International Congress Series 1301 (2007) 188 – 191

www.ics-elsevier.com

How can we realize skillful and precise movement? Ken-ichi Morishige a,⁎, Rieko Osu b,c , Naoki Kamimura d , Hiroshi Iwasaki d , Hiroyuki Miyamoto e , Yasuhiro Wada d , Mitsuo Kawato c a

e

Department of Intelligent Systems Design Engineering, Toyama Prefectural University, Toyama, 5180 Kurokawa Imizu, Toyama 939-0398, Japan b National Institute of Information and Communication Technology, Japan c ATR Computational Neuroscience Laboratories, Kyoto, Japan d Nagaoka University of Technology, Niigata, Japan Department of Brain Science and Engineering, Graduate School of Life Science and Systems Engineering, Kyushu Institute of Technology, Fukuoka, Japan

Abstract. Although human motor variability inevitably exists because of neural noises, it is not clear how humans can effectively reduce this variability for task accuracy. We examined the ability of humans to compensate for this variability during task-constraint reaching movements. The positional variance and muscle activations of reaching movements with task-constraints were compared with those without task-constraints. Flexor and extensor muscles were co-activated for the movements with task-constraints, and the positional variance was decreased to obtain accuracy. The results indicated that subjects were able to regulate their muscle impedance and to modulate the variability to meet the task requirements. © 2007 Elsevier B.V. All rights reserved. Keywords: Impedance control; Task-constraint; Co-contraction

1. Introduction Humans are able to generate skillful and precise movements, such as threading a needle or pointing out a small target. To realize these dexterous movements, the central nervous system computes appropriate motor commands, regulates activations of arm muscles, and then generates actual movement. ⁎ Corresponding author. Tel.: +81 766 56 7500x314; fax: +81 766 56 8030. E-mail address: [email protected] (K. Morishige). 0531-5131/ © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.ics.2006.11.016

K. Morishige et al. / International Congress Series 1301 (2007) 188–191

189

However, these motor commands are corrupted by neural noises, whose standard deviation increases with the level of the motor commands [1,2]. Such signal-dependent noise plays an undesirable role through the muscle dynamics, leading to the variability in movements. Therefore, even after extensive training, movement variability continues to exist [3]. It is essential that the central nervous system has the ability to reduce the effect of variability to the task arising from neural noises. How can humans reduce the variability and realize skillful and precise movements? It has been suggested that the central nervous system co-activates flexor–extensor arm muscles and regulates the impedance to reduce the variability at the end of the movement. The muscle impedance and end-point variability were examined and the results indicate that there is a relationship between the increase of arm impedance and the reduction of end-point variability [4–8]. However, few studies directly address how the muscle impedance and motor variability change to obtain the accuracy required in the middle of the movement. This paper discusses how the central nervous system can reduce motor variability and realize accurate movements. 2. Impedance control improves movement accuracy In order to examine the ability to reduce motor variability, we observed hand trajectories and main muscle activations under two conditions. The first condition was a simple pointto-point movement task (Fig. 1A). The second condition was one with a task-constraint, which requires precision for going through the gate. The gate was allocated at the middle point between the start and end points (Fig. 1B). The muscle activations and movement variability were compared between the two conditions. Around the gate, the activations of all measured muscles with task-constraints had a tendency to be higher than the ones without task-constraints (Fig. 2A–F). Shoulder and elbow stiffness (which is typically measured in Newtons per meter) was estimated from the weighted summation of rectified EMG signals through the index for muscle co-contraction around a joint, as proposed by Osu et al. [9]. The stiffness began to increase before the gate,

Fig. 1. Experimental setups. We observed hand trajectories and muscle activations under two conditions: condition A was simple point-to-point movement task, and condition B was one with a task-constraint. The distance between the start and end points was 25 cm, and the gate, which constrained the task, was allocated at the middle point between the start and end points. The movement duration was set at 350 ± 35 [ms].

190

K. Morishige et al. / International Congress Series 1301 (2007) 188–191

Fig. 2. The comparison of the time course of EMG activity and stiffness under two conditions. A, B, and C show the results of the shoulder, biarticular, and elbow flexors. D, E, and F show the extensors. G and H show the shoulder and elbow stiffness of the typical subject, respectively. Each line represents the mean of 40 successful trials of a typical subject. Gray lines show the results without task-constraints. Black lines show the results with task-constraints. Solid and dashed lines show the mean data and SEM, respectively. Gray areas show the time when the hand was passing through the gate.

and decreased after passing through the gate (shoulder stiffness: Fig. 2G; elbow stiffness: Fig. 2H). Flexor and extensor muscles were co-activated and muscle impedance increased with task-constraints.

Fig. 3. The comparison of positional variance with and without task-constraints. Dotted lines show the data without task-constraints. Solid lines show the data with task-constraints. Gray areas show the time when the hand was passing through the gate.

K. Morishige et al. / International Congress Series 1301 (2007) 188–191

191

We calculated the positional variability from observed trajectories as follows: First, all trajectories from one subject were resampled at 100 equally spaced points along the path. Second, the average trajectory was computed. Third, for each average point, the nearest point from each trial was found and positional variance was calculated from that point. The positional variability at the entrance of the gate was smaller for the movement with the taskconstraint than that without task-constraints (Fig. 3). Thus, the improvement of movement was observed. 3. Discussion It is true that co-activating the flexor-extensor arm muscles would cause larger motor commands, and lead to a larger movement variability. Therefore, it seems paradoxical that the co-contraction of arm muscles improves movement accuracy. However, experimental data indicate that there is a high correlation between co-contraction and accuracy improvement. Our results suggest that the muscle impedance reduces motor variability arising from signal-dependent noise, and improves movement accuracy on demand. Acknowledgments This research was supported in part by the National Institute of Information and Communications Technology (NICT), the Ministry of Education, Science, Sports and Culture, Grant-in-Aid for Young Scientists (B) (16700301), and the 21st century COE project, “World of brain computing interwoven out of animals and robots.” References [1] P.H. Clamman, Statistical analysis of motor unit firing pattern in human skeletal muscle, Biophys. J. 9 (1969) 1233–1251. [2] P.B. Matthews, Relationship of firing intervals of human motor units to the trajectory of post-spike afterhyperpolarization and synaptic noise, J. Physiol. 492 (1996) 597–628. [3] C.M. Harris, D.M. Wolpert, Signal-dependent noise determines motor planning, Nature 394 (1998) 780–784. [4] P.L. Gribble, et al., Role of cocontraction in arm movement accuracy, J. Neurophysiol. 89 (2003) 2396–2405. [5] R. Osu, et al., Optimal impedance control for task achievement in the presence of signal-dependent noise, J. Neurophysiol. 92 (2004) 1199–1215. [6] R. Nonaka, M. Sato, Y. Koike, Speed–accuracy trade-off from the point of view muscle activity, IEICE Trans. Inf. Syst. J87-D-II (4) (2004) 1008–1019 (in Japanese). [7] G. van Galen, W. de Jong, Fitts' law as the outcome of a dynamic noise filtering model of motor control, Hum. Mov. Sci. 14 (1995) 539–571. [8] L.P.J. Selen, P.J. Beek, J.H. van Dieën, Can coactivation reduce kinematic variability? A simulation study, Biol. Cybern. 93 (2005) 373–381. [9] R. Osu, et al., Short- and long-term changes in joint co-contraction associated with motor learning as revealed from surface EMG, J. Neurophysiol. 88 (2002) 991–1004.