Apparatus for measuring and perturbing shoulder and elbow joint positions and torques during reaching

Abstract

Visually guided reaching movements by monkeys has become an important paradigm for examining the function of various sensory and motor areas of the brain. However, a major problem with interpreting neural discharge during this motor task has been the difficulty to quantify and manipulate the mechanics of movement. To address this problem, a new experimental facility has been developed to allow neural recordings in a monkey while it makes movements with a mechanical linkage attached to its arm. The device (KINARM) has hinge joints aligned with the monkey’s shoulder and elbow and allows the monkey to make arm movements in the horizontal plane. Custom-made fibreglass braces attach the linkage to the monkey’s forearm and arm. Motors attached to the mechanical linkage provide angular position of the joints and apply torques either to the shoulder or elbow, or both. The KINARM is used in concert with a computer projection system that provides virtual targets in the plane of the arm. Preliminary results illustrate the ability of a monkey to perform a variety of multi-joint motor tasks under various static and dynamic loads.

Introduction

Reaching movements provide an ideal paradigm for understanding how sensory information is converted into coordinated motor behavior. On the sensory side, movements to spatial targets require the integration of visual and proprioceptive information to select and plan the impending limb movement. On the motor side, force generated by a muscle generates motion at both spanned and non-spanned joints due to intersegmental dynamics (Hollerbach and Flash, 1982, Zajac and Gordon, 1989). This coupling between shoulder and elbow muscle activity and motion requires the CNS to carefully coordinate the motor patterns of muscles at these joints in order to smoothly move the hand in space.

Visually guided reaching movements by monkeys has become an important paradigm to study how regions of the brain, such as primary motor cortex (MI), are involved in the planning and control of voluntary movement (Kalaska and Crammond, 1992, Georgopoulos, 1995, Shen and Alexander, 1997, Zhang et al., 1997). In spite of many years of research, lively debate still remains on the most appropriate way for interpreting the discharge patterns of individual cells in MI. On the one side, many studies have illustrated that the discharge pattern of individual cells are broadly tuned to the direction of hand movement and that a population vector, constructed from the discharge patterns of a large sample of cells, tends to point in the direction of movement (Georgopoulos et al., 1988, Schwartz et al., 1988, Schwartz, 1994). The implication from these studies is that individual cells in MI may be related to higher level features of movement related to the global goal of the task such as the direction of hand movement. On the other side, the discharge of individual cells has been found to share common temporal and spatial features with the activity patterns of proximal arm muscles during reaching (Murphy et al., 1985, Miller et al., 1996, Scott, 1997) suggesting that MI may be involved in the details of motor output at the shoulder and elbow joints.

It seems surprising that such divergent frameworks for interpreting the neural representation in MI during reaching movements are espoused in the literature. However, the ability to find neural correlates of movement to support these different views at least partially reflects the inherent coupling between different variables of movement (Mussa-Ivaldi, 1988, Scott and Kalaska, 1997). Extrinsic variables of movement, such as the kinematics of the hand, covary with intrinsic variables, such as motion of the shoulder or the activity of proximal arm muscles. As a result, neural activity related to a specific feature of movement will also covary with many other variables.

In order to dissociate the various parameters of movement, several studies have examined the response patterns of individual cells when reaching movements are performed under different mechanical loads or arm postures (Kalaska et al., 1989, Caminiti et al., 1990, Scott and Kalaska, 1997). A consistent finding from these studies is that the directional tuning of many MI cells are modified by load or posture. These results suggest that the activity of cells is not simply related to the direction of hand movement, but may be at least partially related to intrinsic features of the task, such as joint kinematics or kinetics. However, further progress on interpreting the nature of the discharge patterns of these cells based on features of motor execution are difficult in these studies because neural activity was only related to hand position or motion.

The major stumbling block for identifying the nature of the neural representation in MI during reaching is the difficulty of quantifying and manipulating the mechanics of multi-joint motion. The present article describes a new device developed specifically to address this problem called KINARM (kinesiological instrument for normal and altered reaching movements). The device uses a four-bar linkage and torque motors to record the motion of the arm in the horizontal plane and apply loads to each joint independently. While related devices have been developed to apply loads to the arm during multi-joint movements (Shadmehr and Mussa-Ivaldi, 1994, Gomi and Kawato, 1996), the loads have always been applied through the hand. A unique feature in the present system is that the loads are applied directly to the upper arm and forearm and allow the loads to be focused either at the shoulder or elbow joints, or both.