation sinν = sinθsinζ. Put simply, ν is the angle between the vector perpendicular to the plane defined by the arm and forearm and the horizontal XY plane, the angle being 0 if the arm plane is vertical.

This measure of arm posture is the same as that used by Soechting et al. (1995) and provides a simple measurement of upper limb configuration, because for a given index fingertip location the inclination angle will be constant if arm and forearm yaw and elevation angles are the same, assuming that the wrist and finger have similar contribution to the movements. Previous studies have reported that wrist movement produced minimal variation in end point posture of the arm and forearm compared to the shoulder (Miller et al., 1992; Wang, 1999; Schot et al., 2010). We stress that it is the variation of this endpoint posture measure that is of most interest. The orientation of the arm plane measure we used distills several variables into one. However, whilst expedient, our measure of arm posture is not related one to one with its actual posture. We have used it, as did Soechting et al. (1995), to obtain a measure of the variability (SD) of arm posture.