Interlimb Generalization of Learned Bayesian Visuomotor Prior Occurs in Extrinsic Coordinates

Area of Science:

• Neuroscience
• Motor Control
• Computational Neuroscience

Background:

• The brain is thought to represent probability distributions and perform Bayesian integration during sensorimotor learning.
• However, the neural representation of this learning process is not well understood.

Purpose of the Study:

• To investigate the neural representation of sensorimotor learning.
• To determine if visuomotor learning generalizes between limbs and its reference frame.

Main Methods:

• Replicated behavioral findings on Bayes-optimal sensorimotor adaptation.
• Tested generalization of visuomotor learning to the untrained limb.
• Analyzed learning representation in extrinsic vs. intrinsic reference frames.

Main Results:

• Humans demonstrated Bayes-optimal performance by integrating sensory uncertainty and statistical distributions.
• Visuomotor learning generalized to the opposite limb only when the perturbation was extrinsically congruent.
• This indicates that learned visuomotor adaptations are represented in an extrinsic reference frame.

Conclusions:

• Sensorimotor learning involves Bayesian integration of sensory information and prior knowledge.
• The neural representation of visuomotor learning is primarily coded in an extrinsic reference frame.
• This extrinsic representation is accessible to untrained limbs, facilitating generalization.