Computational hypothesis testing for neuromuscular systems.

Area of Science:

• Computational neuroscience
• Motor control
• Biophysics

Background:

• Human index finger force fluctuations exhibit non-uniform patterns across different directions.
• Understanding the underlying mechanisms of motor control is crucial for neuroscience and rehabilitation.

Purpose of the Study:

• To demonstrate that computational models can rigorously test hypotheses about biological mechanisms.
• To discriminate between two competing hypotheses for human index finger force fluctuations based on muscle effort cost functions.

Main Methods:

• Formulated two computational hypotheses based on minimizing sum of muscle forces versus sum of squared muscle forces.
• Employed stochastic parameter exploration to generate probability distributions for each hypothesis's predictions.
• Compared observed human force fluctuation data against the predicted distributions.

Main Results:

• Observed human force fluctuation data did not fall within the predicted probability distribution for the hypothesis minimizing the sum of muscle forces (Hypothesis 1).
• Observed data were consistent with the probability distribution for the hypothesis minimizing the sum of squared muscle forces (Hypothesis 2).

Conclusions:

• The computational approach provided rigorous evidence favoring the hypothesis that the central nervous system minimizes the sum of squared muscle forces during index finger exertion.
• This method allows for hypothesis discrimination even with parameter uncertainty, offering a powerful tool for scientific inquiry in motor control and beyond.