Revealing unexpected complex encoding but simple decoding mechanisms in motor cortex via separating behaviorally...

Area of Science:

• Neuroscience
• Computational Neuroscience
• Motor Control

Background:

• Neural responses in the motor cortex are often entangled with irrelevant signals, complicating the study of neural encoding and decoding.
• The ground truth of behaviorally relevant neural signals is not well-defined, making it difficult to isolate them.

Purpose of the Study:

• To propose a novel framework for defining, extracting, and validating behaviorally relevant neural signals.
• To investigate the nature of neural information previously considered irrelevant in motor cortex activity.

Main Methods:

• Developed a framework to separate behaviorally relevant and irrelevant neural signals at single-neuron and single-trial levels.
• Analyzed separated signals from three monkeys performing various reaching tasks.
• Investigated both nonlinear and linear decoding of behaviorally relevant signals.

Main Results:

• Neural responses previously thought to contain little information were found to encode rich behavioral data in complex nonlinear ways.
• Identified that movement behaviors occupy a higher-dimensional neural space than previously understood.
• Demonstrated that behaviorally relevant signals can be decoded linearly with performance comparable to nonlinear decoding when incorporating often-ignored neural dimensions.

Conclusions:

• Separating behaviorally relevant signals is crucial for uncovering hidden cortical mechanisms.
• The findings suggest that linear readout mechanisms may be employed in the motor cortex.
• Behaviorally relevant signals are more complex and occupy a higher-dimensional space than previously assumed.