Modeling task-specific neuronal ensembles improves decoding of grasp.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Motor Control

Background:

• Dexterous movement relies on coordinated neuronal networks across cortical regions.
• Current models often focus on individual neuron firing rates, potentially missing broader ensemble activity.
• Variations in neural ensemble connectivity may offer deeper insights into motor behaviors.

Purpose of the Study:

• To investigate neural ensemble activity in motor and premotor cortices during a reach, grasp, and manipulate task.
• To develop and evaluate point process encoding models incorporating both baseline firing rate and functional connectivity.
• To determine if ensemble connectivity variations improve the classification of grasp types.

Main Methods:

• Recorded neural ensemble activity in primary motor cortex (M1) and premotor cortex (PM) of rhesus monkeys.
• Constructed point process encoding models with task-specific variations in firing rate and functional connectivity.
• Assessed model performance in encoding neuronal firing and classifying grasp actions.

Main Results:

• Task-specific ensemble models accurately predicted grasps (>95% accuracy).
• These models outperformed baseline models (variable firing rate only) in 82% of units.
• Ensemble activity improved spike timing predictability in 62% of units.

Conclusions:

• Functional connectivity variations in motor cortical ensembles provide discriminative information for decoding complex behaviors like grasping.
• This approach holds promise for developing more reliable and accurate neural prosthetics.