Reach-dependent reorientation of rotational dynamics in motor cortex

Area of Science:

• Neuroscience
• Motor Control
• Computational Neuroscience

Background:

• Neurons in the motor cortex display complex activity patterns during reaching movements.
• While single-neuron activity correlates with movement parameters, it only partially explains neural responses, suggesting population-level dynamics are crucial.
• Previous models described these dynamics as "rotational," with neural activity orbiting in a state space.

Purpose of the Study:

• To investigate novel features of motor cortex neural dynamics during reaching that standard models fail to explain.
• To develop and validate a new model, "location-dependent rotations," to better capture neural activity.
• To explore the relationship between neural representations and dynamics in motor cortex control.

Main Methods:

• Reanalysis of reaching datasets from male Rhesus macaques.
• Development of the "location-dependent rotations" model.
• Decoding of reach kinematics from neural spiking data.

Main Results:

• Identified two key features not explained by standard dynamics: differing rotation planes for different reaches and plane variation linked to activity location in state space.
• The "location-dependent rotations" model successfully fits nearly all motor cortex activity during reaching.
• High-quality decoding of reach kinematics showed a quasilinear relationship with spiking activity.

Conclusions:

• The "location-dependent rotations" model provides a more comprehensive explanation of motor cortex activity during reaching than previous models.
• Varying rotational planes enable richer motor outputs, enhancing the motor cortex's capabilities.
• The study integrates representational and dynamical concepts, showing how state space location (representation) is converted into time-varying command signals by dynamics.