Neural Population Dynamics during Reaching Are Better Explained by a Dynamical System than Representational Tuning.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Motor Control

Background:

• Motor cortex models traditionally focus on movement parameters (representational models).
• Emerging evidence supports population-level dynamical systems for movement generation.
• Integration and comparison of these frameworks remain incomplete.

Purpose of the Study:

• To evaluate representational models against dynamical systems in explaining motor cortex neural activity.
• To investigate the emergence of population-level rotational dynamics.
• To assess the explanatory power of different models using novel statistical methods.

Main Methods:

• Simulated center-out reaching using a representational velocity-tuning model.
• Introduced variable latency offsets to simulate neural population dynamics.
• Developed and applied a covariance-matched permutation test (CMPT) for data analysis.
• Utilized recurrent neural networks (RNNs) to model neural activity.

Main Results:

• Variable latency offsets in representational models can generate rotational dynamics.
• CMPT revealed representational model rotations are not condition-dependent.
• Dynamical models and actual motor cortex data show condition-dependent rotations.
• RNNs successfully reproduced both representational tuning and rotational dynamics.
• CMPT highlighted the need for caution with small datasets in motor cortex research.

Conclusions:

• Representational models alone lack sufficient explanatory power for complex motor cortex activity.
• Dynamical systems provide a more robust framework for understanding neural population dynamics.
• The study offers insights into neural data analysis and experimental design for motor control research.