ABSTRACT
The authors report a new sensorimotor phenomenon in which participants use hand-sensed kinesthetic information to compensate for rotational sensorimotor rearrangements. This compensation benefits from conscious awareness and is related to hand posture. The technique can reduce control inefficiency with some misalignments by as much as 64%. The results support Y. CitationGuiard's (1987) suggestion that in bimanual tasks one hand provides an operational frame of reference for the other hand as in a closed kinematic chain. Results with right-handed participants show that the right and left hands are equally effective at providing such a cue. A constant-angular-targeting-error model, similar to that used for hand movements by H. Cunningham and I. Vardi (1990) and for walking by S. K. Rushton, J. M. Harris, M. R. Lloyd, and J. P. Wann (1998), is used to model the trajectories of targeting hand movements demonstrating the phenomenon. The model provides a natural parameter of the error.
ACKNOWLEDGEMENTS
The authors thank Kenneth Cheung for writing the program to run the experiment and Joelle Schmidt-Ott, Aminah Perkins, Matthew Panos-Ellis, and Sarah Reidenbach for assistance in data collection and data processing. Dr. Frank Kooi also assisted in early pilot studies. This research was partially supported by a research grant from NASA Headquarters Code UL Space Human Factors Program. A preliminary version of some the material from Experiment 1 was previously reported in the Proceedings of the 46th Annual Meeting of the Human Factors and Ergonomics Society and as an abstract in the 2002 Psychonomic Society Meeting.
Notes
1. A kinematic chain consists of rigid links (members) that are connected with joints (kinematics pairs) allowing relative motion of the neighboring links. In an open chain, each link in the chain is connected to any other link by one and only one distinct path. In a closed chain, each link is connected to any other link by at least two distinct paths. An arm may be idealized as a three-link open kinematic chain consisting of the upper arm, forearm, and hand. When the left and right hands grip each other, they form a closed chain.
2. For a recent report on this topic with historical perspective, see CitationOnyancha and Kinsey (2007).
3. In the classic formulation of the speed–accuracy trade-off, the Fitts Index of Difficulty (ID) for a movement amplitude (A) and required precision (W) is calculated as ID = log2(2A/W), (CitationFitts, 1954).
4. Speed could be used for additional analyses but for the purposes of this article, we decided to not pursue it and focus our attention on normalized path length.
5. Because of the appearance of inhomogeneity of variance evident in the error bars of , we applied log transformations to the data to verify all statistically significant effects. Reported statistics are based on the untransformed data because the statistical conclusions were unaffected by the transformation.
6. We ignored approximately 7.5% of the very large path lengths, greater than three times the mean, for the regression analysis to reduce the distorting effects of these outliers.
7. The parameters for the velocity profile—a, k1 , k2 , and b—were estimated by a least-squares fit to cross-participants averaged velocity profile data for trajectory data. Time along the movement was expressed as a percentage of a complete movement. This technique amounts to normalizing all movement times so each completed movement takes 100 units of time. For our specific fits with speed expressed as cm/s, initial settings of the variables were (a) for the nondominant, moving hand, a = 12.60, k1 = 7.696, k2 = 1.998, and b = −0.2343 and (b) for the dominant moving hand, a = 12.60, k1 = 8.168, k2 = 1.998, and b = −0.2340.
8. In fact, we have tried repeating the main ANOVAs using targeting error as a dependent measure instead of normalized path length where possible, but the statistical conclusion were unchanged.