Date of Award

5-7-1990

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Physiology and Pharmacology

Abstract

Current experimental evidence suggests that muscle spindles do play a significant role in the sensation of joint position and velocity. However, the dynamic characteristics of muscle spindles are complex and highly variable. Researchers have found, in the primary sensory cortex of primates, neurons whose activities are proportional to joint position and velocity. Some of these neurons were excited only by activating muscles spindles. This leads to the research question: can joint position and velocity be extracted from muscle spindle activity using a neurally-inspired mathematical model.The method of solution included modeling a simple hinge joint with an antagonist and agonist muscle set. Passive motion was imposed upon the joint model producing muscle stretch. Muscle stretch caused sensory zone stretch in the muscle spindles and the resultant activity in the afferent neuron was calculated. The firing rates in the afferent neuron were used as training data for a neural network which is a computer simulation of the neural processing involved in the dorsal column - medial lemniscus kinesthetic pathway. The output of the computer-simulated neural network was a neuron firing rate proportional to position or velocity and the network was trained until the error between the predicted value and actual value was acceptable.It was found that a back propagation neural network based on the anatomy of the dorsal column - medial lemniscus kinesthetic pathway could, with small error, predict joint position and velocity from only muscle spindle information. The neural net had 8 input (dorsal column nuclei) neurons, 9 middle (thalamic) layer neurons and 1 position or velocity (cortical) neurons.

Download

Share

COinS