A quantitative model of conserved macroscopic dynamics predicts future motor commands

Area of Science:

• Neuroscience
• Computational Biology
• Systems Biology

Background:

• Whole brain imaging in simple organisms like *Caenorhabditis elegans* provides rich data for nervous system analysis.
• Understanding the predictive capacity of neuronal activity is key to deciphering motor control.

Purpose of the Study:

• To develop a predictive model of the nervous system based on neuronal activity.
• To investigate the relationship between macroscopic dynamics and individual neuronal variations.

Main Methods:

• Utilized whole brain imaging recordings from *Caenorhabditis elegans*.
• Developed a novel methodology to extract neuronal activity loops in phase space.
• Constructed a predictive model using two macroscopic variables: loop identity and phase.

Main Results:

• The model accurately predicts motor commands controlling locomotion up to 30 seconds prior to the event.
• Predictions were validated on individuals not included in model training.
• The model successfully predicts dwell time statistics, motor command sequences, and individual neuron activation.

Conclusions:

• A predictive model based on macroscopic variables can capture universal nervous system dynamics.
• This framework reconciles consistent inter-individual differences in neuronal activation with overarching macroscopic principles.
• The study demonstrates the power of phase space analysis in understanding complex neural systems.