A multidisciplinary system identification of the human precision grip

Abstract

This thesis focus on the unique human ability to pick up and manipulate small objects using the precision grip, i.e., the fine grip between the index finger and the thumb. This ability is formed by the interplay between the complex biomechanical machinery of the hand and the neural networks within the central nervous system. To gain further knowledge about how humans control the precision grip, both the biomechanical system and the neural control system has to be examined.

For the control system, we used functional magnetic resonance imaging on human subjects to mapout the cortical network involved in different aspects of the precision grip. When subjects applied a small grip force in comparison to a larger gripforce, activity in secondary sensorimotor related areas in the frontal and parietal lobes increased. Our result suggests that these areas play an important role in the control of fine precision grip forces in the range typically used in manipulation of small objects. We also observed increased activity in a small area of the right intraparietal cortex when the subjects coordinated the grip and the load forces in an attempt to lift a fixated object. This is the first evidence for involvement of the posterior parietal cortex in the sensorimotor control of coordinated grip and lift forces in manipulation.

We applied a system identification approach to model the biomechanical system. A new technique was presented, common subsystem identification, with which the common mathematical factors in two models, identified using data from functionally different experiments, was estimated. It is concluded that these factors represent the common subsystem, here the grip force generator, involved in the experiments. The characteristics of the identified model were in agreement with experimental data on human neuromuscular grip force dynamics.

The model was then extended to include wrist movement, and the consequent lift force from the frictional interaction with an object. Predictions of the motor commands (control signals) could then be formed by feeding human data through the inverted model. Hence, the motor commands used in response to an unexpectedly low friction at the grip surface, could be estimated. In agreement with recordings from neurons in the primary motor cortex of the monkey, a sharp burst in the estimated motor command for the gripforce efficiently arrested any slip. The results further indicate a state dependent control system that uses a small set of efficient corrective commands.

The predicted motor commands was also compared with those theoretically optimized, with respect to minimum variance in fingertip forces. This was based on the suggestion that signal-dependent noise in the motor command influences movement control (Harris and Wolpert, 1998). We show, for the first time, that minimization of the gripforce variance, due to signal-dependent noise, indeed explains the characteristic motor commands and force profiles of a voluntary precision grip, as well as the motor command for an automatic grip force correction due to a slip.