Rotational dynamics in motor cortex are consistent with a feedback controller

Area of Science:

• Neuroscience
• Computational Neuroscience
• Motor Control

Background:

• Recent research highlights rotational dynamics in the motor cortex (MC).
• Existing theories often attribute these dynamics to intrinsic MC connections.
• Alternative perspectives suggest MC functions as a feedback controller, relying on sensory feedback and inter-areal interactions.

Purpose of the Study:

• To investigate the origins of rotational dynamics in motor control.
• To reconcile conflicting theories regarding intrinsic versus feedback-driven dynamics in the motor cortex.
• To explore the role of sensory feedback in generating observed neural dynamics.

Main Methods:

• Developed recurrent neural networks (RNNs) to model a limb and its sensory feedback.
• Trained RNNs to perform motor tasks: counteracting perturbations and reaching for targets.
• Analyzed neural activity and sensory feedback within the RNNs.
• Recorded neural activity in monkeys performing similar motor tasks.

Main Results:

• Rotational structure was observed in RNN activity and sensory feedback, even without recurrent connections.
• Neural recordings from monkeys showed rotational structure in both motor cortex and somatosensory cortex.
• These findings challenge the exclusive reliance on intrinsic connections to explain motor cortex dynamics.

Conclusions:

• Rotational structure in neural activity may reflect broader dynamics within the voluntary motor system.
• Continuous sensory feedback and inter-areal interactions play a crucial role in online motor control.
• The study suggests that rotational dynamics are not exclusive to the motor cortex but are present throughout the motor system involved in action control.