Neural Network Models for Spinal Implementation of Muscle Synergies

Area of Science:

• Neuroscience
• Motor Control Research
• Computational Neuroscience

Background:

• Muscle synergies are proposed functional modules simplifying complex motor control.
• The precise neural basis of muscle synergies, particularly spinal cord involvement, remains unclear.
• Existing models struggle to account for observed neural variations in spinal premotor interneurons (PreM-INs).

Purpose of the Study:

• To compare different neural network models of muscle synergies.
• To identify a biologically plausible model for the neural implementation of muscle synergies.
• To reconcile electrophysiological findings with clinical observations of motor control robustness.

Main Methods:

• Development and comparison of three neural network models: non-synergy, simple synergy, and population synergy.
• Simulation of muscle synergy robustness against simulated cortical stroke.
• Analysis of neural variation within the population synergy model.

Main Results:

• Both simple and population synergy models successfully emulated the robustness of muscle synergies observed in human stroke patients.
• The variation in the population synergy model closely matched the recorded variation in spinal PreM-INs in monkeys.
• The non-synergy model did not replicate the observed robustness.

Conclusions:

• A spinal neural population model with moderate variation offers a biologically plausible explanation for muscle synergy implementation.
• This population synergy model reconciles findings on neural variation and functional robustness of motor control.
• The findings challenge simplistic 'module' theories and highlight the role of neural population dynamics.