Modeling inter-trial variability of pointing movements during visuomotor adaptation

Area of Science:

• Neuroscience
• Cognitive Science
• Motor Control

Background:

• Trial-to-trial variability in visuomotor adaptation is attributed to planning and execution noise.
• Standard methods like Kalman filters struggle with closed-loop systems and recursive noise propagation.
• Existing models fail to explain increased variability during early adaptation phases.

Purpose of the Study:

• Develop a novel method to estimate noise parameters in visuomotor adaptation.
• Investigate the sources of trial-to-trial variability under closed-loop and error-clamp conditions.
• Determine whether noise is signal-dependent and identify its specific source.

Main Methods:

• Developed a method to compute the exact likelihood for a linear adaptation system under closed-loop conditions.
• Identified variance parameters by maximizing the computed likelihood.
• Compared model predictions of variance and autocovariance with empirical data.

Main Results:

• The observed increase in variability during early training could not be explained by constant planning or execution noise.
• Signal-dependent planning noise, rather than execution noise, accurately models temporal changes in trial-to-trial variability.
• Signal-dependent planning variance was specific to fast adaptation mechanisms; slow adaptation was explained by constant planning variance.

Conclusions:

• The study introduces a new method for analyzing noise in closed-loop visuomotor adaptation.
• Findings suggest that signal-dependent planning noise is a key factor in early adaptation variability.
• Different noise characteristics underlie fast versus slow adaptation mechanisms.