Exploration biases forelimb reaching strategies.

Area of Science:

• Neuroscience
• Motor Control
• Machine Learning

Background:

• The brain employs diverse strategies for action generation, like reaching.
• Understanding how these movement strategies are learned is crucial for motor control research.

Purpose of the Study:

• Investigate strategy learning in a joystick-controlled reaching task.
• Determine how sensorimotor cortex involvement influences reaching refinement.
• Explore the correlation between training exploration and learned strategies.

Main Methods:

• Head-fixed mice trained to reach an invisible target using a joystick.
• Analysis of joystick trajectories to identify movement strategies (direction vs. endpoint).
• Comparison with model-free reinforcement learning agents.

Main Results:

• Mice developed biased reaching strategies, favoring either direction or endpoint control.
• This bias correlated with the spatial variability of joystick exploration during training.
• Reinforcement learning agents replicated this bias-learning correlation.

Conclusions:

• Reinforcement of individual exploratory behavior during training significantly influences learned reaching strategies.
• This provides insight into how motor learning adapts based on prior exploration patterns.