Adaptive control of saccades via internal feedback

Area of Science:

• Neuroscience
• Motor Control
• Computational Neuroscience

Background:

• Ballistic movements like saccades are accurate despite lacking real-time sensory feedback.
• Internal forward models are theorized to predict sensory consequences and correct motor commands for accurate movements.

Purpose of the Study:

• To investigate how the brain's forward model adapts to prediction errors in saccade control.
• To predict and analyze changes in saccade trajectories due to forward model adaptation.

Main Methods:

• Subjects performed saccades in a paradigm where the target jumped vertically mid-movement.
• Optimal feedback control theory was used to predict adaptation effects.
• Behavioral data on saccade trajectories and adaptation were analyzed.

Main Results:

• Saccade endpoints adapted towards the jumped target with training.
• Saccades became increasingly curved (suboptimal) due to on-line corrections.
• Two adaptive components were identified: slow-adapting initial commands and fast-adapting late corrections.

Conclusions:

• The brain utilizes a forward model with distinct adaptive timescales for saccade control.
• A fast-adapting component corrects late in the saccade, contributing to curvature and improved accuracy.
• This suggests internal feedback mechanisms with characteristics of a rapidly adapting forward model are crucial for saccade control.